National Science Foundation Science and Technology Center. Second Annual Report 5/1/2003 4/30/2004

Transcription

1 National Science Foundation Science and Technology Center Center for Integrated Space Weather Modeling Second Annual Report 5/1/2003 4/30/2004 Boston University 725 Commonwealth Avenue Boston, MA 02215

2 Table of Contents I. GENERAL INFORMATION...4 II. 1A. LIST OF INSTITUTIONS:...4 1B. BIOGRAPHICAL INFORMATION FOR NEW FACULTY MEMBERS EXECUTIVE SUMMARY...6 RESEARCH...9 1A. OVERALL RESEARCH DESCRIPTION...9 1B. PERFORMANCE AND MANAGEMENT INDICATORS C. PROBLEMS A. SOLAR/HELIOSPHERIC THRUST...12 Activities...12 Significant Accomplishments...13 Goals and Activities planned for this coming year...13 Research Highlights B. MAGNETOSPHERIC THRUST...21 Activities...21 Significant Accomplishments...21 Goals and Activities for the Coming Year: C. IONOSPHERIC/THERMOSPHERIC THRUST...30 Significant Accomplishments...30 Goals and Activities Planned for the Coming Year D. CODE COUPLING THRUST...35 Significant Accomplishments...36 Goals and Activities Planned for the Coming Year...37 Ad hoc (Generation 0) Coupling...38 First Generation (Object Oriented) Framework E. MODEL VALIDATION AND METRICS THRUST...49 Activities...49 Significant Accomplishments...49 First CISM Validation Studies F. EMPIRICAL AND FORECAST MODELING THRUST...54 Significant Accomplishments...54 Goals and Activities Planned for the Coming Year G. DATA ASSIMILATION THRUST...60 REFERENCES...63 III. IV. EDUCATION...65 Significant Accomplishments...65 Education Metrics...72 KNOWLEDGE TRANSFER...74 V. EXTERNAL PARTNERSHIPS...77 VI. VII. DIVERSITY...79 Significant Accomplishments...79 Goals and Plans for the Coming Year...79 Diversity Metrics...82 MANAGEMENT...83 A1. ORGANIZATIONAL STRATEGY...83 A2. PERFORMANCE AND MANAGEMENT INDICATORS...84 B. PROGRESS AND PROBLEMS...84

3 C. COMMUNICATION WITHIN CISM...84 D. CISM ADVISORY COUNCIL...86 VIII. CENTER-WIDE OUTPUTS AND ISSUES...87 A1. CENTER PUBLICATIONS...87 A2. CONFERENCE PROCEEDINGS...89 B. AWARDS AND HONORS...96

4 STC Annual Report I. General Information 1a. List of Institutions: Date Submitted 1 May 2003 Reporting Period 1 August July 2004 Name of Center Center for Integrated Space Weather Modeling Name of Center Director W. Jeffrey Hughes Lead University Boston University Names of participating institutions, and role Institution 2 Name Alabama A&M University Role of Institution at Center AAMU will work with Boston University on model validation and with UTEP on education and increasing diversity. Institution 3 Name Role of Institution at Center Institution 4 Name Role of Institution at Center Institution 5 Name Role of Institution at Center Institution 6 Name Role of Institution at Center Institution 7 Name Role of Institution at Center Institution 8 Name Role of Institution at Center Institution 9 Name Role of Institution at Center Dartmouth College Dartmouth College will lead the magnetospheric modeling effort. National Center for Atmospheric Research (NCAR) NCAR will lead the ionosphere/thermosphere modeling effort. Science Applications International Corporation (SAIC) SAIC will work with University of California, Berkeley on the solar/solar wind effort and with Boston University on code coupling Stanford University Stanford will work with University of California, Berkeley on the solar/solar wind effort. University of California, Berkeley University of California, Berkeley will lead the solar/solar wind effort. University of Colorado, Boulder The University of Colorado will lead the knowledge transfer and empirical model efforts and will work closely with NOAA/SEC to ensure our models are transitioned into the forecasting community. University of Texas, El Paso University of Texas at El Paso will lead the education

5 Institution 10 Name Role of Institution at Center and diversity efforts and will work with Boston University on model validation. William Marsh Rice University Rice will work with Dartmouth College on magnetospheric physics and with Boston University on code coupling. 1b. Biographical Information for New Faculty Members None

6 2. Executive Summary The Center for Integrated Space Weather Modeling (CISM) has as its overarching vision To understand our changing Sun and its effect on the Solar System, life, and society. Within this greater vision, CISM sees as its particular missions: to introduce into space physics and space weather research the first comprehensive community model analogous to the community models that exist in other fields such as climate research; to introduce into operational space weather prediction and forecasting the use of physics-based numerical simulation models in the same way as they are used in, for example, tropospheric weather forecasting; and to introduce in education, particularly undergraduate and graduate education, the notion that sun-earth science must be viewed as a single, unified field of research and study and not several separate disciplines. CISM focuses its activities around the development of a series of ever-improving versions of a comprehensive, physics-based simulation model that describes the space environment from the Sun to the Earth. After having fully tested and validated these models, we will use them for research, make them available to the wider research community, transition them as appropriate into operational specification and forecasting tools, and use them as learning tools. This shared vision and task binds the CISM team into a tight center with everyone doing their part towards the common goal even though CISM is split geographically between ten institutions spread across the country. Our plan is to produce a new version of the comprehensive model every two years. Each version will use more sophisticated coupling technology and incorporate more of the physics needed to fully describe the coupled Sun-Earth system. The three science research thrusts, solar/heliosphere, magnetosphere, and ionosphere/thermosphere, are responsible for developing the individual models that will become component models within the coupled comprehensive model. The code coupling thrust both develops the computational science tools required to couple the component models and uses these tools to couple them into a comprehensive model. The model validation and metrics thrust both fully tests and validates the model and its components, and also compares their performance to that of standard models using the metrics we have developed. Models will then be released to the wider community and, if appropriate, developed into operational and teaching tools. Validated versions of the comprehensive model will be made accessible to the wider community with help from our partners the Community Coordinated Modeling Center (CCMC) and the National Computational Science Alliance (NCSA). In parallel to the development of the comprehensive physics-based model, the empirical and forecast modeling thrust has developed comprehensive empirical models of the Sun-to-Earth system. These models emphasize specification and prediction and provide both a benchmark (in terms of metrics and skill scores) by which the physics-based models can be judged, and forecasting tools that can be transitioned to operations much earlier in the life of CISM. The first version of the comprehensive model (Ad hoc version or Version 0) utilizes ad hoc coupling technology to couple the four fluid codes (MAS, Enlil, LFM, and TING) that form the core of the comprehensive model as well as a fifth code that models the ring current region (the Rice Convection Model). The second version will make use of object oriented programming concepts (specifically the InterComm package and Overture framework) to provide more modular and versatile code coupling, and will include additional component codes that model the radiation belts, solar energetic particles, and possibly others if their development is

7 sufficiently mature. The coupling technology to be used for the third version will be selected later from among options still under development in the computational science community. Education and research are integrated by involving all CISM students, both undergraduate and graduate, in the research effort, by using the models we develop as instructional tools, and by conducting science education research focused on measuring the effectiveness of using images in education within the education plan. Our very successful space weather summer school uses models in its curriculum. Our graduate student retreat builds community among and provides professional development opportunities for CISM graduate students. A major component of our diversity plan is for CISM to help and support Alabama A&M University as they create a graduate program in space science within their physics department. This program will provide a route for African American students to enter a field within which they are very poorly represented, and will remain a lasting legacy of CISM. The CISM knowledge transfer plan has three major components: transistion of forecasting tools to the NOAA/Space Environment Center (SEC); providing the wider scientific community with comprehensive models accessible to all; and training and interacting with CISM s partners within the aerospace industry, government, and others who must cope with or mitigate against space weather effects. Our close cooperation with SEC has been strengthened with the hiring of Michael Gehmeyr to liaise with SEC and use his understanding of SEC s priorities to help design visualization tools and the comprehensive empirical model. We are formalizing our partnerships with CCMC and NCSA who will work with us to make our models available to the wider community. We are forming relationships with our industrial partners to learn their needs. Several government and industrial employees attended the 2003 CISM Summer School. We expect a similar number to attend this summer. During our first year we produced a strategic and implementation plan that describes our goals and outlines a path forward to achieving them. This document lays out a series of milestones with which we measure our progress along this path. Our management structure divides CISM into a set of overlapping teams each with an important task to perform as part of developing the comprehensive CISM model or towards achieving our goals in education, increasing diversity within space physics, and transferring tools, techniques and knowledge to the broader research and professional communities. The CISM All-Hands meeting, held in Boston in September, provides an annual forum for comparing progress with the milestones, and for developing detailed plans for the coming year. The CISM Advisory Council met again in March 2004 and provided very useful advice and guidance to the CISM management team. In September CISM was invited by the Journal of Atmospheric and Solar-Terrestrial Physics to produce a special issue. The issue will consist of twenty two papers describing CISM s plans and initial results that have all been successfully refereed. This will provide a valuable record of our progress at this point. The lasting legacies that CISM will leave behind are: The development of a new interdisciplinary science that views the Sun-Earth system as a single closely coupled system, and that erases the existing boundaries among space physicists. A new generation of well-trained space physicists from diverse backgrounds that is capable of using the tools of computational science to study the space environment and who approach problems from an interdisciplinary viewpoint.

8 A new graduate program in space science at an historically black university. The introduction of community models into space physics and the use of numerical models as research tools by the broader research community. Advances in space science, particularly in our understanding of processes critical to the development of the global model. Advances in computer science brought about by our need to efficiently couple disparate numerical models and assimilate observational data. New models and understanding of the space environment that will lead to improved specification and forecasts at the nation s space weather operations centers. A suite of physics-based forecasting and specification tools. A better public understanding of the Sun and its affect on the Earth s space environment. CISM s major accomplishments during the past twelve months were: Editing a collection of 22 papers describing CISM s plans and initial results that will be published in the Journal of Atmospheric and Solar-Terrestrial Physics. Holding a successful CISM Summer School in July/August 2003 attended by 30 students, and organizing the next summer school for 26 July 6 August, Holding the annual CISM all-hands meeting in September 2003 at which we developed detailed plans and milestones and assessed progress towards them. Holding the first CISM graduate student retreat in September Hosting the second CISM Advisory Council meeting in March Completing the construction of AccessGrid video-conferencing facilities at all CISM institutions, and instituting a schedule of regular video meetings for most of the CISM teams. Completing the code repository and data archive facilities for storing and organizing CISM model codes, model output, and observational data at Boston University. Establishing a common standard visualization package, CISM-DX, that will be used to display the results from all CISM models, and incorporating the current CISM empirical and forecast models into CISM-DX Developing a detailed list of parameters, models, and data sources to be used as metrics for calculating model skill scores. Completing a fully 3-D sun-to-earth coupled simulation of a CME using the results from the coupled MAS/Enlil codes to drive the coupled LFM/TING codes. Completing first generation ad hoc coupling of MAS and Enlil codes to form CORHEL ready for model validation and testing. Completing first generation ad hoc coupling of LFM and TING codes to form CMIT (Coupled Magnetosphere/Ionosphere/Thermosphere) ready for validatin and testing. Completing one-way, ad hoc, coupling of the ring current (RCM) model to the global magnetosphere (LFM) model. Modeling several scenarios of magnetic flux emergence into the corona including the formation of a new active region under an existing coronal arcade. Developing 3D particle codes to model the particle acceleration in the radiation belts and SEP entry into the magnetosphere and ionosphere. Incorporating neutral wind and corotation feedback into CMIT. Successful testing of the Overture and Meta-Chaos code coupling tools with space physics codes. Improving first generation coupling framework to incorporate InterComm. Hiring the full-time liaison based at NOAA/Space Environment Center. Transitioning the first CISM models to SEC for testing.

9 II. Research 1a. Overall Research Description The overarching goal of CISM is to develop a reliable and well-validated, comprehensive, physics-based numerical simulation model that describes the space environment from the Sun to the Earth. The research goals of CISM are all directed towards achieving this goal. This means that CISM s research program must be considered as an integrated whole, managed as a single large project. For the purposes of research management we have divided this task into components, which we identify as our research thrusts. However, these components are all interconnected, and the boundaries between them are necessarily fluid, and in some cases not easily defined. Some thrusts are identified by areas of science, others by the processes needed to build the comprehensive model. Figure 1 is a high-level schematic showing our plan for constructing, validating, and disseminating the generations of our models. In this diagram, time increases from left to right while increasing sophistication of our models increases down the page. Each horizontal row of boxes represents the passage of one generation of our model from one stage to the next in its development. The various research teams or thrusts are represented by the circle on the lower left or by a diagonal arrow. The circle represents the three regional science thrusts, the solar/heliospheric research thrust led by co-director Janet Luhmann, the magnetospheric research thrust led by co-director Mary Hudson, and the ionosphere/thermosphere research thrust led by co-director Tim Killeen and Stan Solomon. These thrusts have as their responsibility developing the models of the components of the Sun-Earth system. The blue arrow represents the code coupling thrust led by co-director Chuck Goodrich. Their task is to take the component models from the three thrusts just mentioned and couple them into an end- Figure 1. A schematic timeline of the plan for developing the comprehensive CISM model.

10 to-end coupled model. The smaller green, yellow, and orange boxes represent successive generations of the model each of which increases in its complexity, i.e., the number of component boxes, and in the sophistication of the coupling technology (i.e., computational tools) used to couple the models. When the coupling of one generation of the model is completed, the model is passed on to the model validation and metric thrust, led by co-director Harlan Spence, represented by the yellow arrow. After a model has been tested and vetted by the validation team it is ready for dissemination to the broader communities. The empirical and forecast modeling thrust led by co-director Dan Baker, who also leads the knowledge transfer team, has particular responsibility for developing versions of our models appropriate for the space weather forecast community and for transferring them to SEC and other potential users. The education team led by co-director Ramon Lopez will adapt models for educational use, using the CISM summer school as a test bed. Access by the broader scientific research community will originate from the CISM code repository at Boston University and will be facilitated by our partnerships with the Community Coordinated Modeling Center (CCMC) and National Center for Supercomputing Applications and National Computational Science Alliance (both NCSA). Other arrows in Figure 1 represent the feedback that occurs between teams. The experience gained during the validation process feeds back to improve later versions of the models being developed by the science and code coupling teams. Finally, the data assimilation thrust is not explicitly represented in Figure 1. Their initial responsibility is to investigate the ways in which the CISM models can make effective use of data assimilation techniques. Each box in Figure 1 represents about 2 years. Our progress to date places us on the horizontal arrow representing the transfer of the first generation model from the code coupling thrust to the model validation thrust. Currently many parts of the first generation model, labeled Gen 0 Ad Hoc, have been transferred, and all will be by the end of this summer or funding year. The code coupling thrust has already begun development of the second generation model that uses more sophisticated coupling software tools, while the science thrusts are developing component models to be used in this and later generations. Figure 1 illustrates how integrated the CISM team is and must be in order to complete its goals. The interactions between the trusts are, and have to be, continual and detailed. Figure 1 is misleading if it suggests that the separate tasks of model development, code coupling, model validation, and model dissemination are completely separable. They are not, and many CISM team members play important roles in several of these tasks. This is borne out by many names appearing in several teams in the CISM organization chart in Appendix B. Nevertheless the CISM organization helps to remind us all of the stages the model must go through in order to become a useful, reliable product, and that CISM research is applied research in the sense that it is all directed towards building a product. Figure 2 shows the component models that will be all ultimately coupled together in the comprehensive CISM model. The first generation model includes the four core models at the top of the figure, the MAS model of the solar corona, the Enlil model of the solar wind, the Lyon- Fedder-Mobarry (LFM) global magnetosphere model, and the ionosphere/thermosphere model here labeled T*GCM, together with the Rice Convection Model (RCM) of the ring current region. The red arrows indicate the ad hoc code coupling already completed. The other boxes represent models under development in the science thrusts, solar active region and solar energetic particle (SEP) models (green) in the solar thrust, radiation belt and magnetosphereionosphere coupling models (pink) in the magnetosphere thrust, and plasmasphere and exosphere models (orange) in the ionosphere/thermosphere thrust. These will be coupled into the comprehensive CISM model later.

11 Figure 2. A flow diagram showing the component models that will be coupled to form the comprehensive CISM model. Red arrows show the ad hoc coupling already completed. 1b. Performance and Management Indicators. We have developed a comprehensive set of milestones for the development of the CISM comprehensive model. These milestones include the projected schedules for the transfer of models from the science thrusts to the code coupling team, and for the delivery of successive versions of the comprehensive model from the code coupling team to the validation team and later through the knowledge transfer team to the broader community. They further describe the incorporation of new results into the forecast and specification model developed by the empirical and forecast modeling thrust. We use these milestones to assess our progress in meeting our research objectives. The milestones are contained in Appendix A. 1c. Problems. We have not encountered any significant problems.

12 2A. Solar/Heliospheric Thrust The overall science and simulation/modeling goal of the solar thrust group of CISM continues to be the production of codes that can provide a physically realistic solar wind at 1 AU, into which realistic coronal mass ejections (CMEs) are launched and propagate. The codes must reproduce interplanetary magnetic fields, and solar wind plasma densities and bulk velocities in 3D that are used to both validate the Solar-Heliospheric simulations and couple to CISM magnetospheric simulations. We are also developing parameterized models for CME shock production of Solar Energetic Particles (SEPs) within the global coronal and solar wind models, and their distribution in geospace. Achievement of these goals requires increased understanding of CME initiation and the related problem of active region evolution. It also requires routine modeling of the corona and solar wind based on solar observations. Activities: Our activities in the second year have centered around the production of our first deliverable steady state corona/solar wind simulation code, on the improvement of the simulation codes in both the spatial resolution and solar boundary description areas, on the identification and detailed analysis of the observations for our first realistic event studies, and on the initial development of the solar energetic particles code elements. In addition, the first year's work was completed and submitted with several other relevant studies for publication in the CISM special issue of JASTP. Janet Luhmann at UCB co-directs the CISM Solar thrust area, while Jon Linker at SAIC and Dusan Odstrcil at CIRES and NOAA SEC have primary responsibility for the core solar/heliospheric code development and delivery. Philip Scherrer's group at Stanford provides key solar observational support toward defining the model boundary and initial conditions. UCB is charged with developing the SEP module. In addition, a UCB CISM researcher, Tetsuya Magara, is hosted by Spiro Antiochos at NRL to bring additional perspectives into the modeling of CME initiation. Similarly, George Fisher at UCB provides the intellectual connections to the related DoD MURI center for the study of eruptive solar events. Sarah Gibson and Giuliana detoma are scientific and technical liaisons with HAO solar/heliosphere group, and Nick Arge, now at Air Force Research Laboratory, interacts locally with Boston University CISM members on the subject of the empirical solar/heliospheric models and their validation. A CISM Senior Fellow position for solar energetic particles (SEP) modeling at UCB has been filled by Dietmar Kraus-Varban. Two UCB CISM graduate students, Christina Lee and Camron Gorguinpour started work, and an undergraduate student, Ximena Cid, is engaged in part-time CISM research. The website for the solar thrust at highlights selected results and links to CISM solar/heliosphere publications. During the past year, Access Grid nodes have been set up at UC Berkeley, SAIC, and Stanford. Our weekly meetings on the grid have served to focus our team efforts and enhance communications. In addition, we have an annual solar thrust workshop in December, and solar-oriented sessions at the all-hands meeting. Other visits between members of the solar thrust occur as needed at member sites.

13 Significant Accomplishments: The accomplishments of the second year fulfill the following milestones set out in our detailed plan created at the all-hands meeting: Coupled Simulations Completed the version of steady state coupled MAS/ENLIL codes called "CORHEL", and delivered it and sample simulation data sets for validation at BU. Performed sun to earth coupled 3D simulation of an idealed CME eruption using the MAS, ENLIL, LFM and TING codes. progressed toward the goal of modeling several specific space weather events o investigated the use of magnetograms for model initiation of the May 12, 1997 and late 1996 period events, o collected and analyzing the solar and interplanetary observations for the May 12, 1997 event. First-order simulation of these events is underway. Modeled the idealized emergence of a new active region into a pre-existing coronal arcade using the Berkeley sub-photospheric model and the SAIC coronal model. Corona and CME Model Area: Incorporated localized high resolution grid in the coronal code to better simulate active region involvement. Simulated a global corona with a bipolar active region having vortical flows Used the subsurface codes to simulate vector magnetograms, and compared with observations. o Performed active region flux emergence and coronal response simulations with the NRL ARMS code with Adaptive Mesh Refinement. o Corona and Solar Wind Model Area o Implemented extension of ENLIL grid to poles. o Tested a sub-gridscale model of Alfven wave acceleration in MAS SEP Model Area Developed schemes for field line tracing in coupled MAS/ENLIL run data sets. Determined simulated interplanetary shock parameters in the coupled 3D CME run mentioned above. In addition, we carried out the following related work based on insights gained during the year: Implemented local correlation tracking of magnetic elements in sequences of magnetograms to obtain further observational information for CME initiation models Determined that photospheric vector magnetograms may not be as useful for establishing force-free initial descriptions of active region fields as chromospheric magnetograms Examined the validity of standard force-free models of interplanetary CME ejecta fields in light of the results of our first coupled CME event simulation Worked with validation group to identify coronal model metrics and data sets. Goals and Activities planned for this coming year: Our work this coming year will concentrate on the improvement of the coronal and heliospheric coupled models, on the simulation of a real CME event, and on the development of the SEP

14 model code. In addition, we will continue to support the code coupling and metrics and validation activities of CISM. Some specific goals are: Simulate a 3D ad-hoc CME in the MPI MAS model with nested grid and couple the result to ENLIL Simulate the May 12, 1997 CME event with coupled MAS/ENLIL, and provide the 1 AU interplanetary signature results to the LFM/TING geospace modeling group Compare MAS and WSA inputs for the ENLIL inner boundary Continue development of a more realistic solar wind in the MAS model. Provide several years worth of MAS/ENLIL "CORHEL" results on coronal and solar wind stream structure to BU for validation studies Refine strategy for describing interplanetary transport of SEPs (including scattering) in the SEP model Explore SEP transport in the ENLIL model fields using a simplified shock source description Research Highlights Some highlights of our results to date are briefly summarized here. The fastest CMEs typically originate from active regions on the Sun. As a prelude to an event study of an active region CME, we have studied the effects of including a simple bipolar active region within a global coronal model. Figure 3 shows the configuration we obtain when the solar wind is included, forming a 3D helmet streamer configuration. Although this is about the simplest model of an active region within the streamer belt, we note that the topology is still very interesting. Magnetic fields are attracted away from the neutral line between the two spots by the surrounding magnetic flux. This causes open field lines to exist within the active region. To test the susceptibility of this configuration to eruption, we imposed a rotational flow on the two spots. This vortical flow does not alter the flux distribution, but twists and energizes the field in the active region. Figure 4 shows that a portion of these twisted magnetic field lines expand and are carried out into the solar wind. This is not a true eruption in the sense that very little magnetic energy is released. To obtain a violent eruption, the shear may need to be more concentrated along the neutral line (commonly observed in filaments) and flux cancellation (also observed at filament sites) imposed. We plan to investigate these effects in the remainder of this year and in the 3 rd year.

15 Figure 3. The streamer structure that is obtained for a magnetic flux distribution consisting of dipole plus a bipolar active region. The top panel shows magnetic field lines from two views, the bottom panel shows the corresponding polarization brightness as would be seen by a coronagraph. The radial magnetic field is contoured on the surface. Red (blue) corresponds to outward (inward) pointing fields. One of the primary challenges of real CME event simulation is the description of the boundary and initial conditions. The MAS code will more accurately capture the structure and dynamics of the ejecta if it is initialized with a nonpotential magnetic field resembling the eruptive active region field, and includes boundary velocities consistent with observations. Abbett et al. (2004) find that force-free fields based on either photospheric or chromospheric magnetograms can closely approximate the active region fields. These force free fields are also close to those that occur in the full MHD model. Figure 5 shows comparisons of potential, force-free, and MHD fields where the first two are derived from synthetic magnetograms obtained from the MHD model. The force free model clearly shows the sigmoidal structure seen in the soft x-ray images of many eruptive active regions, suggesting it can provide an initial condition close to the eruptive state.

16 Figure 4. Expansion of a twisted flux bundle from a model active region. In the initial equilibrium streamer, blue field lines are closed, black field lines are open, magenta field lines are traced from the active region. Randomly colored field lines are traced from the center of the positive polarity spot. Twisting flows cause these field lines to rise until they are carried outward by the solar wind. Figure 5. Field lines from potential (left) and force-free (center) models of an active region magnetic field based on synthetic magnetograms derived from the MHD simulation on the right. The top and bottom rows represent models based on synthetic chromospheric and photospheric magnetograms, respectively. The force-free models closely resemble the MHD field configurations. (From Abbett et al., 2004)

17 Figure 6. Field lines calculated using a force free model and scalar MDI magnetograms obtained before the May 12, 1997 CME. The field models (right) were parameterized to obtain the best comparisons with the Yohkoh soft x-ray images on the left. Such field descriptions can be used to initialize MHD models of the eruption. (From Liu, 2004) In related observation-based work, Liu (2004) calculated a force-free model of the soft x-ray sigmoid imaged in our selected event active region in May, 1997 by Yohkoh SXT. The comparison between the force-free model field line geometry and the x-ray observations is reproduced in Figure 6. Notably, this model field is not particularly consistent with the photospheric vector magnetograph data for this active region. The successful force free model is based on the photospheric scalar field observations from MDI and the soft x-ray image only. Local correlation tracking of field patterns in the sequence of MDI magnetograms surrounding the CME, carried out by Li et al. (2004), suggests the prevailing horizontal velocity structure at the photospheric boundary shown in Figure 7. This velocity pattern includes an inferred convergence of flow toward the magnetic neutral line of the active region, consistent with the flux cancellation mechanism of CME initiation used in the CISM ad-hoc CME models (Linker et al., 2003) and the first 3D end-to-end event study (Luhmann et al., 2004). Such observationally derived velocities can be used to guide the event simulations.

18 Figure 7. Photospheric horizontal velocity vectors inferred from tracking patterns in the active region magnetic field in MDI images from the period surrounding the May 12, 1997 eruption. Notice the flow from the left dark (negative polarity) region converging on the neutral line adjacent to the white (positive polarity) sunspot. (From Li et al., 2004). Coronal simulations using the polytropic model typically do not reproduce the observed contrast between the fast ( km/s) and slow ( km/s) solar wind. To produce a more realistic solar wind in the coronal model, we have included the effect of solar wind acceleration from Alfven waves (Usmanov, 2000). The Alfven wave pressure is solved for in a separate equation using the WKB approximation (Jacques 1977). Figure 8 shows how the solar wind speed at high latitudes is increased by including an Alfven wave pressure source of approximately 0.1 dyne/cm 2.

19 Figure 8. Solar wind radial speed as a function of co-latitude and radial distance from two MAS coronal simulations. On the left, a polytropic simulation with no Alfven wave acceleration. There is very little contrast in the solar wind speed. On the right, a polytropic simulation that includes Alfven wave acceleration. Solar wind speeds as high as 750 km/s are produced in the high latitude regions. Figure 9 from Arge et al. (2004) identifies the likely coronal source of the solar wind stream structure associated with the May 12, 1997 CME. Interestingly, the stream following the interplanetary event or ICME is inferred to come from a southern hemisphere coronal hole extension, though the ejection apparently starts in the active region vicinity in the North. Figure 10 from Riley et al. (2004), based on the results of the previous year's ad-hoc CME simulation, illustrates how the in-situ appearance of the event depends on both the distortion of the ejecta in transit and on the location of the observer. This result provides a cautionary note regarding future analysis and validation of real CME event simulations. Figure 9. Illustration of the solar wind sources surrounding the May 12, 1997 event inferred from CISM modeling. (From Arge et al., 2004)

20 Figure 10a (above) and Figure 10b (below) Figure 10a. Cross section of CISM ad-hoc CME simulation, showing locations from which synthetic time series were derived. Figure 10b. Synthetic time series derived for the two different locations of the observer, suggesting the important dependence of the interplanetary signature on both the solar wind distortion of the CME and its path.

21 2B. Magnetospheric Thrust The goal of the magnetospheric thrust is produce a model of the magnetosphere that is capable of accurately describing the transmission of solar influences to the Earth's atmosphere as well as describing the internal state of the magnetosphere. The ionosphere and thermosphere are tightly coupled to the magnetosphere through current systems that run from the magnetosphere to the ionosphere, across both polar caps, and then back into the magnetosphere. Modeling this interaction is one of the primary goals of the magnetospheric thrust and is shared with the ionospheric thrust. Much of our effort is devoted to this M-I (magnetosphere-ionosphere) coupling. Within the magnetosphere itself the object is to provide models that can simulate the particle distributions and dynamics of the near Earth region. This means coupling a global MHD model for the magnetosphere to various other simulation codes with different, complementary physical domains of applicability. The global MHD code gives a good broad cut at the structure and dynamics, but leaves out a good deal of physics. We have identified three major areas in which to concentrate our efforts to extend the magnetospheric model during the initial phases of CISM: the ring current (through the Rice Convection Model, the RCM), the radiation belts, and the physics of magnetic reconnection. Activities: Our activities during the past year reflect our overall goals. We have developed a coupled global magnetospheric MHD (LFM) and ionosphere-thermosphere model (TING) and are subjecting the resulting code to testing and validation. The combination of the radiation belt and LFM codes have been used to study some important questions regarding energetic particle access and evolution. Mary Hudson at Dartmouth remains the director for the magnetospheric thrust. Chief responsibility for the LFM code and its use with the other magnetospheric codes lies with John Lyon. Frank Toffoletto at Rice is the lead on the RCM effort. The LFM-TING effort is led by Mike Wiltberger and Wenbin Wang at NCAR. The radiation belt effort has been led directly by Mary Hudson with Scot Elkington at Colorado, and Brian Kress and Kara Perry at Dartmouth. The exploration of enhanced models for M-I coupling has been the responsibility of Bill Lotko at Dartmouth. Active research in reconnection physics has been undertaken by Barrett Rogers at Dartmouth and Jim Drake and co-workers at Maryland. CISM MI-coupling researchers at BU, at Dartmouth (including graduate student John Gagne), NCAR, Rice and SwRI have formed a MI Coupling working group to coordinate development and validation of the empirical and physical transport models and code couplings that are required to address this set of problems. Monthly meetings of this working group occur via the Access Grid (see pacescience/wl/cism-mi-wg/ for meeting notes. Significant Accomplishments: We have accomplished the following tasks on the milestones developed for this year at the Sept All-Hands Meeting: LFM /TING development and coupling We have developed a robust version of the ad-hoc coupled LFM-TING code. The interface for general users is being tested and is expected to be given to the validation team before the end of the year with sufficient documentation for them to run the code without aid from the development team. We have continued to test the coupled LFM-TING code against idealized

22 solar wind conditions and observed sequences. We are developing the code necessary to add neutral wind effects to the LFM-TING code. RCM / LFM coupling The LFM code has been used to drive the RCM code in a one-way fashion. The Rice team is currently using this capability to investigate the development of the inner magnetospheric plasma for a number of solar wind conditions. They have also developed the code necessary to provide the necessary information to feed RCM data back into the LFM. The code for the twoway coupling, that is, modifying the running LFM code with RCM results, is being developed. Radiation Belt Two 3D radiation belt codes are in use at Dartmouth. The first uses a 3D guiding center motion formulation and has been used to study the resonant diffusion of energetic particles in a set of analytic fields. The second tracks the full Lorentz motion of the particles and has been used to determine the cutoffs associated with entry of solar energetic particles to the Earth. Work with equatorially mirroring particles has continued with a concentration on studying the effects of time dependent injection and ULF modal structure. Reconnection Physics: Studies of reconnection have been used to model regimes closer to actual conditions in the magnetosphere rather than the very simplified case studied in the GEM Reconnection Challenge. We have found that: 1. diamagnetic drift can act to stabilize reconnection at the magnetopause 2. External forcing of reconnection can lead to very fast onset of reconnection. The Maryland group has begun developing codes using the InterComm package. This package will form the basis of embedding reconnection codes within a larger scale MHD code. Other Projects A number of other projects needed for magnetosphere-ionosphere coupling are in the conceptual development stages. These include: 1. design of a generalized electrodynamic magnetosphere/ionosphere interface. 2. initiation of a plasmasphere model for inclusion to coupled geospace codes 3. development of diagnostics for the LFM results; the output of which can be used for an improved model of the auroral acceleration region. Research Highlights Inner Magnetosphere Modeling Global MHD models of the magnetosphere, such as the LFM, and multi-component kinetic convection models of the inner magnetosphere, such as the RCM, are, in some sense, complementary. For example, as a stand-alone model, the RCM has been shown to adequately reproduce such well-established inner-magnetospheric phenomena as shielding of the lowlatitude ionosphere from full effects of magnetospheric convection, the formation of field-aligned currents connecting the ring current to the ionosphere (region-2 currents), and particle dynamics during magnetic storms. One limitation of the RCM has been the use of a prescribed magnetic field in most cases that is not self-consistent with the computed particle pressure. The use of MHD-computed magnetic fields in the RCM ensures compatibility with a representation of the solar wind - magnetosphere interaction and tail dynamics. Global MHD models, on the other

23 hand provide magnetic fields computed self-consistently with the particle pressure. However, their neglect of energy-dependent gradient/curvature drifts results in unrealistic/missing region-2 current distributions and the lack of a realistic storm-rime ring current. Figure 11. Initial equatorial pressures (left) the final RCM-computed equatorial pressures (right-top) and the corresponding LFMcomputed temperatures (right-bottom) without the RCM (Toffoletto et al., 2004). For our initial effort at coupling the two codes, a one-way coupling experiment in which LFMcomputed values of electric and magnetic fields, ionospheric conductance, and magnetospheric plasma distribution were used to drive the RCM. The LFM was started with a northward interplanetary magnetic field (IMF) which then switched to southward at ~1200 seconds and continued southward for another three hours. The solar wind conditions for this run were a speed of 400 km/s, density of 5 particles/cc, and an IMF B Z = 5 nt, IMF B y =B x =0. The LFM provided the RCM with high-latitude boundary conditions on the plasma distribution function and the electrostatic potential; it also provided magnetic field values (for flux tube volume calculation and mapping) as well as ionospheric conductances everywhere in the RCM region. The data exchange scheme, described in more detail in the code coupling section of this report, involved exchanging data through an intermediate grid. After the RCM has run for a specified interval the RCM computed MHD quantities are returned to the intermediate grid ready for ingestion by the LFM. Figures 11 shows the initial pressure, and subsequent pressure and temperature from the RCM and LFM codes, respectively, with magnetic field specified in the overlapping RCM volume by LFM, about 3 hours after the southward turning of the IMF. Outside the RCM modeling region, the RCM parameters are set equal to the MHD values, so the plots automatically agree there. The MHD code warmed the plasma sheet considerably in response to the southward turning, and generally raised the temperature and pressures to almost realistic levels. The RCM values for T and p exceed MHD values by an order of magnitude or more near geosynchronous orbit. The shapes of the inner-magnetospheric temperature and pressure distributions are also quite

24 different, with the RCM showing strong radial gradients and weaker azimuthal variations. The RCM s inner magnetospheric pressures (and densities) are higher than the MHD values mainly because the RCM s numerical method rigorously enforces the adiabatic compression requirement. The RCM s local-time-smoothing results in part from transport by gradient/curvature drift, which is included in the RCM equations but not in MHD. In addition, the RCM has clearly formed a partial ring current centered near geosynchronous orbit as well as strong region-2 currents. This is classic RCM response to a period of steady strong convection. The next step is to initiate feedback of the modified pressures and densities in the LFM. We expect that the increased pressures and densities, as specified by the RCM, may cause a significant inflation of the magnetic field, mostly on the nightside. Radiation Belt Codes The Dartmouth-Colorado Radiation Belt Codes are relativistic trajectory-tracing codes which extract electric and magnetic fields from the LFM-MHD simulations of the magnetosphere and advance position following either guiding center test particle trajectories (electrons, Figure 12a) or the full gyromotion in 3D (ions, Figure 12b). The goal is to simulate the time dependent fluxes of outer zone relativistic electrons as well as the access and trapping of Solar Energetic Particles (SEPs), which accompany interplanetary shocks, entering the magnetosphere through the high latitude cusp region. With gyroradii comparable to the radius of curvature of the magnetic field, SEPs can become trapped and contribute to the inner zone with the arrival of an interplanetary shock originating at the sun as a coronal mass ejection. Figure 12a: Snapshot of plasma sheet electrons convecting earthward from 20 R E, a portion of which become trapped and energized as part of the outer radiation belts around L=3 (Elkington et al., 2004).

25 Figure 12b. Modeled 25 MeV proton cutoff, sunward +x, at a snapshot in time from LFM-MHD simulations using input data from WIND L1 measurements. Boundary of shaded region is cutoff surface at time just before/after solar wind pressure pulse arrival (Kress et. al, 2004; Hudson et al., 2004). Significant progress has been made in modeling equatorial plane dynamics of radiation belt electrons, as shown in Figure 12a from an equatorial plane simulation of a plasma sheet electron source population (10 s of kev) transported into the inner magnetosphere during a strong geomagnetic storm, modeled with LFM-MHD fields driven by measured upstream solar wind conditions. Incorporation of an injection boundary condition from the nightside plasma sheet region has been accomplished in the first 18 months of CISM. Modeling radial transport and trapping of SEPs requires extension to 3D, as will modeling the balance between source and loss processes for relativistic electrons. A preliminary study of electron transport in a 3D analytic model of interaction with ULF waves, which have been shown to be correlated with relativistic electron flux enhancements, is underway as part of Dartmouth PhD student Kara Perry s thesis. Related work to that shown in Figure 12a is being carried out by Rice PhD student Yue Fei. In MHD-particle simulations of radiation belt electrons during the September 1998 storm, the evolution of the radial flux profile appears to be diffusive, and ULF waves have been invoked as the probable diffusion mechanism. In this work, led by Anthony Chan (Rice University) and Scot Elkington (University of Colorado), in collaboration with Yue Fei and Michael Wiltberger (NCAR), a radial diffusion equation with ULF-wave diffusion coefficients and a time-dependent outer-boundary condition is solved to better quantify the radial diffusion coefficient. Completion of both PhD projects is expected by Sept 04. The prompt trapping of Solar Energetic Particles (SEPs) in the inner magnetosphere around L=2-2.5, including protons and heavier ions, has been observed at both the Cycle 22 and 23 solar maxima, in association with high speed interplanetary shocks and Storm Sudden Commencements (SSCs). Recent observations include the Bastille Day 2000 CME-driven storm

26 as well as two in November 2001, which produced a long-lived new proton belt, as well as trapping of heavy ions up to Fe in all three cases. A survey of such events around the most recent solar maximum indicates similarities to the well-studied March 24, 1991 SSC event. In this event, electrons and protons in drift resonance with a magnetosonic impulse were transported radially inward. Arrival of the interplanetary shock compresses the magnetosphere and changes the cutoff from its nominal value around L=4, as evident in Figure 12b, which corresponds to the arrival time of a high density pressure pulse at the magnetosphere. This effect was modeled using a 3D Lorentz integration of SEP trajectories in electric and magnetic fields taken from the LFM-MHD code, with solar wind input parameters taken from spacecraft measurements upstream from the bow shock, for the November 24, 2001 SEP event. The results indicate that an enhancement in solar wind dynamic pressure for this event plays a role in the observed injection of ions to low L-values, to form a new proton belt which has lasted for more than two years. Postdoc Brian Kress will shift focus to the SEP trapping mechanism, while developing 3D relativistic electron modeling in LFM fields in year 3. Magnetosphere-Ionosphere Coupling CISM research into magnetosphere-ionosphere coupling is currently following two trajectories that are expected to intersect within 1-2 years. One trajectory, as described in more detail elsewhere in this report, is proceeding initially as a numerical effort to couple the existing LFM, RCM and TING codes. The second trajectory is a scientific investigation with the primary objective of developing simple, lumped transport models to be embedded within the coupled large-scale models. These transport models characterize collisionless plasma processes that arise in the topside ionosphere and low-altitude magnetosphere, which impact the large-scale dynamics and energetics of the coupled MI system by converting electromagnetic power into field-aligned beams of electrons and upward flows of ions. The formation of parallel electric fields in the low-altitude magnetosphere produces electron beams that carry large energy fluxes into the ionosphere and magnetosphere. The downgoing beams, in particular, cause aurora and modify the ionospheric conductivity and the electrodynamics of MI coupling. Topside collisionless ion heating causes massive outflows from the ionosphere. The stratification of the ionosphere is modified in the process, and an inertial coupling is imparted to the MI interaction. The effects of these processes either are neglected or, as currently implemented, are of limited applicability in the existing large-scale models of the magnetosphere and ionosphere. The nearterm goal of this work is to improve the existing electrodynamic coupling between the LFM, TING and RCM models and, for the first time, to implement inertial coupling between the models.

27 Figure 13: Flow chart diagram of the coupling between the LFM, RCM and TING, illustrating how an M-I coupler module would link these code together. The M-I Coupling working group is currently addressing a wide variety of issues, summarized in the accompanying figure and list of topics, which include: 1. Facilitating cross-disciplinary understanding of the physical bases of the LFM, RCM and TING models, including their implementation as numerical algorithms both as uncoupled and coupled models; 2. Identification of thermospheric and MHD variables to be used as causal inputs in lumped models of collisionless plasma transport (enhanced energy fluxes carried by electrons, ion heating and mass outflow), together with identification of the thermospheric and MHD variables to be modified by the resulting lumped models; 3. Relaxation of artificial boundary conditions at code interfaces; 4. Streamlining and improving the MI coupling algorithm at the heart of all three models (the MI coupler in the above flowchart); 5. Interpretation of coupled model results, code validation and validation of model results against measurements; 6. Identification of candidate products and metrics for eventual transition to CISM partners at NOAA/SEC; 7. Development of software and data protocols for exchanging variables and data streams between the LFM, RCM and TING codes and for visualizing and interpreting results; and 8. Scientific investigations into MI coupling enabled by new modeling capabilities arising from the coupled codes. The physical basis for the coupling between the LFM global magnetospheric model and the TING ionosphere-thermosphere model assumes that the time scales of interest are sufficiently long (>100 seconds) that the low-altitude electrodynamics may be treated as electrostatic. In developing extensions of the models to include mass exchange, particularly ion outflow from the

28 ionosphere to the magnetosphere, the coupling must be extended to treat faster processes. A new coupling scheme including the effects of electromagnetic induction at low altitude has been developed. Initial steps in implementation are in progress. We have also developed new diagnostics to characterize the electromagnetic energy flux flowing through the low-altitude boundary of the LFM model. For purposes of validation leading to model improvement, we will also be comparing the statistical behavior derived from this diagnostic with recently reported observations of electromagnetic power flowing into/out of the high-latitude ionosphere. Reconnection Code Development A significant physics discovery over the past several years is the importance of including the coupling to dispersive whistler and kinetic Alfven waves in boundary layers where magnetic reconnection takes place. The complication is that the correct treatment of these waves requires the resolution of small scales, which are normally not treated in global simulations. On the other hand, the resolution of small scales is only required in rather small regions of space. The computational challenge is therefore to develop "adaptive physics" techniques in which adaptive mesh refinement is combined with the inclusion of a kinetic physics model as small scales develop in large-scale MHD modeling. Graduate students Nathaniel Ferraro and Brian Sullivan have been engaged in research on the topic of magnetic reconnection over the past year under the supervision of Prof. B. Rogers. Nathaniel is a former Dartmouth undergraduate who completed a MS degree in one year at Dartmouth and is now in the PhD program in plasma physics at Princeton. His Dartmouth thesis research broke fundamental new ground in the theory and simulation of magnetic reconnection in low plasma pressure environments. Dartmouth PHD student Brian Sullivan is engaged in research to understand the physics of collisionless reconnection in externally driven systems. His numerical simulations have shown that external forcing can lead to the sudden onset of fast reconnection, and his present goal is to understand the physics behind the triggering process, and the conditions under which it occurs. During the past year, Jim Drake, Mike Shay and graduate student Paul Cassak successfully coupled together two independently running two-fluid simulations using InterComm and passed a whistler wave without distortion back and forth between them. Both simulations evolved the same fluid equations and had the same grid scale and time step. During the next phase of our study, they will couple together simulations with disparate grid scales and different physics. In such a coupling, there is a strong possibility of anomalous reflections of waves from the boundaries. The goal will be to determine under what conditions the spurious reflections can be kept to a minimum. Using this knowledge, they will then begin to couple reconnection "modules" inside of large scale MHD codes such as LFM. Goals and Activities for the Coming Year: In many ways our concentration will be on what can be considered the three base models that are strongly coupled together: RCM, TING and LFM. Our goals for these models for the coming year are: 1. Use the coupled LFM-TING model for validation and scientific studies 2. Develop the coupled RCM-LFM code to the point where it can be handed to the validation team. 3. Develop the three way coupled RCM-LFM-TING code

29 The key goal for the radiation belt effort will be to determine ways of describing the loss processes for the energetic particles and tying those to the data available from the coupled RCM-LFM-TING model. In particular, determining the ULF mode structure of the magnetosphere and the effects of these modes on the acceleration of particles will be an important part of the effort. The largest remaining limitation in the magnetospheric model is the lack of ionospheric outflow. We will begin modeling this outflow in an exploratory program. The intent is to test the effects for various models for outflow on the magnetosphere.

30 2C. Ionospheric/Thermospheric Thrust The ionosphere-thermosphere modeling segment of CISM is primarily housed at the National Center for Atmospheric Research, but with affiliations to the Space Environment Center at NOAA, the University of Colorado, Southwest Research Institute, Utah State University, and Space Environment Technologies. Primary modeling tools include the NCAR Thermosphere General Circulation Models, auroral particle and photoelectron transport models, middleatmosphere tidal and planetary wave models, and the Assimilative Mapping of Ionospheric Electrodynamics (AMIE) procedure for analyzing auroral region currents and conductances using a variety of measurement data. The particular model currently being used for CISM studies is a high-resolution version of the NCAR-TGCM known as the Thermosphere- Ionosphere Nested Grid (TING) model. The ionosphere, that small percentage of the upper atmosphere that exists as charged particles, is created and maintained mostly by solar extreme-ultraviolet radiation. However, its variability on daily and shorter time scales is largely driven by processes controlled by the solar wind and magnetosphere and coupled to the ionosphere through the auroral regions. Many other smallscale forms of ionospheric variability, such as irregularities in the equatorial region and traveling disturbance waves, are also important, but are less accessible at this time to global-scale thermosphere-ionosphere models. Therefore, the initial goal of the CISM project for the geospace regions is to create a coupled model of the magnetosphere and ionosphere that includes upper atmosphere circulation, solar irradiance variation, and forcing by the lower atmosphere. Significant Accomplishments The following milestones were identified as primary objectives for the Ionosphere- Thermosphere modeling thrust during the second year of the CISM project. It is anticipated that all of these will be completed by the conclusion of Year-Two (8/1/04). One milestone, validation of the LFM-TING coupled model, has been slipped to Year-Three because, although validation studies are in progress at NCAR, it is unlikely that the documented coupled model will be delivered to the validation team until near the end of Year-Two. Perform run of LFM-TING for end-to-end simulation - completed Develop robust ad-hoc interface for LFM-TING coupled model - completed Initiate development of three-way LFM-RCM-TING - completed Include neutral wind and co-rotation feedback in LFM-TING - being evaluated Include new solar EUV module in TIE-GCM - completed Test, document, and diagnose LFM-TING coupled model -in progress (exp July 2004) Develop unified potential solver algorithm - in progress Initiate plasmasphere model development - initiated Initiate ionospheric data assimilation project - initiated Explore boundary conditions for ionospheric outflow - in progress Goals and Activities Planned for the Coming Year The following milestones are the tentative goals for Year-Three, subject to discussion and evaluation at the all-hands meeting in September. Validate LFM-TING coupled model Link TING with LFM and RCM forming the coupled geospace model

31 Implement unified potential solver algorithm Transition from TING to TIE-GCM (inclusion of dynamo and potential solver) Transition coupled codes from ad-hoc coupling to OOP I Continue development of plasmasphere model Perform initial data assimilation collaborative tests Continue validation and metrics studies The most significant new partnership we anticipate during Year-Three will be collaboration with the Global Assimilation of Ionospheric Measurements (GAIM) project. As an initial experiment, we have proposed to use an ionospheric state estimate obtained using the GAIM method as an initial condition for a LFM-TING simulation using measured solar wind inputs, for comparison with a simulation using standard diurnally-reproducible initial conditions. Magnetosphere-Ionosphere Coupling Coupling between the LFM and TING models has been implemented using an ad-hoc method accomplished through exchange of information via a series of interchange files. One-way coupling, which consists of the LFM specifying the auroral energy flux, characteristic energy, and electric potential distribution to provide the high-latitude upper boundary condition to TING, was implemented during Year-One. Two-way coupling, which adds the feedback of ionospheric conductivities calculated by TING to the LFM, was then applied. This exchange of boundary conditions occurs once per TING time step of two minutes. During the intervening time the LFM uses the previous TING conductivities to calculate the ionospheric potential. Grid rotation and interpolation is also handled by the magnetosphere-ionosphere coupling module. During Year-Two, initial problems in the implementation of the two-way coupling methodology that produced unrealistically low ionospheric conductivities were resolved, and the two-way coupled model now performs well. Results for the CISM end-to-end run, described in Luhmann (2004) are shown in Figures 14 and 15. Papers by Wiltberger et al. and by Wang et al., also submitted to JASTP, describe the details of the methodology and initial runs for parametric solar wind conditions. The ad-hoc coupled model has been installed on the NCAR supercomputer system and an operational interface was developed to facilitate test runs. Validation studies to examine the intensity and morphology of auroral precipitation are in progress. The next step in magnetosphere-ionosphere coupling is inclusion of neutral wind and co-rotation effects from the thermosphere on the ionospheric potential, and feedback of that potential to the magnetosphere. A method of calculating the incremental current system due to this interaction has been implemented and is being tested in the potential solver.

32 Figure 14. Parameters representing the global response (northern hemisphere) to a simulated ICME displayed together with key solar wind parameters at L1. Top to bottom: solar wind density, cm -3 ; solar wind velocity x-component, km s -1 ; IMF y and z components, nt; cross-tail potential, kv; hemispheric Joule heating, GW; hemispheric power of precipitating electrons, GW; simulated AU and AL indices, nt. The vertical dashed lines represent the times of the model output shown in figure XXX.

33 Figure 15. Response of the thermosphere/ionosphere system calculated by the LFM/TING coupled model at 6, 12, and 22 UT. (a,b,c): E-region electron densities (log 10 cm -3 ) at ~120 km with the ion drift pattern superimposed. (d,e,f): F-region neutral temperatures (K) at ~250 km with the neutral winds superimposed.

34 Ionosphere-Thermosphere Model Development Transition from the TING model to the TIE-GCM is planned for Year-Three. Preparation for this step requires two new developments: a high-resolution version of the TIE-GCM, and implementation of a global solution of the ionospheric potential. The first of these is in progress using a workstation-based test version of the TIE-GCM. The second involves development of an algorithm that calculates the ionospheric electric potential on a global basis, imposing hemispherical symmetry at low-to-mid-latitudes (as in the standard TIE-GCM method) but allowing hemispherical asymmetry at high latitude. A test version of this has been designed and tested in the TIE-GCM and will be tested with LFM inputs to the auroral regions during the remainder of Year-Two. A new algorithm for calculation of solar ionization, including accurate parameterization of photoelectron impact ionization, has been developed and incorporated into the TIE-GCM. This module also has the capability of using measured solar fluxes as input, e.g., from the TIMED and SORCE spacecraft. A short-term (1-3 day) forecast capability using projected indices available from NOAA/SEC and Space Environment Technologies to drive a solar proxy model is under study. Plasmasphere Model Development The NCAR TIE-GCM does not yet include the plasmasphere, other than as a parameterized upper boundary condition. A study of possible approaches for development of a plasmasphere model for CISM has been initiated. The current plan is to couple a version of the SUPIM model to the TIE-GCM. Results of a study using Rice Convection Model (RCM) circulation to drive an extended-altitude SUPIM description of the ambient plasmasphere to simulate observations of the plasmapause location will be presented at the spring AGU/CGU joint assembly, Scientific Studies using Ionosphere-Thermosphere Models Burns et al. (2004) used the high resolution TING model to study aspects of the behavior of the thermosphere that are relevant to the coupling problem. The model run was a simulation of the Bastille Day, 2000 geomagnetic storm. The main conclusions of this study were that a tongue of neutral composition is formed, in much the same way as a tongue of ionization, when parcels of air that are rich in atomic oxygen are drawn from the dayside by the anti-sunward winds associated with the neutral convection pattern and transported across the polar cap towards the night side auroral oval. This neutral tongue tends to be weaker than the ion tongue, it extends a shorter distance across the polar cap and it takes longer to form, but when enough time has elapsed to allow a neutral tongue to form in conjunction with the tongue of ionization, the latter becomes both stronger and longer. An initial study of solar flare effects on the ionosphere and thermosphere during the October/November 2003 disturbances was conducted and the results presented at the fall 2003 AGU meeting. Measurements of a series massive X-class flares by the TIMED and SORCE spacecraft were used to drive TIE-GCM ionosphere/thermosphere simulations together with index-driven auroral specifications. The flare effects on ionospheric densities and thermospheric temperatures were found to be similar in magnitude but shorter in duration when compared to the auroral forcing, and less effective in causing dynamical perturbation to the upper atmosphere system.

35 2D. Code Coupling Thrust The Code Coupling thrust is responsible for coupling the component models into successive versions of the comprehensive CISM model. In order to do this, this thrust has to define the code coupling methodology and the specific computational science tools that will be used to couple the component models together, and develop or adapt, as needed, these tools for our specific use. This thrust must then apply these tools to produce functioning versions of the comprehensive CISM code that can then be passed on to the model validation and metric thrust for testing. Finally the code coupling thrust is responsible for developing and implementing the code and data handling plans for CISM as well as maintaining the central elements of that plan, the code repository and data archive at BU. The code coupling thrust plays a critical, core role within CISM. It is led by Charles Goodrich. The members of the thrust include computer scientists Alan Sussman and Henrique Andrade, and numericists William Abbett(S), Jon Linker(S), Zoran Mikic(S), Roberto Lionelllo(S), Dusan Odstricil(S), James Drake(M), Scot Elkington(M), John Lyon(M), Frank Toffoletto(M), Michael Shay(M), Stan Sazykin (M), Robert Spiro(M), Michael Wiltberger (M), Alan Burns(I), Stanley Solomon(I), and Wenbin Wang(I), as well as postdocs and students at most CISM sites. The graduate students Paul Cassak and Timothy Guild are involved particularly closely. The numericists, postdocs, and students are also members of the other science thrusts, the Solar /Heliosphere thrust (S), the Magnetosphere thrust (M), or the Ionosphere-Thermosphere thrust (I). The strong overlap of the coupling team with the other thrusts highlights our close integration with them. This interaction and overlap is clearly evidenced the following sections of this report. We work very closely with the Model Validation and Metrics thrust to assure the scientific accuracy of the codes, both individually and coupled, as well as the accuracy and utility of the code documentation delivered with the codes. We also support the release of new versions of the comprehensive CISM code to the public and operational and scientific communities through our partner institutions, CCMC and NCSA, by the knowledge transfer team and support use of models in an educational context by the education team. Our plan, which is described by Goodrich et al. (2004), is to produce an ever-improving series of comprehensive models as is shown conceptually in Figure 1. Approximately every two years we will produce a new generation model based on increasingly sophisticated coupling technology. Within each generation of coupling technology, new versions of the comprehensive model will be released that incorporate new or more sophisticated and comprehensive physics as new component models become available from the science thrusts. Each new version will be transferred to the validation team for a period of rigorous testing lasting up to a year. The codes will then be released to the scientific and operational communities and the public in cooperation with the Knowledge Transfer and Education thrusts. While Figure 1 gives a good overview of our plan, two important aspects require further elaboration. First, the initial development of the coupling of all code combinations will be done using most convenient methods available, which generally will include ad hoc ones. As these ad hoc coupling codes are completed and transferred to the validation team, we will also attempt to implement them in our first generation framework software, based on the Overture and InterComm packages. We will also proceed to attempt ad hoc coupling of more complex code combinations. Using the geospace codes as an example, we have completed this year the ad hoc pair-wise coupling of the magnetospheric and ionosphere-thermospheremesosphere (LFM and TING) codes, and plan to implement this coupling using InterComm and

36 Overture this summer and fall. Next year we will begin the ad hoc three-way coupling of LFM, TING, and RCM codes. Thus coupling efforts using both ad hoc and framework methods will continue cooperatively in parallel. The second aspect is that we are writing and evaluating code documentation together with the codes themselves as we transfer completed model versions to the model validation team. All versions will include the documentation necessary to use and understand the codes. The model validation team will evaluate the usefulness of the codes and documentation - making suggestions for improvement in both to the code developers as required. The transfer is complete only when the validation team is satisfied. Our plan, which is described in Goodrich et al. (2004), is to produce an ever-improving series of comprehensive models as shown conceptually in Figure 1. Approximately every two years we will produce a new generation of models based on increasingly sophisticated coupling technology. Within each generation of coupling technology, new versions of the comprehensive model will be released that incorporate new or more sophisticated and comprehensive physics as new component models become available from the science thrusts. Each new version will be transferred to the validation team for a period of rigorous testing lasting up to a year. The codes will then be released to the scientific and operational communities and the public in cooperation with the Knowledge Transfer and Education thrusts. While Figure 1 gives a good overview of our plan, two important aspects require further elaboration. First, the initial development of the coupling of all code combinations will be done using most convenient methods available, which generally will include ad hoc ones. As these ad hoc coupling codes are completed and transferred to the Validation team, we will also attempt to implement them in our first generation framework software, based on the Overture and InterComm packages. We will also proceed to attempt ad hoc coupling of more complex code combinations. Using the geospace codes as an example, we have completed this year the ad hoc pair-wise coupling of the magnetospheric and ionosphere-thermosphere-mesosphere (LFM and TING) codes, and plan to implement this coupling using InterComm and Overture this summer and fall. Next year we will begin the ad hoc three-way coupling of LFM, TING, and RCM codes. Thus coupling efforts using both ad hoc and framework methods will continue cooperatively in parallel. The second aspect is that we will build the development and evaluation of code documentation and well as the codes themselves into the process of transferring completed model versions (for all coupling methods) to the Validation team. All versions will include the documentation necessary to use and understand the codes. The Validation team will evaluate the usefulness of the codes and documentation - making suggestions for improvement in both to the code developers as required. The transfer is complete only when the Validation team is satisfied. Significant Accomplishments: The performance indicators for the Coupling thrust are the CISM 5 year Milestones and Goals, as well as the more detailed elaboration of them developed each year at our All Hands Meeting. As indicated below, we have achieved most all of our goals in this year. The LFM-RCM coupling has progressed somewhat more slowly than anticipated, due primarily to the complexity of the two-way overlapping volume coupling required. We have taken steps to increase our effort on this problem and expect to see significant progress in the near future. We have also delayed work on the Second Generation (Object Oriented) framework design to next year, when the

37 development in other related projects in which we are interested, such as the Earth System Modeling Framework (ESMF), will be mature. Pair-wise ad hoc coupling of CISM core codes - completed o CMIT full two way coupling of LFM and TING o CORHEL Coupling of MAS and ENLIL codes. Due to the unresolved complexity of determining and incorporating transient solar boundary condition for MAS, CORHEL is limited to simulation of ambient solar wind structure. Preliminary ad hoc incorporation of solar active region structures into MAS -completed One way ad hoc coupling of LFM results into the RCM code in progress Refinement of Object Oriented framework software, initial testing of its basic elements, and first application to core CISM codes mostly completed o Upgrade data communication services to InterComm library o Basic tests InterComm and Overture utility and performance o Application of InterComm and Overture to CMIT coupling late summer or early fall of 2004 Implementation of archive for source code and documentation version control, and data storage at BU (re.bu.edu) - completed o Concurrent Version System (CVS) for source code and documentation control. CMIT and CORHEL committed to CVS repository o Storage Resource Broker (SRB) for binary data storage implemented at BU. The SRB archive currently holds Simulation results Observational data for code Validation and Metrics Code distributions in tar and gzip formats Goals and Activities Planned for the Coming Year Complete LFM-RCM ad hoc coupling Complete three-way ad hoc geospace coupling: LFM, RCM, and TING Couple 3D radiation belt code to LFM Transition coupling in CORHEL and CMIT to our First Generation Framework technology Begin design of our Second Generation (Object Oriented) Framework. Evaluate possible use of other development such as the ESMF

38 Ad hoc (Generation 0) Coupling We have focused this year on developing and testing ad hoc couplings of pairs of the core CISM models. Sub-photosphere-Coronal Coupling Understanding how flux emerges in active regions from below the photosphere into the corona, and how this flux disperses during the lifetime of an active region, is critical to understanding how and why coronal magnetic fields erupt to produce coronal mass ejections. Investigating these processes is also important for determining the best methods for using photospheric vector magnetic field measurements needed to simulate active region evolution and possible eruption. While we would like to follow numerically the entire process of the formation and ascent of isolated flux systems through the turbulent convection zone to their emergence into the corona, it remains computationally intractable to simulate the entire solar interior and atmosphere in one calculation. The physical characteristics of the plasmas and the characteristic timescales between the interior and the atmosphere are vastly different; the plasma β (ratio of the plasma thermal pressure to the magnetic pressure) is typically << 1 in the lower corona, while β >> 1 in the convection zone. To understand the interaction of the solar interior and corona, Abbett, Mikic, and Linker are working to couple two separate numerical models individually designed to efficiently describe their respective domains. To model the complex 3D interaction of magnetically buoyant flux systems (flux tubes and Ω-loops) during their ascent through the convective envelope, we use the U.C. Berkeley code ANMHD to numerically solve the 3D system of MHD equations in the anelastic approximation (Fan et al. 1999; Abbett et al. 2001). This model generates realistic sub-surface structure for emerging active regions with self-consistent flow fields that can be used to drive a model of the solar atmosphere. For the coronal region, we use a Cartesian version of the SAIC coronal model (suitable for localized modeling of solar active regions). At this stage, our primary goal in coupling the ANMHD model to the SAIC coronal model is to explore the topology and properties of the emergent magnetic field. Therefore, in these initial cases we also made the further simplification of using the zero-beta equations (pressure is set to zero, density is set to a constant value, and only the momentum equation and Faraday s law are solved).

39 Figure 16. Emergence of a bipolar active region into a weaker background arcade field, viewed from above. The magnetic flux is contoured on the surface with red (blue) regions corresponding to outward (inward) pointing magnetic fields. In the left frame, the emerging field and the arcade field have the same polarity. In the right frame, the emerging field has polarity opposite to the background field. A true coupling of the two models not only requires the coronal model to respond to evolution in the sub-photosphere, but also the sub-photospheric model to respond to changes in the corona. In practice, the high inertia of the photosphere will tend to resist reaction to coronal changes while the corona responds readily to changes to the magnetic field footpoints. Therefore, we drive coronal solution by supplying the upper boundary values from the sub-photospheric solution, but we do not feed back this information to ANMHD at this time. However, even in this approximation, appropriate boundary data must be specified (for example, supplying all of the MHD quantities over-specifies the problem). We extract the tangential electric field (E x and E y ) and vertical component of the flow velocity (V z ) from the upper boundary of the subphotospheric simulation (considered to be the plane z = 0 in the coronal simulation). This data is supplied the boundary conditions for the coronal simulation in the domain z > 0, in which we calculate the evolution of the coronal magnetic field. In the first year we demonstrated this coupling technique for the case of magnetic field emerging into a field-free corona. We have now extended these techniques to drive a model corona in the more difficult (and interesting) case where the new magnetic flux emerges into a pre-existing arcade. Figure 16 shows the contrasting results when a new active region emerges into a pre-existing arcade with the same (left frame) or opposite (right frame) polarity. Note the complicated field topology associated with the opposite polarity case. Some of the field lines along the neutral line separating the strong and weak flux regions show dips, concave upward portions of the field thought to be required to support prominence material. See Abbett et al. (2004) for a detailed discussion of the methodology and results of these calculations. Solar/Heliosphere Coupling We have made great progress in coupling the solar corona and heliosphere. Riley, Linker, Mikic, and Odstrcil have completed this year the first coupled model of the solar corona and

40 heliosphere. The CORHEL (CORona-HELiosphere) model, which combines the SAIC coronal model (MAS) with NOAA/SEC's heliospheric code (ENLIL) has been delivered for use by the Validation and Metrics thrust. The model including the base codes and documentation have been committed to the BU CVS repository. Due to the complexity of modeling solar active regions, discussed in detail above, the use of CORHEL is confined to the following approximations and conditions: 1. Steady-state (i.e., equilibrium) solutions. Both MAS and ENLIL solve the time-dependent MHD equations. The equations are integrated in time until the solution settles to equilibrium. 2. Sequential coupling, i.e., ENLIL runs after MAS terminates; 3. Polytropic solutions. To obtain realistic solutions for the solar wind velocity and density, only the MAS magnetic field solution is passed to ENLIL. The speed, density, and temperature are specified at the boundary between the two codes according to the methods outlined by Riley et al. (2001). CORHEL is run via a C-shell script with two arguments, resolution and Carrington Rotation, i.e., -> CORHEL resolution CR. Two resolutions are currently supported, low and medium, which correspond to 61x71x64 and 101x101x128 in MAS, and 100x30x90 and 200x60x180 in ENLIL; we plan to implement higher resolutions in the future. We suggest that each user run initially model at low resolution and if sufficient detail in the solution is lacking, the case be rerun at medium resolution. The Carrington Rotation (CR), a 4 digit number, specifies the Carrington Rotation to be modeled for the magnetogram processing program (KPSYN) that supplies the photospheric boundary conditions for CORHEL. In addition, the MAS and ENLIL codes have a number of individual parameters that can affect details of the solution, which knowledgeable users can adjust through initialization files used the codes. The default values of these parameters for CORHEL have been chosen with the goal of robustness, at the expense of producing solutions that may wash out some of the smaller scale structure. We note that the use of smaller values of viscosity and resistivity can combine with the noise in magnetograms processed with minimal filtering to yield solutions with more sharply defined features. However, these solutions have greater risk of numerical problems because these structures may be under-resolved at the chosen mesh size and produce unstable oscillatory behavior. This is even possible even for the default parameters we have chosen, particularly for solar maximum cases. In future releases, we may enable the user to specify the code parameter values via a GUI interface. When run, CORHEL performs the following operations: 1. Creates the appropriate directory structure for the run. Runs are stored in the run directory and are named according to the Carrington Rotation. 2. Queries the NSO online database and retrieves the appropriate Kitt Peak synoptic data set. At present, NSO data is absent after Carrington Rotation 2007 (September, 2003), with the new SOLIS facility coming on line. Future data will be available from SOLIS (see 3. Preprocesses these data using the program KPSYN to provide a suitable input for SAIC coronal model, MAS. This step involves several steps, such as removing any monopole contribution, along with appropriate smoothing and filtering the data. At present, the parameters are chosen to produce a highly smoothed magnetic map. 4. Generates an appropriate input parameter file. Suitable values for these parameters are chosen automatically. 5. Runs the MAS program.

41 6. On termination of MAS, CORHEL post-processes the MAS output to provide suitable input for ENLIL. Magnetosphere/Ionosphere/Thermosphere Coupling (LFM/TING) The magnetosphere and upper atmosphere of the earth are closely linked at high latitudes through electric currents and particle flows from the magnetosphere that heat and increase ionization of the ionosphere and neutral atmosphere, which in turn support the electric potentials in the polar caps which control conditions at the base of the magnetosphere. The coupling is nonlinear and inherently bi-directional in that these potentials depend on Hall and Pederson conductivities of the ionosphere which depend on the electric potential itself; the potentials and conductivities both depend on the field aligned currents and particle precipitation from the magnetosphere. It is further complicated by the large difference in characteristic time scales for dynamic change of the magnetosphere (fraction of seconds) and ionosphere (minutes). Wiltberger and Wang have addressed these challenges to complete our first coupled geospace model CMIT (Coupled Magnetosphere-Ionosphere-Thermosphere-Mesosphere). CMIT couples the LFM global magnetosphere code with the TING Ionosphere-Thermosphere-Mesosphere code. A detailed description of CMIT is given in Wiltberger et al. (2004). The coupling between the LFM and TING model involves the LFM passing fluxes and characteristic energy precipitating particles generated in the magnetosphere as well as the high latitude convection pattern to the TING model. TING uses this information to determine new ionospheric conductivities, which are used to recalculate the high latitude electric potential. The potential is used by the LFM to advance the magnetospheric solution.

42 Figure 17: Schematic representation of the LFM TING coupling process. Arrows represent data flow with parameters passed between modules indicated next to each arrow. The LFM passes the FAC current (J ), characteristic energy (ε) and number flux (F) of electrons to the MI coupler. The MI coupler passes the polar cap potential (Φ) as well as ε and F to the TING code which returns the Pedersen (Σ P and Hall (Σ H ) conductivities to the MI coupler. The MI coupler than recalculates Φ and returns it to the LFM. Figure 17 shows a schematic representation of CMIT indicating its three component parts. The LFM model, which for the purposes of this diagram, is limited to the magnetospheric portion of the simulation. The magnetosphere - ionosphere (MI) coupler includes the ionospheric portion of the LFM model and infrastructure for communicating between the LFM and TING models. As a practical matter the code for the MI coupler code currently is embedded within the LFM. The TING model is essentially unchanged with the exception of code added for communicating with the MI coupler. In operation, the MI coupler uses the LFM empirical relationships for particle precipitation to make an initial determination of the ionospheric electric field. The MI coupler then makes the time dependent transformation of these parameters from the LFM ionospheric grid to the TING grid, which rotates with the earth, which TING then uses to determine values for the height integrated Hall and Pedersen conductivities or conductances as described in Wang et al. (2004). These new conductances are used by the MI coupler to recompute the electric field, which is passed to the magnetospheric portion of the LFM to use as part of its inner boundary condition. The large disparity in computational time between the LFM and TING models brings up another important aspect of the coupling between these models. During the 2 minute interval between

43 exchanges of information between the MI coupler and the TING model the LFM needs revised ionospheric electric fields. In this time frame the MI coupler uses the conductances from the previous step to compute the electric field based upon the field aligned currents. The current implementation has the MI coupler waiting for the TING model to complete its work before proceeding, but it would be a simple matter to have this process executing concurrently with the LFM by having the MI coupler lag one data exchange behind in the Hall and Pedersen conductances it uses. This coupling approach results in a consistent set of conductances and electric fields being used within both components of the CMIT model. At this time the coupling methodology uses a series of lock and data files to create a fully automatic mechanism for communications between the components of CMIT. To summarize, CMIT in essence replaces the empirical model for the ionospheric Hall and Pedersen conductivities in originally in LFM from with physical values from TING and replaces the parameterized models for ionospheric electric field and aurora precipitation in TING with parameters determined with a higher time resolution by the LFM. It further allows for the inclusion of the ionospheric dynamo effect in both CMIT is currently undergoing intensive testing by members of the Magnetosphere and ITM thrusts to prepare for its release to the Validation and Metrics thrust. An example of this testing is summarized in Figure 18 taken from Wiltberger et al. (2004). The figure contrasts the northern hemisphere Joule heating determined by the LFM and CMIT during an ideal sequence of solar wind conditions in which IMF direction was rotated every four hours while the solar wind velocity and density were held constant. While this comparison shows a general qualitative agreement between the models, it also shows the first indications of interesting differences in the physical response of the system. It is quite clear that the CMIT model shows significantly more variability during the entire interval, especially during the periods of northward and eastward IMF, with numerous peaks in the Joule heating which are not present in the LFM simulation. In addition, the average amount of Joule heating seen in the coupled model during southward IMF is 142 gigawatts less than was seen in the uncoupled model. Both of these features are indicative of ionospheric and thermospheric processes which are not included in simple ionosphere used by the LFM.

44 Figure 18: The northern hemisphere Joule heating in the northern hemisphere is shown for the entire interval. The LFM results are shown with the thin black line while the CMIT results are shown with the thick grey line. Magnetosphere/Inner Magnetosphere (LFM/RCM) Coupling As part of the space environment modeling activities of CISM, Toffoletto, Sazykin, and Lyon are working to integrate the computational machinery of the Rice Convection Model (RCM), an inner magnetosphere model that computes potential electric fields, electric currents, and magnetospheric particle distributions in careful detail, including the effects of magnetosphereionosphere coupling and the transport of magnetospheric particles by gradient/curvature drift, in the Lyon-Fedder-Mobarry (LFM) global magnetohydrodynamics (MHD) model of the magnetosphere. While single fluid MHD is generally held to be the simplest complete first-order representation of the large-scale structure and dynamics of Earth s magnetosphere and its interaction with the solar wind, it fails to provide an adequate representation of conditions in the inner magnetosphere where energy-dependent gradient and curvature drifts become important. Additionally, since it is essentially a pair of 2-D rather than 3-D calculations, the RCM is able to use a much finer grid to represent the inner magnetosphere than the LFM. Figure 19 illustrates this point: it shows an equatorial view of a portion of the LFM grid overlaid with the RCM ionospheric grid (in blue) mapped to the equatorial plane along magnetic field lines. The RCM s grid can be seen to be significantly finer that the LFM s grid in the inner magnetosphere.

45 In the fully coupled CISM code, the LFM will provide the RCM with magnetic fields that are consistent with the momentum equation (as well as providing boundary-condition information). The RCM will provide the MHD code with corrected inner magnetosphere pressures and densities, which will be used to modify the field and plasma values computed by LFM within the RCM modeling region. The information exchange will be done frequently during an event simulation, so that the coupled code will, in principle, provide a realistic representation of the entire magnetosphere, including the processes that couple the inner and outer regions. We have undertaken the first step in the coupling process: a one-way coupling in which the MHD code feeds information to the RCM; however the RCM does not yet feed its computed densities and pressures back to MHD. In coupling the two codes it is necessary for the LFM to provide pressure and density to both initialize the RCM and to provide subsequent time-dependent boundary conditions for the RCM calculations. As the system evolves the RCM uses LFM plasma information about particles that are entering the RCM modeling region. While initially the plasma distribution is assumed to be Maxwellian, the RCM formalism permits it to depart from Maxwellian as it drifts through the inner magnetosphere, although it is constrained to remain isotropic. The conversion from RCM to a single-fluid MHD description is given by summing over the RCM plasma species. (Details are given in Toffoletto et al, 2004). An important aspect of the information exchange between the LFM and the RCM involves tracing magnetic field lines, both for the purpose of computing flux tube volume V and for transferring RCM computed pressures and densities onto the LFM grid. In the latter case, it is assumed that pressure and density are constant along a magnetic field line, so that each 3-D grid point that is topologically connected to a point on the 2-D RCM modeling grid has its pressures and density replaced by RCM computed values. For convenience, the exchange of data is done through a simple intermediate rectilinear grid similar to the one used by Toffoletto et al. [2001. Since field line tracing, which requires the determination of values at arbitrary locations throughout the magnetosphere, is very efficient in such a rectilinear grid, this approach substantially speeds up the field line tracing process. In addition, this tracing algorithm can be easily parallelized if needed. The disadvantage is that the data have to be interpolated from the non-orthogonal curvilinear LFM grid, which can inflict significant computationally and communication overhead. Initial tests suggest that the advantages of simplicity and speed outweigh the disadvantages. Reasonable accuracy can be achieved in the magnetic field interpolation by only interpolating values with the dipole field analytically subtracted. For ionospheric quantities, the RCM will use LFM-computed conductances. The electric potential distribution used as a high-latitude boundary condition for the RCM will be supplied by LFM values interpolated to the RCM boundary. Initially, to expedite development, the exchange of information between the two codes will be through data files. For our initial one-way coupling runs the full LFM has been replaced by an LFM proxy code (plfm) that simply copies pre-computed LFM results for use by the RCM. When the RCM has completed its calculations for a given time interval and is ready to exchange data, it uses a lock-file to pass a signal to the placeholder-lfm (plfm). This lock-file signals the plfm to copy in data and make it available for the RCM while the RCM idles, waiting for the new data to be made available. When the copy sequence is completed, the plfm signals the RCM, via a lock-file, that the data is available and then waits until the RCM is ready for the next sequence. For the first attempt at two-way coupling, the plfm will be replaced by the full LFM with appropriate lock-file code inserted. While it is anticipated that later versions of the coupled code will use more sophisticated means of information exchange the lock-file approach has proven to be adequate for our initial testing. The currently used data exchange protocol is designed to make the transition to the INTERCOM protocol relatively straightforward.

46 Figure 19: An equatorial view of an inner magnetospheric portion of the LFM grid overlaid with the RCM ionospheric grid (in blue) that has been mapped to the equatorial plane along magnetic field lines. The RCM s grid is significantly finer that the LFM s grid in that region First Generation (Object Oriented) Framework The long-term goals of CISM require a software framework in which codes can be coupled together efficiently and with a maximum amount of flexibility for adding new physics and new simulation models. This general software framework will be a major legacy of the CISM project. During the second year of CISM, we have added new capabilities to our coupling software, tested it in relatively simple cases to confirm their accuracy and efficiency, and begun to test use with our core codes. This work is also supported by a NASA grant from the Living with a Star (LWS) program, which has goals similar to those of CISM. Code Development Sussman and co-workers have developed recently the first version of the InterComm library, which replaces the Meta-Chaos library. InterComm has significant new functionality and provides much better performance. InterComm is a runtime library that achieves direct data transfers between data structures managed by multiple data parallel languages and libraries in different programs, which do not need to know in advance (i.e. before a data transfer is desired) any information about the program on the other side of the data transfer. All required information for the transfer is computed by InterComm at runtime. The current implementation supports data decomposition by regular and generalized block distributions, and completely irregular distributions. The first release of the library supports C, C++, Fortran77 and Fortran90 standard interfaces. It includes a higher level interface for P++/Overture codes that utilizes features of that package to hide some of the complexity of using the standard InterComm interface such as the LFM code. This capability effectively provides high level support for all the CISM space science codes. Support for additional data distributions can be added easily if required by new codes to be coupled to the existing set of codes. A higher level interface for parallel Fortran90

47 codes is still under design and development. The current version of InterComm is available with documentation at the project web site, ResearchAreas/ic. Sussman and co-workers have also added support for generalized block data distributions to the Multiblock Parti library, which provides services for distributing data and using data across multiple processes in the P++/Overture framework. As P++/Overture is used in several of the space science codes in the project, we are working with the P++ development team at Lawrence Livermore National Laboratory. Adding generalized block distribution support for P++ via Multiblock Parti enables better load balancing within a P++/Overture parallel program, via the use of grid partitioning algorithms that generate generalized block distributions with good load balancing properties. We have delivered the new version of the library to the Overture group, after performing comprehensive testing of it standalone as well as with the currently available P++/Overture distribution. The Livermore group will perform additional testing in preparation for incorporating the new library into an upcoming release of P++/Overture. Basic tests of InterComm and Overture Functionality This year we have tested the performance of InterComm and Overture with simple versions of o space weather codes; this work is described in Goodrich et al. (2004). Shay used InterComm to pass data between two identical copies of the 3D two fluid code developed at the University of Maryland (Shay et al., 2001). He initialized a circularly polarized whistler wave in one version of the code such as to propagation into the second code domain. He found the results using the InterComm and the two separate codes were identical to those obtained in a single version of the code with a larger grid encompassing the same spatial domain as the coupled pair of separate versions combined. Cassak studied the ability of the Overture package to handle code couplings involving with overlapping grids, such as the LFM-RCM. He used a trapezoidal leapfrog time stepping implementation of the convection-diffusion equation in a two dimensional, annular domain overlapped with a square region for the interior of this annular region. He found Overture extended the solution smoothly and stably across the overlapping grid system. He further found the stability criterion is the one dependent on the cell spacing of the component grids, not the cell spacing of the overlapping grids. As individual grid nodes of the overlapping grid cells can be quite close, this is an important result for our work. For the first use of InterComm with a core CISM, Lyon is attempting to transform the P++ version of the LFM code by separating the ionosphere/thermosphere (ITM) model from the magnetohydrodynamics (MHD) model, and adding InterComm calls to allow these two modules to interoperate. This effort will allow other ITM models in CISM to replace the current model in the LFM code, once they are modified to use InterComm. Incorporating additional codes with the LFM using InterComm, including the Rice Convection Model, has been discussed and will be done in the next few months. Source Code Repository and Data Archive We have established this year the central source code repository and data archive for CISM at BU, both are resident our new Linux server, re.bu.edu. This server is a dual Pentium Linux

48 machine with currently 2.4 Terabytes of disk storage available to store simulation codes, simulation results, and spacecraft and ground based data. Our code repository is implemented using the Concurrent Version System (CVS). Originally written as a tool for open-source software project management and development, CVS is now used throughout the software development world for: Tracking changes to software files and packages; Enabling simultaneous development work upon projects and subprojects while resolving any conflicting changes that arise; Facilitating communication among developers by requiring them to enter logs of changes made, and providing a method for announcements of such changes; Limiting repository read / write access to those involved with specific projects; Allowing a 'rollback' capability in which developers can revert to previous versions of a project in case changes in software result in incompatibilities; Associating 'release tags' with a project at and midpoint or end stage of development. We are currently using CVS for physics-based and empirical model development, and also for development of the CISM_DX visualization tool suite. Each 'official release' of a CISM model or project, either for internal CISM use (for, e.g., model validation) or for use by a wider audience, will be associated with a distinct CVS 'tag', which will enable developers to continue improving software even as the static release is tested or used. We have implemented the data archive using the Storage Resource Broker (SRB, see package developed and supported by the San Diego Supercomputer Center. SRB enables us to maintain an integrated logical view of binary data which can be distributed across all the CISM institutions, having various computational platforms, operating systems and storage devices. Our server at BU, re.bu.edu) is the first and central CISM SRB node, and maintains master CISM (SRB) database. We note that this database stores the locations of our data as well as the metadata associated with the data. It essentially maintains documented pointers to the actual data files, which typically are distributed across the CISM sites. Presently, our server holds locally spacecraft, ground-based and solar-terrestrial magnetic activity index data, results from CISM model runs, as well as distributable versions of the CISM physical and empirical models and visualization tools (i.e., CISM DX). In addition to maintaining a database of model results and observational data, SRB provides a virtual space in which the intermediate and end-result of model validation, skill-score validation using CISM metrics, data and simulation result post-processing, and documentation writing may be shared among people in the CISM community. In Year 3 we plan to extend installation of SRB at data servers at several other CISM institutions. By running the SRB clients that interface with the CISM SRB database at BU, we can have 'local' access from any site to all the data stored on all the CISM data servers. Eventually, the CISM SRB database will interface directly with the CISM DX model visualization software, feeding it data and simulation results based upon user inquiry.

49 2E. Model Validation and Metrics Thrust Activities The core of the V/M team is centered at Boston University and led by Harlan Spence. Other BU members include: Andrew Clarke, Nancy Crooker, Charles Goodrich, W. Jeffrey Hughes, John Lyon, Mathew Owens, and George Siscoe. These eight scientists meet approximately every three weeks via the Access Grid with 12 other V/M team members from CISM partner institutions, including: UTEP (Ramon Lopez), NCAR (Alan Burns, Giuliana detoma, Stan Solomon, Wenbin Wang, Michael Wiltberger), AFRL (Nick Arge), UC Boulder (Robert Weigel, Michael Gehmeyer), Alabama A&M (Marius Schamschula), SWRI (Geoffrey Crowley), and SAIC (Pete Riley). These twenty scientists plus their associates and students are well integrated with the other CISM thrusts so optimal cross-communication is assured. Significant Accomplishments One of the most significant accomplishments which met last year s goals was the establishment of CISM metrics. Metrics are defined to be clearly specified and standardized quantifications of how well empirical or numerical models predict those physical measurements especially important to space weather. Tracking these metrics over time will provide a quantitative record of numerical models improving abilities to predict key quantities. Metrics inform both model developers and model users of a model s predictive capabilities. During this past year, the rationale for CISM metrics selection were developed and a list of 29 metrics, along with the baseline models, first-generation physics models, and the data sets needed to compute skill scores, were established. Several important factors influenced our selection of CISM space weather metrics. These include: 1. The metrics must be a reasonably small collection to be able to be feasibly tracked; 2. On the other hand, they must be comprehensive enough to assess robustly the wide range of CISM models; 3. They must be based on direct measurements or derived quantities that will be continuously and reliably available into the foreseeable future; 4. They must be quantities related to key space weather effects that we are trying to predict; and finally, 5. They must be parameters that are recognized to be important by the space physics science community and/or the operational user community. This final factor naturally divides CISM metrics into two broad categories: operational and scientific. Operational metrics will gauge the progress of our ability to predict high-priority space weather effects that are of major concern to user communities. Scientific metrics are complementary to these, chosen to assess key quantities that are important scientifically and are a critical test of a model s ultimate ability to predict, but which are not necessarily of immediate and/or direct relevance to user communities.

50 Operational Metrics Science Metrics Operational SW Community Baseline Models Skill Score Data Sets Physics Models 1 Shocks and CMEs at L1 ACE MAS+ENLIL a Speed Augmented Vrsnack- Gopalswamy (a) " b Arrival time " " c Bz " " d Duration " " 2 SEP Properties GOES UCB a Event/No Event PROTONS (b) " b Rise Time " " c Peak Flux " " d Duration " " e Cutoff Shea-Smart (c) POES 3 Magnetic Indices a Dst Temerin-Li (d) NGDC LFM+RCM b Ap/K ARX-McPherron (e) " 4 Regional Ground db/dt Weigel-Baker (f) IMAGE (mag) LFM+TING 5 Radiation Belt EP fluxes a GEO Li (g) LANL RBM b MEO and LEO Vassiliadis (h) SAMPEX " 6 Ionosphere/Neutral Atmosphere a "State" of ionosphere IRI (i) Digisondes TING Scientific SW Community 1 Solar/Coronal SOHO UV a Coronal Hole Index PFSS/Wang-Sheeley (j) maps MAS+ENLIL SOHO b White-light Streamer Belt Index PFSS/Yi-Ming Wang (k) LASCO " 2 Solar Wind/IMF at L1 a Density WSA (l) + nv = constant ACE MAS+ENLIL b Velocity WSA " " c IMF vector WSA + B " " 3 GEO/MEO Environment a Magnetic field Tsyganenko (m) GOES LFM+RCM b Particle fluxes (ring current/rad belt) MSM (n), CRRESELE (o) GOES/LANL LFM+RCM,RBM c M'pause crossing Shue (p) LFM+RCM 4 MI Coupling a Polar Cap Potential Weimer (q) DMSP LFM+TING b Polar Cap Boundary Weimer " c Field Aligned Currents (2D) Weimer LFM+TING+MIC d Particle precipitation AURORA (r) MIC 5 Ionospheric Plasma a E-, F-region Heights IRI Digisondes + TING b E-, F-region Peak Densities " ISRs " Table 1. Operational (top section) and science (bottom section) metrics defined for CISM are shown in the left-hand column. Baseline models, data sets, and first-generation physics models from which skill scores will be computed are shown in subsequent columns. (a) Vrsnak and Gopalswamy (2002), Gopalswamy et al. (2001), Bothmer and Rust (1997), and Owen and Cargill (2002); (b) Balch (1999); (c) Shea and Smart (1990); (d) Temerin and Li (2002); (e) McPherron (2004); (f) Weigel and Baker (2003); (g) Li et al. (2003); (h) Vassiliadis et al. (2004); (i) Bilitza (2001, 2003); (j) Wang and Sheeley (1992); (k) Wang et al. (1997); (l) Wang and Sheeley (1992);

51 Arge and Pizzo (2000); Arge et al. (2004); (m) Tsyganenko (1995; 2003); (n) Magnetospheric Specification Model from AF- GEOSpace; (o) CRRES radiation belt electron model from AF-GEOSpace; (p) Shue et al. (1997, 1998); (q) Weimer (1996), Weimer (2001a;b); (r) Air Force Statistical Auroral Models from AF-GEOSpace Guided by the five factors outlined above, CISM identified 11 metrics areas to track (six for the operational and five for the scientific metric categories) that address key phenomenon and/or regions that span from the Sun to the Earth. Operational metrics areas cover: interplanetary shocks and coronal mass ejection (CME) properties; solar energetic particle (SEP) properties; standardized magnetic indices; regional ground magnetic field variations; radiation belt energetic particle fluxes; and properties of the ionosphere/neutral atmosphere. Scientific metrics areas cover: solar and coronal hole properties; solar wind and interplanetary magnetic field (IMF) properties at 1 AU; particle and field properties of the geostationary (GEO) and medium Earth orbit (MEO) regions; properties key to magnetosphere-ionosphere coupling; and finally, ionospheric plasma properties. These metrics are shown in Table 1 (taken from Spence et al., 2004). Note that every metric consists of four elements: a parameter, a baseline model (herein typically empirical) used to predict that parameter, an observation of that parameter for skill score computation, and finally, a predictive physics-based model. The results of this work are documented in a paper recently accepted for publication (Spence et al., 2004). As noted above, the process of defining and refining the next level of detail needed to compute metrics is a goal for next year. Such details include specifying the relevant properties (e.g., the time cadence, the spatial location(s), the altitude extent(s), etc.) of each particular metric as well as the optimal means by which departures of model outputs from observations can be measured. First CISM Validation Studies Validation of CISM models is as important to the success of CISM as is their development. A model is only useful to the extent we know how well it simulates reality. Consequently, validation and metrics define a distinct and vital research thrust within CISM carried about by a common team. Science validation occurs during the major development period of each code version. The distributed CISM validation team is working with code developers to: (1) identify key aspects of codes or code coupling which require scientific validation; (2) compile and use extant data sets to exercise and explore the ranges of validity of the codes; and (3) iteratively feedback the knowledge gained into the ongoing development of scientifically robust models. The goal of this effort is to assure that CISM scientists understand the primary science outputs of the models early enough in the process, through comparisons with the most relevant observations, so that developers will be able to adjust their codes to better model the desired regions or processes. While the V/M teams await the delivery of all the coupled CISM models, several validation efforts of the stand-alone LFM model were completed by graduate students at BU and UTEP. Here, we briefly summarize several examples of these science validation efforts, representative of the many others that are ongoing at CISM institutions (please refer to Guild et al., 2004; Lopez et al., 2004; Spence et al., 2004; and Vassiliadis et al., 2004 for more details). The examples chosen here demonstrate both traditional model validation techniques (case study analysis) as well as techniques not generally applied to the comparison of observations with large-scale simulation results (statistical climatology ). They explore and quantify the strengths and weaknesses of the core geospace model, the Lyon-Fedder-Mobbary (LFM) code, a global three-dimensional MHD simulation of Earth s magnetosphere (Lyon et al., 2004). While these are specific studies aimed at a specific model, the approaches described apply generically to virtually all CISM models and are representative of other ongoing and planned studies. We note that all of these V/M efforts use the CISM-adopted CISM-DX software architecture.

52 Validation studies at Boston University The goal of one study is to quantify the degree to which the LFM code reproduces the distortions of the inner magnetosphere during the life-cycle of magnetic storms (Huang et al., 2003). For this study, several magnetic storm intervals were selected that spanned from minor to moderate storm conditions (as measured by the Dst index). Huang et al. s case study analyses reveal and quantify the regions where, and the times when, the uncoupled LFM code performs well in addition to when it performs poorly. These analyses reveal that: 1. MHD field lines are not stretched enough at geosynchronous orbit during storm main phase, especially on the night side (M s in Figure 1 indicate midnight location of GOES- 8); 2. MHD pressure in the inner magnetosphere is low compared to expected pressures during storm main phases; and 3. Ongoing studies show that coupling the Rice Convection Model (RCM) with the LFM simulation leads to significantly increased pressure gradients and increased field line stretching. The goal of a second study (Guild et al., 2004) is to validate the LFM code with statistical data surveys through comparative climatology of bulk magnetotail plasma sheet properties. They compare more than three years of Geotail plasmasheet data (constituting nearly 800,000 data points) with 10 consecutive days of LFM simulation results (constituting over 15 million data points in the plasma sheet) driven by real solar wind and IMF inputs. Comparison of Geotail and LFM thermal pressure, magnetic pressure, and perpendicular flows in the equatorial plane reveal a number of important similarities and differences. For instance, the LFM thermal pressure falls off much more steeply than the data, the Geotail magnetic pressure shows a dawn-dusk asymmetry, and the LFM median perpendicular flow magnitudes are significantly higher than those in the Geotail data. Based on these comparisons of average properties, Guild et al. (2004) reach important conclusions about model strengths and weaknesses that would not be discernable in case studies, including comparing quantitatively the degree of flow variability as well as the median flow values. Validation studies at UTEP and Alabama A&M University The groups at UTEP and Alabama A&M University have also been working on comparisons of LFM output and observations, using these validation efforts especially as a means to involve undergraduate students in research and to build up a space science capability at AAMU. Since there are currently no faculty at AAMU with a space physics research background, the AAMU M.S. student (the first in the new Space Science concentration) has been remotely mentored by the UTEP group. In the spring of 2004 it is expected that a new faculty hire will be able to provide local support to students. The UTEP group has been working on several comparisons between observations and the results from the LFM simulations, as discussed below We have examined two magnetic storm runs (January 10, 1997 and September 25, 1998) to compare the calculated joule heating rates and distributions to estimates of joule heating from AIME and AE. These studies have shown quantitatively that the simulation does a reasonably good job of simulating the overall pattern of joule heating. Of particular interest is the fact that during the main phase of the storm there was a strong correlation between the solar wind ram pressure (which was driven by density fluctuations) and the ionospheric heating. The same effect has been documented during other events. Further study, using LFM as an experimental

53 tool to investigate the effect of solar wind variations on ionospheric joule heating and the crosspolar potential, demonstrated the effect as well as its cause [Lopez et al., 2004]. The results of this validation effort have considerable bearing on models that will be used to describe solar wind-magnetosphere coupling during magnetic cloud events. Another area of research is the comparison of the spatial distribution of joule heating in LFM and AMIE. While some features are similar, such as the concentration of heating on the dawn side, there are features in the LFM result, such as the north-south structure, about which we have doubts. We intend in the future to do similar comparisons using the LFM-TING model and quantify the differences in the magnitude and distribution of the joule heating. A final area of research is collaboration between UTEP and AAMU. UTEP undergraduates have been collecting a data set of magnetotail observations during the main phase of magnetic storms, and preparing a statistical representation of that data set. A similar data set is being compiled from the LFM magnetic storm runs by the M.S. student at AAMU. He will then compare the two results to quantify how well the LFM reproduces the tail field during magnetic storm main phase. For example, to what extent are observed lobe magnetic field values in the tail reproduced by LFM? The answer to this question has a direct bearing on how well LFM handles the rates of dayside and nightside reconnection. A similar study is being conducted of the structure of the mid-tail neutral sheet. UTEP students are assembling the data for comparison to LFM results. These studies are the mid-tail analogue of the comparison of the LFM geosynchronous magnetic field to observations described above.

54 2F. Empirical and Forecast Modeling Thrust The empirical and forecast modeling thrust is being led by Dan Baker, who also heads the Knowledge Transfer group. The empirical modeling group consists of approximately 16 individuals who participate regularly in weekly or bi-weekly teleconferences or on-site meetings. At LASP, the participants include Xinlin Li, Bob Weigel, Scot Elkington, and Josh Rigler. From BU the participants are George Siscoe, Jeff Hughes, Chuck Goodrich, Harlan Spence, and Nancy Crooker. Alex Klimas and Dimitris Vassiliadis participate from NASA/Goddard along with Bob McPherron of UCLA. Terry Onsager and Howard Singer from NOAA/SEC and Nick Arge from AFRL are also regular participants and provide significant guidance and feedback. Summary of Empirical Modeling Objectives 1. Create end-to-end empirical models relating ground-based and space-based observations of the Sun and solar wind to important geophysical parameters at Earth. 2. Compare output of these models to current forecasts and demonstrate that these models are more skillful than current techniques and are more human-independent. 3. Provide a test bed where various physics-based modules can be evaluated. Use statistical methods to characterize and identify essential aspects that need to be included in physics-based models. 4. Provide models that can be used to compute baseline metrics for the physics-based models. 5. Use empirical techniques to assist physics-based model groups in extracting useful information from numerical output. Significant Accomplishments: The most significant accomplishment in this reporting period was the completion of a sun-earth empirical model chain and its pre-validation delivery to the Space Environment Center. At the same time of delivery to SEC this model was delivered to the Validation team. After validation, the final version will be delivered to the SEC, and the models in this version will be frozen. The model chain is integrated into the CISM Data Explorer (CISM_DX), which is a software framework developed by the Knowledge Transfer group to facilitate communication of research results within CISM and then to the outside user community. The models were installed on a computer in the Forecast Center at SEC in April, The CISM forecast model currently predicts three parameters with lead times of 1-7 days: the solar wind velocity at L1, the daily-average Ap index, and MeV electron flux in the radiation belt for L= The overriding goal is to provide a highly reliable output through well documented and easily accessible models; we also wanted to be able to answer the question of how these realtime models are different from what exists today. Our models are different for several reasons: (1) they are in a documented package that is available for any CISM member or SEC employee to download, explore, test improvements, and do additional validation, and (2) for a given predicted quantity, the output of several models are used to make a final prediction. This ensemble approach has the potential to make an optimal prediction based on all available inputs and it adds a layer of reliability, because the individual models do not all use the same input measurements; if some of the input measurements are not available, a prediction can still be made. The ensemble approach will become even more useful when large-scale numerical results begin to contribute predictions.

55 Figure 20: The CISM Forecast Model Vsw prediction. This top level plot only shows minimal information in the same sense as a daily weather forecast in the newspaper. The predictions shown were made from a combination of model outputs including a persistence model, autoregressive model, and the WSA model. Note the addition of a probability estimate; although this feature is not complete, there is a placeholder for such estimates. Because the source code was made available through CISM_DX, any CISM scientist can explore methods for its calculation by just having a copy of CISM_DX. An additional advantage of incorporating these models into CISM_DX is that a scientist or validator can easily take a more in-depth view of how the above predictions were made, and can view and compare the output of each individual model. In the case of the solar wind velocity prediction at L1, the prediction shown if Figure 20 is actually based on several models. One is the Wang-Sheely-Arge (WSA) solar wind model, another is an autoregressive model that is based on past measurements of the solar wind velocity at L1, and finally there is a persistence model. The combination of model predictions to make a final prediction addresses the ever-present reality that all monitors of the space and geospace environment are not always available. In our case, when no solar source surface measurements are available, the WSA model cannot make a prediction, and when no L1 data are available, an autoregressive model cannot make a prediction. By using all available prediction models, the CISM forecast model will have fewer prediction gaps, and can provide a prediction at L1 even when measurements at L1 are not available. A possible extension of this idea is to invert a geomagnetic quantity to give an estimate of the solar wind velocity at L1 in the case that all other models and data are not available, as happened during the October 2004 storms. With this inversion, we can still have a real-time estimate of the solar wind velocity at L1. This is a project under consideration.

56 Figure 21: The CISM Forecast Model Ap prediction. The predictions shown were made from a combination of individual model predictions, including a persistence model, and autoregressive model, and ARX model with ACE Vsw input, and an ARX model with WSA Vsw input. Note the addition of a probability estimate; although this feature is not complete, there is a placeholder for such estimates. In the case of the prediction of Ap, the prediction shown in Figure 21 is based on four models. The first model is persistence. The second model is an autoregressive filter that requires measurements of Ap over the last 30 days to make a prediction of Ap 1-7 days in advance. The third model is an autoregressive exogenous filter (ARX), which requires measurements of Ap over the last 30 days and delayed measurements of the solar wind velocity at L1 to make a prediction of Ap 1-7 days in advance. The fourth model is an ARX model in the same form as the third model, but instead of using delayed solar wind velocity measurements at L1, it uses the WSA predictions. These outputs of the contributing models shown in Figure 22 are also made available for viewing and evaluation.

57 Figure 22: Detailed view of model predictions that contributed to the ensemble prediction shown in Figure 2. The final quantity that is predicted is the MeV electron flux In the radiation belt for L= This model currently makes 1-day lead time forecasts, but is being extended to make 1-7 day forecasts. A validation package of this model was sent to Kevin Scro at Peterson Air Force Base in March, We are currently working with Scro to assess what the Air Force needs are and to find out how future design and implementation decisions in CISM Forecast Model can be made to address these needs. The validation package was delivered as an IDL package. The model has since been integrated into the Data Explorer so that it can be executed and evaluated alongside of the other forecast models. There are three graduate students that worked with the Empirical Modeling Group: Manny Presicci, Alexa Halford, and Josh Rigler. Manny Presicci is advised by Daniel Baker and works on empirical modeling of the radiation belt. In May, 2004 he will attend the Spring AGU to present his most recent research results. Alexa Halford is advised by Xinlin Li and works on

58 modeling of the ring current. She attended the GEM meeting in June, 2003 and the Yosemite Inner Magnetosphere meeting in February, These two students work closely with several scientists at LASP and give presentations at group meetings approximately every semester. Josh Rigler is advised by Dusan Baker and is near graduation; he has recently contributed tutorial codes for using the Kalman filter technique in MeV electron prediction and specification into CISM_DX. Changes in Research Direction In the past year there have been some changes with regard to the Empirical Modeling thrust area. This issue was discussed at a meeting at BU in January, 2004, which was attended by Jeff Hughes, George Siscoe, Harlan Spence, Nancy Crooker, Bob Weigel, Chuck Goodrich, and Dan Baker. In this meeting, the concept of a CISM Forecast Model (FM) was developed to address the question of what CISM will do with empirical modeling once the baseline model is in place. (In the original proposal the role of empirical modeling after the first two years was not specified explicitly.) It was concluded that any FM prediction made of a given quantity will most likely always contain pieces from the numerical models and pieces from the empirical models, in much the same way that several empirical models contribute to the Ap and Vsw prediction. That is, both empirical models and numerical models have strengths and weaknesses. In many cases empirical models have strengths where numerical models are weak, and vice-versa. Therefore, initially the CISM Forecast Model will contain only empirical models. As the numerical models mature, they may replace some empirical models in the CISM Forecast Model, but many of the empirical models will remain. In this sense, the CISM FM will be an integrated model. The development and improvement of EMs will not stop once the EM chain is complete, although the amount of time spent on improving existing empirical models will decrease and more EM effort will be made in determining how the output of the numerical codes can be transformed into a useful prediction, and for determining how to improve the CISM FM with anything available. For this reason the EM thrust is now more accurately termed the FM thrust. Goals and Activities Planned for the Coming Year Our goals and plans for the next reporting period were laid out at the All-Hands meeting at BU in September, They include: Year 2 Milestones (September 2004, uncompleted milestones in italic) Use WSA to provide 1-3-day forecasts of electron flux levels o Deliver FIR models for validation to KT and VM team o Develop ARMA models o Continue development of EKF model o Develop LPF of relativistic electrons at GEO Use WSA to provide 1-3 day forecasts useful to SEC o Deliver daily AR Ap algorithm to KT and VM team McPherron o Deliver Ap ARMA algorithm to KT and VM team McPherron o Deliver Ap air mass model to KT and VM team (expected completion in August) o Create surrogate time series as input to physics-based model (expected completion in July) o Deliver empirical B and db/dt model to KT and VM team (in progress)

59 Provide L1-Earth test of B and db/dt w/coupled Weimer, LFM-TING model o Provide initial report of 24 hr run using LFM (in progress) o Explore how to create forecast-relevant output from LFM output (in progress) Year 3 Milestones (September 2005) o Use WSA to predict surface B and db/dt using solar inputs and LFM-TING o Provide initial report of 24 hr run using LFM Gehmeyer/Weigel/Arge o 24 hr run using LFM/TING Gehmeyr/Weigel/Arge o Continue to work with SEC forecasters on developing modules for transforming output of CISM models into forecast-relevant measures. Develop methods and modules for transforming MHD/TING output to forecast-relevant quantities.

60 2G. Data Assimilation Thrust The first annual CISM report discussed plans and likely areas where incorporation of measurements into the solar-terrestrial modeling suite using data assimilation techniques could be of advantage. During the second year, several meetings were conducted, collaborations with other groups were identified, and initial steps toward data assimilation implementation were taken in selected fields. Several CISM participants contribute to the data assimilation thrust, including CISM Director Jeffrey Hughes and Harlan Spence at BU, Stan Solomon at NCAR, Dan Baker, Bob Weigel, Josh Rigler, and Giuliana detoma at CU/LASP, Mary Hudson at Dartmouth, and Nick Arge at AFRL. Additional collaborations with Bob Schunk and Jan Sojka at Utah State and Tim Fuller- Rowell at CU/CIRES have also been established. Meetings A workshop on data assimilation in the space physics context was held at the University of Colorado Laboratory for Atmospheric and Space Physics, following Space Weather Week, Presentations were made by Jeff Anderson, Tim Fuller-Rowell, Jan Sojka, and Bob Schunk. Approximately 30 people attended, the majority affiliated with CISM. A summary of this workshop was presented at a special session at the 2003 GEM meeting. Additional presentations from the CISM team were made by Dan Baker, Josh Rigler, and Bob Weigel. This forum was used to discuss space physics data assimilation issues with the broader community. A special data assimilation session was also held during the CISM all-hands meeting. The key conclusions reached at this meeting included: (1) formal data assimilation into coupled numerical models is still a distant goal; (2) the best places to consider data assimilation are at the interfaces where codes are coupled; (3) to improve solar wind predictions upstream of the magnetosphere, some form of data assimilation of recent in-situ measurements at L1 will be required. Solar Photosphere The synoptic map assembly routine for solar magnetograms was generalized, and can now remap calibrated solar disk images into heliographic coordinates and assemble them into synoptic maps using virtually any type of solar data. At present, synoptic maps have been created using data from the SOHO/EIT instrument, He 1083nm and H-alpha data from Mauna Loa Solar Observatory, and photospheric field data from Mount Wilson, Wilcox, National or Kitt Peak, SOHO/MDI observatories. White light synoptic maps using data from Mauna Loa and SOHO/LASCO will also soon be generated. The MAS model uses solar magnetogram data as boundary specification, and significant progress was made toward the goal of modeling several specific space weather events using magnetograms for model initiation, particularly the May 12, 1997 and late 1996 events. Upstream Solar Wind Measurements Measurements of the interplanetary magnetic field vector, solar wind vector, and plasma density, are routinely performed by spacecraft in orbit about the first libration point (L1) of the Sun-Earth gravitational system, particularly by the ACE spacecraft. These measurements have been employed for some time to drive magnetosphere and ionosphere models, including direct

61 incorporation into the LFM model to study magnetospheric disturbances, and continuous realtime ingestion by the TING model, using empirical relationships to estimate auroral precipitation and convection. One outcome of the meetings and discussions described above has been that this interface between the corona/heliosphere models and the magnetosphere/ionosphere models is a key location to assimilate data. In particular, measured real-time data can constrain the heliosphere model by providing improved timing information for expected geoeffective solar wind events. Bob Weigel is currently working with Nick Arge on methodologies for using data assimilation methods to improve predictions of the solar wind velocity using in-situ measurements at L1. Radiation Belts Rigler et al. [2004] implemented an adaptive system identification scheme, based on the Kalman filter with process noise, to determine optimal time-dependent electron response functions. The nonlinear dynamic response of the radiation belts can then be tracked in time by recursively updating the optimal linear filter coefficients as new observations become available. This demonstrates a significant improvement in zero-time-lag electron log-flux predictions relative to models that are based on time-stationary linear prediction filters, while incurring only a slight increase in computational complexity. Modifications necessary for an operational specification and forecast model, including the assimilation of real-time data, more sophisticated model structures, and a more practical gridded description of the radiation belt state, are being examined. Figure 23. Static FIR and adaptive Kalman filter (KF) predictions of 2 6 MeV electron log flux for 1994 through 1999 are compared with observations. The percent of variance (PV) is plotted as a function of L-shell. Historical averages were used to substitute for model output during periods without reliable solar wind input (recognizable as horizontal contour lines in the predicted output), and were not included in the calculation of PV. Assimilative Mapping of Ionospheric Electrodynamics The Assimilative Mapping of Ionospheric Electrodynamics (AMIE) procedure, developed by CISM co-investigator A.D. Richmond and co-workers, is a data assimilation model that specifies high-latitude electric potential and convection, and ionospheric conductance, using a variety of measurement sources. These include ground-based magnetometer data, DMSP and NOAA auroral particle data, and auroral images from the POLAR UVI and IMAGE FUV instruments,

62 incoherent scatter radars, and Superdarn HF radars. AMIE combines the available data with constraints from an empirical auroral oval to obtain a time-dependent best-fit estimate of auroral ionization and convectional forcing, suitable for use as the high-latitude inputs for TIE-GCM runs. AMIE will be an important tool for validation of the coupled magnetosphere-ionospherethermosphere model, an effort that is currently just underway. AMIE studies of the large geomagnetic events of April 2002 and October 2003 have recently been performed in support of community-wide analysis campaigns. Ionospheric Data Assimilation The most promising area for direct incorporation of measurements into CISM model segments is in the ionosphere-thermosphere system, where the meteorological analogy is most valid. Global ionospheric density measurements using ground-based and, increasingly, space-based GPS receivers are available, and models such as the Global Assimilation of Ionospheric Measurements (GAIM) are of considerable utility for describing the current state and short-term future of the ionosphere, particularly the non-auroral ionosphere. A collaboration between the CISM ionosphere/thermosphere thrust and the GAIM group has been agreed upon, involving in particular Bob Schunk and Jan Sojka at Utah State, and Tim Fuller-Rowell at CU/CIRES. We will conduct an experiment using a GAIM ionospheric specification as the initial condition for a coupled magnetosphere-ionosphere model run, and quantify the differences between runs using nominal and assimilated initiation, and the time scales on which ionospheric features persist in the model during magnetically active periods. Solar Ultraviolet Radiation Currently the coupled magnetosphere-ionosphere model employs empirical models of solar farultraviolet, extreme-ultraviolet, and X-ray irradiance based on past and current measurements. A demonstration using daily measurements (by the TIMED solar EUV experiment and the SORCE satellite) of the solar irradiance in the NCAR TIE-GCM was conducted for the extremely active period of October-November 2003 and the results presented at the fall 2003 AGU. For short-term (~2-day) forecasts of solar irradiance necessary for modeling of space weather events, a collaboration with Space Environment Technologies (Kent Tobiska and Dave Bouwer) has been established to use the proxy-based forecasts they provide to the NOAA Space Environment Center for prognostic modeling.

63 References Abbett, W. P., G. H. Fisher, & Y. Fan, The Effects of Rotation on the Evolution of Rising Omega Loops in a Stratified Model Convection Zone, Astrophys. J., 546, Abbett, W. P., Z. Mikic, J. A. Linker, J. M. McTiernan, T. Magara, and G. H. Fisher, The Photospheric Boundary of Sun-to-Earth Coupled Models, J. Atmos. Solar Terr. Phys., submitted, Arge, C.N., J.G. Luhmann, D. Odstrcil, C.J. Schrijver, and Y. Li, Stream Structure and Coronal Sources of the Solar Wind During the May 12th, 1997 CME, J. Atmos. Solar Terr. Phys., submitted, Fan, Y., E. G. Zweibel, M. G., Linton, and G. H. Fisher,.The Rise of Kink-unstable Magnetic Flux Tubes and the Origin of delta-configuration Sunspots, Astrophys. J,521, 460, Goodrich, C.C., A.L. Sussman, L.G. Lyon, M.A. Shay, and P.A. Cassak, J. Atmos. Solar Terr. Phys., submitted, Guild, T., H. E. Spence, L. Kepko, M. Wiltberger, C. Goodrich, J. Lyon, W. Jeffrey Hughes, Plasma Sheet Climatology: Geotail Observations and LFM Model Comparisons, J. Atmos. Solar Terr. Phys., submitted, Huang, C.-L., H. E. Spence, L. Kepko, J. Lyon, C. Goodrich, T. Guild, M. Wiltberger, W. J. Hughes, H. Singer, Comparison of LFM simulation with Tsyganenko models and GOES observations, EOS Trans. AGU, 84(46), Fall Meeting Suppl., Abstract SM52A-0569, 2003.Jacques, S. A., Astrophys. J., 215, 942, 1977 Jacques, S. A., Momentum and energy transport by waves in the solar atmosphere and solar wind, Astrophys. J., 215, 942, Li, Yan, J.G. Luhmann, G.H. Fisher, and B. Welsch, Observational Evidence for Velocity Convergence toward Magnetic Neutral Lines as a Factor in CME Initiation, J. Atmos. Solar Terr. Phys., submitted, Linker, J.A, Z. Mikic, R. Lionello,, P. Riley, T. Amari., and D. Odstrcil, Physics of Plasmas, 10, 1971, Liu, Y., Photospheric Magnetic Field Observations during the May 12, 1997 CME and their implications for modeling that event, J. Atmos. Solar Terr. Phys., submitted, Lopez, R. E., M. Wiltberger, S. Hernandez, and J. G. Lyon (2004), Solar wind density control of energy transfer to the magnetosphere, Geophys. Res. Lett., 31, L08804, doi: / 2003GL Luhmann, J.G., S.C. Solomon, J.A. Linker, J.G. Lyon, Z. Mikic, D. Odstrcil, W. Wang, and M. Wiltberger, Coupled Model Simulation of a Sun-to-Earth Space Weather Event, J. Atmos. Solar Terr. Phys., submitted, Rigler, E.J, D.N. Baker, R.S. Weigel, D. Vassiliadis, and A. J. Klimas, Adaptive linear prediction of radiation belt electrons using the Kalman filter, Space Weather, 2, S03003, 2004 Riley P., J. Linker, and Z. Mikic, An Empirically-Driven Global MHD Model of the Solar Corona and Inner Heliosphere, J. Geophys. Res., 106, 15,889, Riley P., J.A. Linker, R. Lionello, Z. Mikic, D. Odstrcil, M.A. Hidalgo, C. Cid, Q. Hu, R.P. Lepping, B.J. Lynch, and A. Rees, Fitting Flux Ropes to a Global MHD Solution: A Comparison of Techniques, J. Atmos. Solar Terr. Phys., submitted, Shay, M.A., J.F. Drake, B.N. Rogers, and R.E. Denton, Alfvenic collisionless magnetic reconnection and the Hall term, J. Geophys. Res., 106, 3759, Spence, H. E., D. Baker, A. Burns, T. Guild, C.-L. Huang, G. Siscoe, and R. Weigel, CISM metrics plan and initial model validation results, J. Atmos. Sol. Terr. Phys., submitted, 2004.

64 Toffoletto, F. R., J. Birn, M Hesse, R. W. Spiro and R. A. Wolf, Modeling Inner Magnetospheric Electrodynamics, in Space Weather, Geophys Monograph Series, Am. Geophys. Union, 125, edited by P. Song, H. Singer, G. Siscoe, pp. 265, Toffoletto, F.R., S, Sazykin, R.W. Spiro, R.A. Wolf, and J.G. Lyon, RCM meets LFM: Initial results of one way coupling, J. Atmos. Solar Terr. Phys., submitted, Usmanov, A. V., M. L.. Goldstein, E.P. Besser, & J.M. Fritzer, A global MHD solar wind model with WKB Alfvén waves: Comparison with Ulysses data, J. Geophys. Res., 105, 12675, Vassiliadis, D., R. S. Weigel, S. G. Kanakel, D. N. Baker, and A. J. Klimas, Probing the solar wind-inner magnetosphere coupling: validation of the MREF model for CISM/KT, J. Atmos. Solar Terr. Phys., submitted, Wiltberger, M., W. Wang, A.G. Burns, S.C. Solomon, J.G. Lyon, and C.C. Goodrich, Initial results from the Coupled Magnetosphere-Ionosphere-Thermosphere Model: Magnetospheric and Ionospheric responses, J. Atmos. Solar Terr. Phys., submitted, Wang, W., M. Wiltberger, A.G. Burns, S.C. Solomon, T.L. Killeen, N. Maruyama, and J.G. Lyon, Initial results from Coupled Magnetosphere-Ionosphere-Thermosphere: Model: Thermosphere-Ionosphere responses, J. Atmos. Solar Terr. Phys., submitted, 2004.

65 III. Education Overall objectives: The major objective of the CISM education program is to develop the next generation of space physicists while increasing the diversity of our field. Our goal is to instill in them a holistic view of the Solar Terrestrial environment that is unusual in our fragmented field. The core elements we have created to meet this objective are: 1. The space weather summer school 2. A graduate retreat for CISM graduate students 3. A distributed program of undergraduate research 4. A program of space science education research that supports our core educational goals, and also contributes to the research base on teaching and learning. 5. Supporting the emergence of a space science education and research program at Alabama A&M University (which is also a key diversity strategy) We concentrate on these five elements in response to the recommendation of our Advisory Council that we focus our education efforts and do a few things very well. In addition, CISM has dedicated some resources for professional development for high school and community college teachers through long-term internships and the development of space weather classroom activities, and a program for communicating the importance of space weather to the general public via the web and planetarium shows. These additional programs, while they are more loosely connected to our major educational objective, promote broader science literacy and in the long term increase the number and diversity of students entering science through use of CISM resources and space weather. The performance indicators for these elements are summarized in the CISM 5 year Milestones and Goals. Significant Accomplishments: Held another successful summer school and saw the adoption of summer school materials in regular classrooms, including one non-cism institution Held first graduate student retreat Involved more than 15 undergraduates in research, with many presenting papers at meetings, and resulting in two students electing to pursue graduate studies in space weather at CISM schools Approval of a Space Science M.S. degree at Alabama A&M University Conducted research on the use of 3-D visualization for teaching about magnetic fields and the solar system Developed a test version of the Space Weather Monitor Involved 6 teachers in summer internships and research Began work on an undergraduate section of the popular Windows to the Universe site Plans for next year Hold summer school with significantly updated labs and prepare summer school materials for broad dissemination to non-cism university faculty Hold graduate retreat

66 Continue successful program of undergraduate research, recruiting new students and graduating additional students who pursue space weather studies Expand the number of students in the AAMU M.S. program from one to three and help recruit a new tenure-track faculty member in space physics Conduct first round of summative assessments of the impact of the summer school and undergraduate research on the participants. Continue research into visualization and use those results to design improved labs for the summer school Test the space weather monitor and begin to disseminate it through CISM teacher interns Launch the new Windows to the Universe CISM section and release the first CISM planetarium shows Staff Changes and move of the UTEP group to Florida Tech: Over the past year we have had two changes among the support staff of the education programs. Dr. Esther Zirbel is leaving Boston University. The UTEP education coordinator, Jana Martinez, has taken another position as Assistant Director of the Model Institution of Excellence program at UTEP. Robert Bruntz, who is finishing his M.S. degree working on a CISM research project, was hired as the new Education Coordinator at UTEP. He will provide visualization support for UTEP and AAMU, assist in mentoring undergraduates doing research, conduct longterm evaluation of CISM education programs, and provide support for the summer school. Ramon Lopez announced recently that he, Neiscja Turner, and Robert Bruntz will be moving to Florida Institute of Technology during summer This has major implications for the management of the education program which are discussed in the management section. Distribution of CISM Education Activities and Responsible Individuals: Boston University Lead Institution for organizing Summer School (Hughes, Nottingham) Undergraduate research (Hughes, Spence) Graduate retreat (Hughes, Nottingham) Ethics training (Hughes) UTEP (FL Tech, as of 8/04) Summer school organization and execution (Lopez, Bruntz) Physics Education Research/Curriculum Development (Lopez, Turner, Bruntz,) Undergraduate research (Lopez, Turner, Bruntz) Formative and Summative Program Evaluation (Bruntz) Teacher intern/professional development (Lopez, Turner, Bruntz) Stanford Space Weather Monitor development (Scherrer, Morefield) Teacher intern/professional development (Scherrer) Rice University Teacher intern/professional development (Reiff) Undergraduate research (Reiff, Toffoletto) Full-dome planetarium show development (Reiff)

67 NCAR Summer school organization and execution (Wiltberger) Undergraduate research (Solomon) U. Colorado Undergraduate research (Baker) Dartmouth Teacher intern/professional development (Hudson) Undergraduate research (Hudson, Lotko) Programs The Summer School: Description: The summer school is a two-week course in space weather and space weather models. It is intended primarily for first-year graduate students; however it can taught at a level and in a manner suitable for well-prepared upper division undergraduates, and also space weather professionals. The summer school has been held for three years, beginning before the official start of the center. The next one will be held this summer, from July 26 through August 6 at Boston University. The faculty and students of the 2003 school are listed in Appendix F. The basic structure of the summer school is three series of lectures on the space environment (reality), its effects on technological systems and humans (harsh reality), and on models used to specify or predict the space environment (virtual reality) during the morning, and afternoon computer labs in which students explore these models, and also gain familiarity with IDL and OpenDX (two core CISM software packages). In addition leading professionals in space weather are invited to present after-lunch talks on their professional lives to give students a better grasp of the various sides to the profession of space weather beyond research. The pedagogy of the summer school is active and hands-on. Lectures use a Peer-Instruction format (Mazur, 1997). At the end of the morning lectures students write questions on index cards that are collected; answers to the questions are provided at the start of the labs. On the last day of the summer school the students engage in an event study that challenges them to integrate disparate data sets and simulations into a concept map that describes the event. This summer we will prepare a version of the final activity for broader dissemination. During the summer school there is explicit reflection on the pedagogy being used, as well as a special session on physics education. Summer school participants might at some point be teachers themselves, so a goal of the summer school is to inculcate some familiarity with active learning techniques in the participants. Moreover, it is also a goal that CISM faculty will also gain familiarity with these techniques and use them in their regular classes. We are also disseminating information about the summer school to the professional community. Papers describing the summer school have been presented at national meetings of the AAPT and AGU. Evaluation: Formative: The summer school has daily formative assessments of lecture sessions and labs, using a modified Likert scale ranging from 0 to 5 (as opposed to a typical 1-5 scale in order to emphasize that the low end of the scale is a negative evaluation). Our indicator for success is that each individual item score better that 3. Previous evaluations have scored 3.6 to 4.8 on this scale, which we consider to be successful. There is also a summary evaluation of the summer

68 school at the end of the event. Those scores have averaged 4.2 and comments have also been quite positive, with most students ranking the summer school as very good or excellent. Summative: This summer we will survey past participants and the faculty in order to judge the long-term effects of the summer school. Our indicator for success is that students in retrospect rank the summer graduate school to be among their most significant graduate educational experiences (rated a 4 or 5 on a five point scale) 2 years after participation There will also be a selected set of interviews that will collect more in-depth information. The evaluation will examine how the summer school has impacted the participants space weather careers in order to document specific success. Another indicator of success is that the faculties use the summer school active engagement educational techniques in regular school-year teaching. Already there is evidence that the impact on summer school faculty has been considerable. Four of the five CISM summer school faculty who have regular teaching assignments have reported that they have adopted techniques they learned in the summer school into regular classes, notably Peer Instruction and/or the final activity. In fact, it was this latter development that led us to decide that we would prepare a user-friendly version for broad dissemination. Integration with Research: The summer school provides a critical support to the research program through the creation of a shared resource that all CISM institutions can use to educate incoming center members about space weather and models. It also provides a forum for testing and evaluating teaching techniques in an advanced educational setting. New visualizations are also tested in the summer school, such as the new CISM DX module for displaying the ENLIL results. The new lab created using this visualization tool provided an excellent introduction to the 3-D structure of the solar wind in the heliosphere. Graduate Student Retreat Each year we will hold a meeting for all CISM graduate students. The retreat is intended for students who are or will be engaged in CISM research and whom we expect to be within CISM for a few years, completing their thesis or dissertation on a CISM topic. The first of these graduate student retreats was held in Essex, MA, the weekend before the All-Hands meeting in September 2003 engaged in CISM research. Seven students participated in the meeting, which was led by J. Hughes. Two recent Ph.D.s, Drs. Murr and Turner (who is part of the CISM education team), also participated in the retreat. Sessions included The Goals of CISM, How to survive graduate school, and Getting along with your advisor. These meetings will allow the CISM graduate students to share their research and build community. A rotating program will focus on professional development items not normally taught in a formal graduate curriculum, such as the funding and management of research, and how to prepare research proposals, or teaching methods and physics education. In addition to the graduate retreat, there are Access Grid node opportunities for CISM-wide education, such as the session held on March 30 for learning about CISM DX, our standard visualization package. Evaluation: Formative evaluation will focus on student perception of usefulness. Summative evaluation will consist of interviews in succeeding years. Success will be defined if interviewed students identify the meetings as very useful or extremely useful on a five-point scale. Integration with Research: The annual retreats will provide important professional development to CISM graduate students as they prepare to become professional scientists and faculty. Our students will be better prepared to make informed career choices and will have a community

69 and support network that will enhance their ability to be successful in the future. As such, the CISM mentoring and professional development program for graduate students might prove attractive to students and aid in recruitment. Physics Education Research/Curriculum Development CISM will have an active program of physics education research that will enhance the teaching of physics, astronomy, and space weather within CISM and without. We continue our research program focusing how students interpret space science visualizations. For example, we have conducted a study to determine why students are able to understand 3-D renderings of the substorm current wedge better than 2-D drawings. The evidence suggests that the cognitive processing of mental images is a key factor in producing the cognitive load that makes it difficult for students to come to proper conclusions. This result will be used to guide the revision of summer school labs. We consider such research to be essential if we are to use CISM visualization products effectively in education. This research also has applicability to undergraduate teaching, such as introductory astronomy and physics. Evaluation: Our target for science education research and curriculum material development is four papers presented at AGU, AAPT, and/or APS meetings on CISM-developed course materials or science education research projects and two papers per year submitted for publication. We are meeting this target. We also have a five year goal that CISM-developed instructional materials be used by least 5 non-cism institutions. We are making progress toward that goal with the use of summer school material at the Air Force Academy. As we create a more portable version of the space weather event activity, we expect to see other institutions using the materials. Integration with Research: The physics education component of CISM will focus on research problems that are directly related to CISM science. Our results will be used to increase the effective use of CISM resources like visualizations in the research realm, such as helping students acquire the basic content and conceptual knowledge they need to contribute to the center. Undergraduate research CISM has an active, distributed undergraduate research program with about 18 students. Several of these students have decided to continue graduate study at CISM institutions, including two Hispanic and one African-American student (the undergraduate research program is an important part of our diversity strategy as well). Evaluation: Students will be surveyed to determine how they feel about their research experience. Those surveys will begin this summer now that many students have been involved for more than one year of research. Our metric for success is that students rate their experience as a 4 or 5 on a 5-point scale. We also examine student demographics, with success defined as at least 30% of participants being women or members of underrepresented minorities. We are meeting that target. Over the five years our measure of success in using undergraduate research as a recruiting tool is to have 1-3 women or minority students who participate in such research recruited into each CISM graduate school. We have met this goal at Dartmouth and Rice, and continue to stress the recruitment of women and minority undergraduate students at all CISM schools.

70 Integration with Research: Undergraduates can provide significant research support to CISM while at the same time such research can be used to recruit and retain students in physics. New graduate program at Alabama A&M University Alabama A&M University has received approval to establish a masters-level graduate concentration in space science within their physics graduate program beginning in Fall This marks a major success for CISM. The Center is acting as a key partner and resource to help AAMU make this new concentration successful and build an active research presence in space science with space weather as a research focus that will be integrated with the new degree concentration. Establishing these new programs will require a stronger space physics expertise within the AAMU faculty since there are currently no faculty with significant space science research experience. The Center has reallocated funds, increasing the AAMU share, to provide partial support for a new tenure-track position at AAMU in space physics that will be advertised this fall. Until then, CISM is providing research support through visits to AAMU by CISM staff and faculty, and Access Grid meetings. This support is crucial for those students currently in, or joining, the AAMU space science program. Building a stronger program at AAMU will provide that institution with increased ability to contribute to CISM research as a whole. A new tenure-track position and a new MS concentration represent significant institutional legacies that will continue after the center and provide a new route for increasing the number of African Americans in space physics. Evaluation: Success is indicated by the establishment of a new degree program (accomplished) and by hiring a tenure-track space physics faculty member who can support the program. We expect this to happen next year. All Hands meeting and Ethics training In September 2003, the first CISM All Hands meeting was held in Boston. At that meeting, a plenary session on ethics was led by Dr. Aine Donovan (Executive Director, Ethics Institute, Dartmouth College). This session included a discussion of general scientific ethics, as well as discussion of rules of the road specific to CISM, such as rules for authorship on abstracts. We intend to have similar sessions, either at the annual All-Hands meeting, or via the Access Grid for new CISM members. Evaluation: We only this year instituted this program and do not yet have a metric of success beyond the fact that such a session was held and that a lively discussion ensued. Teacher Interns/Professional development Rice, NCAR, and Stanford are planning to host teacher interns this summer. The role of these teacher interns is to work with the CISM sites and connect CISM resources to the area education community and/or engage in summer research experiences as a means of professional development. We envisage that these teachers will become long-term members of the center, and this is in fact happening. Rice: Rice will hire a teacher intern to assist in summer continuing education and to continue incorporate space weather material into a course offered in the fall for Houston teachers as part of the MS in teaching offered by Rice. The intern will also assist in the development of

71 evaluation instruments for the planetarium shows being developed by Rice. A teacher has also been working on a research program with Dr. Reiff using an empirical model of the polar cap potential. NCAR: NCAR will hire one, perhaps two, teacher interns this summer to assist in work on an undergraduate website on space weather. One of those teachers, Randy Russell, is a long-term intern, having worked with the NCAR group since the inception of CISM. Stanford: Stanford will hire two interns, one high school physics teacher and one community college physics teacher, to assist in the development of an inexpensive space weather monitor. The basic idea is to monitor ionospheric disturbances with existing VLF signals. The detectors cost about $200 each. Proof of concept tests have been conducted, and this year classroom lessons have been developed. Cal State University Hayward students have been writing supporting software that scheduled for completion in May This summer the interns will test the materials, software, and activities. In the following year we will begin disseminating the detectors and lessons. In a research project at UTEP we conducted in-depth interviews with all the physics majors to determine why they had chosen physics. About 50% said that they decided to study physics because of positive experiences in middle or high school. Thus by providing real-world data experience we hope to impact high school students. Students at schools where the space weather monitor is deployed will be surveyed to measure the effect on their attitudes about science. Evaluation: We had hoped to survey the teachers who participated in the program last year, but due to staff changes this was not done. Our metric for success is that participating teachers rate their experience as a 4 or 5 on a 5-point scale and that they use CISM-related materials in their classes. We target that at least 1/3 of teachers participating in CISM professional development activities teach historically underserved populations. This has been met. We can begin to evaluate another metric (Did 1/2 of teachers continue contacts with CISM for at least 2 years after the summer research experience?) starting next year. Informal Science Education The group at NCAR has begun developing web-based materials for undergraduates about space weather models, leveraging the expertise in informal science in the Windows to the Universe group. These materials will be used both inside CISM to provide our undergraduates with basic space physics content and as a public outreach tool for undergraduates or advanced high school seniors. The NCAR and UTEP teams will jointly develop evaluation instruments to determine the effectiveness of the materials. We also will use these materials as a research tool to study student interpretation of images and visualizations. What we learn will be utilized to improve the summer school, thus directly supporting the research effort by better preparing center students. Two other informal science efforts are centered on the creation of materials for planetarium shows. At one end of the spectrum, Stanford is creating a space weather show for the Starlab planetarium (Starlab is a small, inflatable planetarium used widely in schools). At the large-scale end of the spectrum, Rice University is developing content for an immersive, full-dome planetarium show in conjunction with the Houston Museum of Natural Science. Success in each case will be measured by number of viewers of the show. More specific measures of success will be determined this summer.

72 Education Metrics What constitutes success in Education and Outreach? Metrics for Evaluation. Graduate school indicators Number of students. The summer graduate school attracts a full house each year Student evaluations (formative). Attendants rank the summer graduate school as being a very good or excellent educational experience (4 point scale) Student evaluations (summative). Students in retrospect rank the summer graduate school to be among their most significant graduate educational experiences (Average > 3 on a five point scale; to be asked 2 years after participation). Faculty evaluation (formative) - Participating faculty learn about innovative pedagogical techniques or CISM resources that they intend to incorporate into courses they teach in their home institutions (statement on interview) Faculty evaluation (summative) - Participating faculty actually use these techniques and/or resources within 2 years of their participation in the graduate summer school (statement on interview) Graduate retreat indicators Number of students. The graduate retreat attracts a full house each year, with all CISM Ph.D. students attending at least once. Student evaluations. Students rank the graduate retreat as being a very good or excellent professional development experience (4 point scale). Undergraduate education indicators Student participation in research (formative) Students feel that this research experience was exciting and valuable (Average > 3 on a five point scale) Faculty evaluation (formative) - Participating faculty learn about innovative pedagogical techniques or CISM resources that they intend to incorporate into undergraduate courses they teach in their home institutions. Faculty evaluation (summative) - They actually use these techniques and/or resources within 2 years of the development of the materials. Faculty any non-cism institutions use CISM products in undergraduate courses. Science Education Research Indicators Four papers presented at AGU, AAPT, and/or APS meetings on CISM-developed course materials or science education research projects Two published papers describing development of materials and evaluation in the Journal of College Science Teaching, or similar publications Formal science education in secondary school Teacher Workshop participation (formative) - Did teachers rank the workshops as being a very good or better educational experience (4 point scale)? Teacher Workshop evaluation (formative) - Did participating teachers learn about innovative pedagogical techniques or CISM resources that they intend to incorporate into courses they teach?

73 Teacher research interns (summative) - Did 1/2 of teachers continue contacts with CISM for at least 2 years after the summer research experience? Teacher research interns (summative) - Did all teachers include some CISM product in their school (educational resource, visit by scientist, etc.) in the year following the research experience? Informal science education and outreach Produce 2 CISM immersive experience planetarium shows Six museums have shown CISM immersive planetarium shows CISM immersive planetarium shows have been copied to HDTV format and made available to NASA, television networks, etc. CISM videos have been made available to news organizations, like CNN, and have been used in situations where appropriate. Program development at AAMU (also in Diversity) AAMU has developed new graduate degree program in space science AAMU has hired new tenure-track faculty member in space physics with CISM support

74 IV. Knowledge Transfer In addition to the normal dissemination of research results within the scientific community (achieved through publishing papers in journals and reporting results at meetings) the CISM Knowledge Transfer plan has three distinct objectives: 1. Transition of forecasting tools to NOAA/Space Environment Center. The transition will be facilitated through a close partnership and interaction with SEC including a CISM staff member to specifically be the liaison with SEC. We anticipate that the first models to be transitioned will be components from the end-to-end empirical model. Later we will develop modules that transform output from the comprehensive numerical model into quantities of interest for space weather. 2. Providing the scientific research community access to CISM models. Access to the empirical models will be provided through a web site based at LASP. In the longer term, we plan to provide access to the CISM physics-based models through our partnerships with CCMC and NCSA. 3. Training and interaction with industrial partners and government labs and agencies. This will be achieved both through industrial or government employees attending the summer school and through specific programs that will be developed in conjunction with industrial partners The management and performance indicator for (1) is the transition of community-developed end-to-end models from the sun to earth to operational forecasts at NOAA/SEC. The initial performance indicator for (2) is the development of a KT web page that disseminates the forecasting advances of CISM to the scientific community. The indicators for (3) are a strong CISM presence at NOAA s Space Weather Week (in April), an industrial sponsorship of Space Weather Fellowships for graduate students and post-docs, a program of a seminar series whereby CISM member visit industrial partners to present on-site seminars and other training, and participation of government and/or industrial employees at the 2004 Summer School. The following is a list of knowledge transfer activities that serve to meet the Knowledge Transfer goals of (I) transition of forecasting tools to NOAA/SEC; (II) dissemination of community models to the scientific community; (III) training and interaction with industrial partners and government labs and agencies. 1. (I, III) Interaction at NOAA/SEC, Boulder, CO. Interaction involves meetings and presentations on a bi-weekly basis. 2. (I, III) Hire of full time KT Liason Michael Gehmeyr at NOAA/SEC. Weigel still works out of an office at the SEC at least one day per week. 3. (I, III) Interaction with Peterson Air Force Base, Colorado Springs, CO. Weigel and graduate student Manny Presicci visited the Base in December, They met with Kevin Scro and several members of his computer support staff. 4. (II) Development of the CISM Data Explorer (see 2b below). 5. (II,III) Baker, Weigel, and Gehmeyr will all give presentations at Space Weather Week in Boulder, CO in April, (IIII) Bob Weigel has been meeting with Eric Kihn at the NGDC in Boulder CO to better understand what resources they have that can be exploited by the Data Explorer. 7. (I,II) Integration of the CISM Forecast Model into the CISM Data Explorer (see below) 8. (III) Dan Baker is chairing the Boulder Space Matrix, a consortium of all major organizations (commercial, governmental, and educational) that are doing space-related

75 work. 9. (III) Dan Baker has been meeting with Dr. Dan Moorer of Ball Aerospace (Military Space Systems) 10. (II) Dan Baker is chair of the Living with a Star Management and Operations Working Group (MOWG). 11. (III) Dan Baker is a member of the Colorado Lt. Governor s working on Aerospace. 12. (III) Dan Baker is co-chair of the Air Force Technical Applications Center Satellite Review Panel. (In addition to these activities many of the senior members of the CISM team serve on various advisory or steering committees advising various government agencies and programs and/or serve in leadership roles in professional societies and in NSF programs such as SHINE, GEM and CEDAR.) In the past year, the primary effort in Knowledge Transfer has been in completing the preparation and CISM- developed empirical models for transition to the Space Environment Center. During the preparation of empirical forecast models to be delivered to the SEC, we realized that in the future a framework was needed to simplify model delivery, validation, and comparison. In response to this, the KT group initiated intense development of the framework for a CISM Data Explorer (CISM_DX). The overarching goal of the framework was to increase model input/output standardization and to improve communication first between CISM modelers Figure 24: Startup window of the CISM Data Explorer.