NEXT GENERATION ALTIMETER SERVICE CHALLENGES AND ACHIEVEMENTS

Transcription

1 NEXT GENERATION ALTIMETER SERVICE CHALLENGES AND ACHIEVEMENTS M. Naeije (1), R. Scharroo (2), and E. Doornbos (1) (1) DEOS, Delft University of Technology, Kluyverweg 1, 2629 HS Delft, Netherlands (2) Altimetrics LLC, 330a Parsonage Road, Cornish NH 03745, USA ABSTRACT The Radar Altimeter Database System RADS was launched by DEOS in 2001 as a precursor to an operational observing system as envisaged by the upcoming International Altimetry Service IAS. It is a facility to easily manage and access calibrated and validated altimeter data with consistency as common denominator. Whenever new data, models or knowledge arrive, the system is updated. We have now over 20 years worth of valuable sea level, wave and wind data suited for science, operations and commercial activities. Data and background information can be obtained through the web portal This paper highlights some of the historic achievements of RADS and challenges to reach the next generation of altimeter service. The latest development foresees a complete revision to gain in efficiency and accessibility not only for the expert user but also for the new user. We go from RADS V2 to V3 and include the widely in the ocean and atmospheric community accepted and used NetCDF format, which allows for a much larger database including the incorporation of high frequency data (10, 18 or 20Hz data). At the same time allowing users to use standard tools for extraction and processing and to develop own tools. It resolves also some of the practical data organization shortcomings in RADS V2. 1. INTRODUCTION Sea level and its horizontal slope and time changes are the surface expression of ocean processes occurring over large spatial and temporal scales. Large-scale currents like the Gulf Stream, extending up to kilometres in depth, and Kelvin and Rossby waves, thousands of km in horizontal extent, reveal the air-sea exchanges that can affect subsequent weather at great distances. Sea level also provides evidence of local solar heating, tides, and water produced by melting ice caps or glaciers. Altimeters emit a sequence of short pulses at microwave frequency and then measure the return times to obtain the instrument-to-sea-surface distance. Knowledge of the exact position of the satellite provides a means to extract the sea level. The development of satellite altimeters for the active sensing of ocean surface topography has been very successful over the past 20 years. Today, altimetric measurement of the sea level has become an indispensable research tool. Skylab (1974) demonstrated the possibility of altimetry from space, Geos-3 ( ) made measurements sufficiently accurate for geoid studies, SEASAT (1978) showed the potential for ocean studies, and GEOSAT ( ) and ERS-1 ( ) produced routine ocean products though lacking the long-wavelength components, because they still suffered from relatively large orbit and atmosphere errors. The last decade a lot of effort, has been put in improving the needed altimeter corrections and the radial accuracy of the satellite height. This has been realized by improving the orbit determination, on one hand by improving the modelling of gravity and tides, on the other hand by improving non-conservative forces acting on the satellite, like atmospheric drag. It is now feasible to monitor the rise of the global sea level and its potential acceleration, to an accuracy of better than 1 mm/yr. More recent altimetry missions, ERS-2 (1996-) and especially TOPEX/Poseidon ( ), added precise, largescale measurements, allowing to retrieve both fast and interannual sea level variations. This is enabled by the high accuracy, the ability to map the global ocean with a temporal sampling of days to a few weeks, and the prospect of a long-term time series decades into the future. The harsh accuracy requirement necessitates additional improvements to contemporary and future altimeter data, to accurately link missions from different times (in absolute sense), and optimally combine measurements from operational missions that display systematic differences. Altimetry missions are also of great interest in geodesy and geophysics. Sea level is dominated by local gravity and acts as a sort of levelling rod perpendicularly oriented to the local mass attraction defining the socalled marine geoid. Geoids derived from multiple altimetry missions now have a space resolution dense enough to resolve submarine ridges and mountains. Satellite orbits, as well as positions of ground tracking stations, are now determined very accurately using DORIS, GPS and laser data collected from altimetry satellites as well as other satellites. Such observations have immediate impact on reference system quality and related parameters. Additionally, the oceanographic community working with altimetric data is expecting a lot from dedicated gravity missions, like CHAMP, GRACE and GOCE. In particular, these missions allow significant improvement of the knowledge of the mean ocean circulation by providing a reference geoid of Proc. Envisat Symposium 2007, Montreux, Switzerland April 2007 (ESA SP-636, July 2007)

2 much better quality than before. Mean sea level monitoring will also significantly benefit from a better knowledge of reference systems and tectonics movements. In 2002 Jason-1 took over T/P s mission and Jason-2 is scheduled for launch in Also, since 2002 ESA s ENVISAT functions as the successor for ERS-2, and plans are in place to secure the continuous monitoring of the sea surface by incorporating an altimeter onboard of the future ESA Sentinel series (beyond 2010). These missions not only contribute considerably to ocean and geodesy studies, they also ring in the era of operational sea level monitoring as part of the GMES programme, the European initiative Global Monitoring for Environment and Security. RADS is a system that strives to support the GMES services developments. It also is targeted to integrate with ESA s Basic Radar Altimeter Toolbox BRAT. Geosciences research and applications are getting more and more dependent on space-borne techniques: techniques that require a global approach. Some techniques have become successful with the establishment of internationally coordinated services which handle the generation of standard products, recommendations, standards, coordinated analysis, data collection, and product and software distribution. Gradually, for other space techniques, like altimetry, the need for coordinated global approaches and global observation systems is being recognized. This resulted in the foundation of the IAS-PG, the Intl. Altimeter Service Planning Group commissioned by GLOSS, IAG, and IAPSO. Here DEOS/RADS is discussion partner and will put on the role as IAS analysis or data centre. 2. ABOUT ALTIMETRY What actually is measured or analysed onboard of the altimeter satellite is the so-called radar return waveform. Typically, this returned waveform consists of a rising part where at a time (after the first signal is received) less than the pulse width, the scattering zone intersects the sea surface in a disc shape with growing radius (footprint), and at a time more than the pulse width, the scattering zone intersects the surface in a ring shape with growing outer radius and thinning ring width, and therefore unchanging surface. In reality the received power will peak and then fall off due to the antenna power pattern and increasing distance of the scattering small surfaces resulting in a typical waveform shape. By averaging over many pulses the noise is substantially reduced and the average waveform can be analysed and interpreted more easily: the in this manner combined echo from an ocean surface can be described analytically by the Brown model [1]. The position of the rising edge centre corresponds to the (average) distance of the sea surface to the radar dish. The slope and the amplitude of the echo are related, respectively, to wave height (SWH or H 1/3 ) and to wind speed (U 10 ). In presence of waves the signal wave front will initially interest only the wave tips (first returns are received earlier), then gradually extend to the entire surface, within a time proportional to the wave height. This results in a reduced slope of the echo leading edge for increased wave height. This is used to retrieve wave height information. The effect of the wind over a sea surface is to increase its roughness, in a way proportional to the speed of the wind itself. A perfectly planar surface behaves, with regard to electromagnetic waves, like a mirror (specular reflection): the incoming energy is almost totally scattered at the angle between the direction of incidence and the perpendicular to the plane. In the case of the altimetric nadir observation the greatest part of the energy is scattered back toward the transmitter. If the surface is rough, the "radiation backscattering lobe" of the backscattered signal is widened, and the energy is then distributed over a larger angle around the zenith, decreasing the amplitude of the peak of the reflection lobe. The surface reflectivity (sigma-0) decreases (due to the rough surface a larger portion of the incident radiation is reflected away from the satellite). The observable effect is then a change of amplitude: with increasing wind speed, the amplitude of the echo signal decreases. With proper calibration it is therefore also possible to extract wind speed information from altimetry. Focussing on sea level, we thus only need the timing of the centre of the leading edge. In principal altimetry is a simple procedure: measure the time the radar pulse needs for travelling twice the distance from spacecraft to sea surface and subtract the derived distance from the height of the satellite to arrive at the sea surface height or sea level. The practice is somewhat more complex: we have to account for instrument design, calibration and validation results, media (atmospheric) corrections, geophysical adjustments like tides, inverse barometer, and sea state), reference system, precise orbits (satellite height), sampling characteristics, and so on. This is where RADS comes in. RADS provides a continuous set of sea level measurements of high constant quality incorporating all of the above-mentioned issues and providing consistency between data from different satellites and different satellite mission phases. Additionally, for deriving ocean dynamic topography we need to subtract the contribution of gravitation and rotation (gravity) of the Earth, the marine geoid, which is also supplied in RADS in different flavours. For more details on altimetry and applications the reader is referred to [2].

3 Programs Creation Tools (used to generate data base) Utilities (Data manipulation) User-specific Tools (created using RADS library) Development Tools (Utilities under development) Interface RADS Library (data reading, selection, conversion, editing, arithmatic, manipulation) Settings Global Settings (regulated by namelists in $RADSROOT directory) User Settings (regulated by namelists in ~/.rads) Local Settings (regulated by namelists in working directory) Command line arguments (common to most utilities) Data base Geosat ERS-1 TOPEX Poseidon Phase A (GM) Phase B (ERM) Phase A... Phase G Phase A Phase B (interleaved) ERS-2 GFO Jason-1 Envisat Figure 1. The Radar Altimeter Database System RADS. 3. DATABASE, TOOLS AND SERVICES The methodology and feasibility for building the Radar Altimeter Database System is described in detail in the original RADS [3] and RADSxx [4] project final reports. The work was divided into four packages dealing with project management, database maintenance, calibration and validation (including precise orbit determination and gravity field modelling), and communications (userinterfacing, dissemination of altimeter products, valueadded products definition, and tailor-made monitoring and processing tools development). Data maintenance primarily deals with adding new data from different data sources taking into account different references, different instrument characteristics, and different corrections models, and so on. As explained previously, validation of these corrections is key issue. Also for enabling the study of sea level change, all missions are continuously (cross)-calibrated. Maintenance also includes preparations for upcoming altimeter missions like CRYOSAT-2 and Jason-2. The calibration and validation forms the actual core of RADS. It focuses on cross-calibration methods, for which DEOS has been contracted by ESA for ENVISAT and CRYOSAT-2. Cross-calibration deals with inter-comparing measurements of sea level at the same location (but at different epochs and/or under different sea state conditions) of various radar altimeters. The vast amount of these co-locations ensures averaging of effects that are beyond the control of global ocean and environmental models or measurements. It enables the simultaneous identification of relations between the range bias and sea state. When cross-calibration is performed on a long term (i.e. far beyond the commissioning phases) relative drifts of the altimeter ranges can be identified, and, if one of the concurrent altimeters is regularly calibrated in absolute sense, this will identify the drift in each of the altimeters. In addition, operational orbits for ERS-2 and ENVISAT enable near real time applications of sea level, a prerequisite for the development of operational oceanography. Together with NOAA, DEOS has setup a processing chain for near-real time processing of ERS-2, ENVISAT and GFO-1 data. Precise orbits are produced for overall improvement of the data and to ensure a consistent quality. Much attention is paid to satellite surface force modelling, Earth stations coordinates, Earth gravity field model improvement, and satellite orbit consistency. Another aspect of the database work is communication or involvement of partners and potential users, definition and study of (new) valueadded products, web-interface maintenance and negotiating with institutes interested to join RADS. Successful examples are initiatives of NOCS and ESA to incorporate RADS in their own developments. Fig. 1 gives an overview of the RADS system: more then just a collection of data. The connection to the outside world runs through the web portal

4 For the base level of the database system, source data have been obtained from Fast Delivery and (Interim) Geophysical Data Record (GDR) providers like AVISO and CERSAT and transformed to level 0 (RADS definition) products with the restriction that all original information is maintained and that the format and contents are as uniform as possible. This enables replacing or isolating specific model corrections, key research at DEOS and other institutes. The base level contains measurements, corrections and auxiliary data at 1 Hz sample frequency. The data are stored in separate pass (or track) files and indexed by time relative to equator crossing. A typical pass file will hold around 2000 data records, which measures about 200 Kb. The enormous amount of files is kept in a hierarchical directory structure. Table 1 gives an overview of the available data within RADS. The total nears 1 billion records, approximately taking up 100 Gb of disk space. For RADS V2 we chose a rather unconventional way of storing the data: instead of record by record the data is stored field by field. At the time it provided the most flexible solution to add or remove fields. A more detailed explanation can be found in [5]. In summary, RADS can be characterized by: geophysical corrections and reference frame common to all altimeters, common data and meta file formats, flexible file augmentation, preferences controlled by name lists, up-to-date with most current GDR data and all instrumental corrections applied or provided. Table 1. RADS data content status spring Altimeter Phase Time Cycles Records Gs A B e1 A B C D E F G Tx A B N Pn A e2 A g1 A j1 A n1 B Sum FROM RADS V2 TO V3 As introduced in the abstract of this paper we opt for a major upgrade in RADS, which would allow for more than just 1-Hz data, more flexibility in data format, easier exchange with other altimeter data bases and generic tools, and more flavours of corrections, models, etc. The major upgrade of our software is scheduled for late 2007 and would also incorporate the transition to Fortran 90 code. Why upgrade? The current software is too much tailored to 1-Hz data. We anticipate a growing need for quality high frequency data for coastal, and inland water applications. This definitely means a drastic increase of the data volume. Incorporating 10/20-Hz data also requires a more flexible format, and the larger data volume requires dynamic memory allocation, not supported by the current code based on Fortran 77. Fortran 90 will make it easier for users to code their own software around the RADS database, and Fortran 90 modules are a good way to transfer some of the peripheral information in the data. In addition, Fortran 90 makes array arithmetic very easy, so every full pass of data can easily be worked on. Although the current RADS format is far more flexible that those of GDRs used by the (raw) data distribution centres, there are some impractical aspects. For instance RADS format does not allow floating point numbers, extension of the format with new options is not always possible, and the meta and data files are separated. Also if the data format has to be changed all together, we want to adopt a format that is widely used in the oceanographic and meteorological community, and not restricted to the current RADS or other GDR-type formats. So, we like to introduce The Network Common Data Format NetCDF, which is a machine-independent, self-describing, binary data format standard for exchanging scientific data, developed by the University Corporation for Atmospheric Research UCAR. Some major advantages include: Widely used format C, C++, Fortran 77 and Fortran 90 code is readily available to read NetCDF data NetCDF files are inherently platform independent Various generic tools are available to read (and display) NetCDF files directly Meta information and data are in the same file Information can be extended easily without requiring any software updates A well-defined format allows easy excess by other data base systems, such as GRID. Geosat 20 th anniversary data will be distributed in NetCDF format and RADS V3 compatible Jason-2 data is likely to be generated in NetCDF format and RADS V3 can easily be tuned

5 And finally, the current implementation of RADS uses numbered data fields, and it is not easy to use two flavours of the same correction simultaneously. That's why RADS V3 will have named data fields, edit limits that can be specified for each data field, and more flexible "arithmetic", allowing users to set up the altimetric equation themselves. This too will be easier to implement in Fortran DATA LEVELS We distinguish 4 levels of RADS data products, ranging from the base Level 0, binary records available when the total system has been implemented, like at DEOS, NOAA, NOCS, DNSC, SEAMERGES universities (more interested parties can and will follow), via Level 1 products or ASCII data dumps which are available to everyone through the RADS web portal at DEOS, via Level 2, interpolated data upon request, like along-track collinear data, crossover data (generated at locations where satellite passes cross), and sea level anomalies interpolated to periodic (weekly) regularly spaced mesh values or grids, which provides a convenient way of representing altimeter observed sea level changes, to Level 3, which defines altimeter value added products like tide models, calibration models (sea state bias), models for the sea level annual cycle and trend, Hövmuller (time-longitude) diagrams, and current velocities, magnitude and direction. Fig. 2 shows an example of Level 2 data; analyzed sea level data in the equatorial Pacific obtained from ERS and ENVISAT altimeter data. From top to bottom we see the anomalies late December 1996, representing normal conditions with relatively low levels in the eastern part, the anomalies late December 1997 with El Niño in full effect; abnormal high water level in the eastern part up to 40cm, the level in late December 1998, when the return from El Niño overshoots and results in abnormal low levels in the eastern part (La Niña), and finally the recent situation in April 2007 which shows a retreat to normal after a mild El Niño event in the 2006 holiday season. As a reminder: during a serious El Niño the eastern part of the tropical pacific will go up a few decimetre in response to a collapsing thermocline affected by eastward up- and down-welling Kelvin waves. Away from the equator motions are dominated by Rossby waves. The reader should check the RADS web pages A typical example of a Level 3 product is the absolute current velocity in the Gulf Stream area in the vicinity of the East coast of North-America. gives detailed information how we combined the three operational altimeters ENVISAT, Jason-1 and GFO-1 to arrive at current velocities that follow from the gradient in the (absolute) dynamic topography imposing (quasi) geostrophy. Figure 2. Sea level anomaly maps for the tropical pacific; perfectly suited for studying ENSO development. 6. SOME ACHIEVEMENTS There is a very active community utilizing RADS for various scientific issues concerning water level. A good example of usage is analyzing the December 2004 tsunami near Sumatra that affected the coastlines of the entire Indian Ocean. RADS enabled a quick-look at the tsunami evolution (after the fact). RADS contributed to hurricane research as well. In Fig. 3 the location and intensity of hurricane Katrina is plotted at intervals of six hours. We notice two intensification events that can not be correlated with surface temperature (upper panel), but have a decent correlation with dynamic topography based on Jason-1, TOPEX, Envisat, and GFO data.

6 Figure 4. Global averaged mean sea level change based on 15 years of calibrated and validated RADS altimetry. Figure 3. Hurricane Katrina s path and intensification related to surface temperature and ocean heat content based on altimetry. The dynamic topography is directly related to the ocean heat content (of the mixed layer and not just the skin temperature). The Loop Current can be seen entering the Gulf south of Cuba and exiting south of Florida; the warm-core ring (WCR) is very prominent shed from the Loop Current in the center of the Gulf and contributes to the growth of Katrina into a category 5 storm. Finally, with RADS it is very easy to calculate mean sea level evolution and make frequent updates. This can be done locally, regionally and globally. Fig. 4 shows the latest sea level change plot based on RADS, incorporating Topex, Jason, GFO, ERS and Envisat. After aligning the agreement is striking boiling down to a common rise value of around 3.1 mm/yr over the last 15 years of altimetry. Note: the in this manner obtained sea level sums contributions such as water expansion (warming), mass change (melt), isostatic adjustment of the crust, tectonics, etc. Other data and models should enable separation of all these contributions. Think of Cryosat-2, Jason-2, Altika on Oceansat-3 and Sentinel-3. Key in this remains the cooperation with other altimeter expert institutes, and the linkage of RADS with other data infrastructure programmes and services. Think of BRAT, think of GMES. The contribution to IAS will ensure a long-term altimetric sea level infrastructure so desperately needed for offshore operations, marine engineering and Earth system research. 8. REFERENCES 1. Rees, W.G., Physical Principles of Remote Sensing, Cambridge University Press, ISBN Fu, L.-L. and Cazenave, A., Satellite Altimetry and Earth Science, Academic Press, ISBN Schrama, E., Scharroo, R. and Naeije, M., RADS: Towards a Generic Multi Satellite Altimeter Database System, SRON/BCRS pub., USP-2 report 00-11, ISBN , Naeije, M., Doornbos, E., Mathers, L., Scharroo, R., Schrama, E., and Visser, P., RADSxx: Exploitation and Extension, SRON/NIVR pub., NUSP-2 report 02-06, ISBN , Naeije, M., Schrama, E., Doornbos, E., and Scharroo, R., The Role of RADS in Building the 15- year Altimeter Record, Proc. Symp. 15 years of Progress in Radar Altimetry, March 2006, Venice, Italy (ESA SP-614, July 2006), ISBN OUTLOOK Summarizing, RADS enables both expert altimeter user and novice to apply consistent and validated and calibrated sea level, wave height and wind speed data to their research or (operational) applications, and intends to keep on doing that for many years by improving and extending with any newly available altimeter data.