Operational air pollution forecasting and management system over Bilbao

Transcription

1 Operational air pollution forecasting and management system over Bilbao R. San Jose, I. Salas, A. Martin, J.L. Perez, A.B. Carpintero, D. Camara* & R.M. Gonzalez^ *Environmental Software and Modelling Group, Computer Science School, Technical University of Madrid, Campus de Montegancedo, Boadilla del Monte Madrid. ^Department of Meteorology and Geophysics, Complutense University of Madrid, Ciudad Universitaria, Abstract The increased importance of modelling and forecasting the air quality over urban areas has contributed to the development of more sophisticated software tools to manage urban air quality. During recent years, the use of the world wide web has also increased extraordinarily and consequently the new software technologies are linked more and more to Internet Services. In this contribution we present the application of the well known model OP ANA - a non-hydrostatic meteorological mesoscale model - based on MEMO model - and the CHEMA module - for simulating the chemical reaction - over the Bilbao (Spain) domain. This application is part of the EQUAL project funded by EU (DGXIII - Telematics for the Environment). We also present the importance of the new Geographic Information Systems to enhance the quality of the visualisation of the air pollution information and also the improvement of the emission model (EMIMO) to carry out such work. The management system is present as a set of parallel processes to perform different emission reduction scenarios when different pollution episodes are present. The results for different pollution episodes are also presented.

2 68 Computer Techniques in Environmental Studies 1. Introduction Air Quality Models are becoming an important tool for Municipalities and Regional Governments to manage the air quality into their respective domains. The Air Quality Forecasting Systems are moving toward the use of sophisticated new communication technologies (WAP systems, Internet, etc.) to improve the quality of the forecasting and also to improve the information means to the general public. The Municipalities and regional governments are obligated to report to the citizen in their respective domains about air quality levels but also they are obligated to reduce these levels when they are over the limits established by the different European Directives for Air Quality. In this context Information Technology becomes an important are where the improvement of Air Quality Models can effectively be applied in order to improve the quality of the information which is reaching to the public. EQUAL project (Electronic Services for a better Quality of Life) is one of the funded projects into the Information Technology for Environment area into the 5^ Framework Programme. In this contribution we will focus on the application of the well known OPANA model over the Bilbao area into the EQUAL project. OPANA model was developed at the beginning of 1995 (San Jos6 et al/ ) and it is under continues improvement. OPANA model means Operational version of ANA model. ANA stands for Atmospheric mesoscale Numerical pollution model for regional and urban Areas. The mesoscale air quality model ANA had several different previous versions, so that the E3DUSM, San Jos6 et al/ and NUFOMO, San Jos6 et ala ANA model is composed on several different codes such as: the Chemical Model for Atmospheric processes (CHEMA), the DEPOsition model (DEPO), the REmote sensing MOdel (REMO), the EMIssion model for Madrid Area (EMIMA) and the Regional MESoscale Transport model (REMEST). All of these modules are in fact independent modules which can be applied for specific purposes. The CHEMA model is integrated in the REMEST model under the "on-line" mode which means that the chemistry is solved and updated (m the time when the advection and diffusion is simulated by the REMEST model on the actual time step. The EMIMA model has recently incorporated under the "online" mode which means that the temperature and solar radiation are taken from REMEST module every time step and the traffic emissions and biogenic emissions for isoprene and monoterpenes are calculated with that information. Previous versions used average historical information for temperature and solar radiation to calculate emissions before starting the ANA simulation. The non-hydrostatic mesoscale model is based on the MEMO model (Hassak and Moussiopoulos*) and MM5 model (Grell et al.*). The Navier-Stokes partial differential equations are solved numerically to obtain the three wind speed

3 components temperature and humidity at each three dimensional grid box. The CHEMA module is based on the SMVGEAR method (Jacobson and Turco*) and it includes the CBM-IV mechanism (simplified version), Gery et al/. The deposition module DEPO includes a resistance approach from Wesely\ The OPANA version was developed to fulfil the different operational requirements, so that, the model should be used to forecast ozone, NOx and SO2 surface concentrations and to compare successfully with the urban and regional air pollution network information in different European cities. The EMMA^ project focused on this direction. Afriendlygraphical user interface OP ANA- VIS was developed to facilitate the use of the ANA modelling system. Figure 2 shows an example of the X- windows (Tcl-Tk) OPANA- VIS. In this contribution we will show the application of OPANA over the Bilbao domain (Regional domain: 96 x 96 km, 4 km spatial resolution; Nested domain: 16 x 12 km; 1 km spatial resolution). Figure 1 shows both domains. 69 Figure 1.- Digital Regional and urban domain for Bilbao Area. The model requires also a landuse classification. In this contribution we are using the NOAA/AVHRR global landuse classification with 1 km spatial resolution. A combination of land use classification 2 and 3 is used here. Landuse classification 2 corresponds with the Global Ecosystem Legend with 94 landuse types and Landuse classification 3 corresponds with the IGBP Land Cover Legend with 17 landuse types. Both combined landuse classification are related to the seven landuse types which are used by the REMEST model. Figure 2 shows a map with the final result.

4 70 Computer Techniques in Environmental Studies. :\. :/:/.Y::x::Y::^ \x;!;% Figure 2.- Landuse classification from NOAA/AVHRR (USGS) types 2 and 3. Emission data are taken from the IER (Stuttgart, Germany) emision data base in this contribution. Another emission data bases can be used such our EMIMA model (the generalised version is called EMMO). The Air Quality Network of Bilbao is used to initialize the chemistry module for the OP ANA model by using the values at OhOO oh the day of starting the simulation (usually 24 hours before the actual day) and in case of the ozone, we use the average concentrations during the past 24 hours. Finally the initial vertical meteorological soundings are obtained from the AVN/MRF weather forecasts which are automatically downloaded by using a JAVA application which is integrated into the OP ANA platform. Madrid domain (80 x 100 km; 5 km grid resolution). The simulation period is August, 31, 1998 OhOO to September, 4, 1998, 24h. Figure 3 shows how the vertical meteorological soundings are linked into the initial meteorological information which should be provided to OPANA model. The vertical meteorological soundings are downloaded from the NOAA server by using a JAVA software tool (OPANA-JAVA) which is controlled by the OPANA-VIS tool. Figure 3 shows an example on how these vertical meteorological soundings are distributed along the 120 hour simulation time. This information is given as

5 Computer Techniques in Environmental Studies 71 input data for the OP ANA simulation and it is not changed during all the simulation period (12-18 CPU hours). Figure 3.- AVN/MRF vertical meteorological soundings which are used as input vertical meteorological soundings for the non-hydrostatic mesoscale module (REMEST) of the OP ANA application. Figure 4.- O3 forecasted concentrations Web based illustration.

6 72 Computer Techniques in Environmental Studies The BILBAO/OPANA application is build in Tcl/Tk as a X-windows interface and the WWW application which is used to provide the air quality forecasts by using the INTERNET. Figure 4 shows an example of how this information is provided to the user. The Internet based Bilbao/OPANA system allows the user to select the pollutant, the time and the map he/she would like to see overlapping the surface forecasted air concentrations. 2. Brief description of the AVN/MRF meteorological models Numerical/Computational Properties: Horizontal Representation: Spectral (spherical harmonic basis functions) with transformation to a Gaussian grid for calculation of nonlinear quantities and physics. Horizontal Resolution: Spectral triangular 126 (T126); Gaussian grid of 384x190, roughly equivalent to 1x1 degree latitude/longitude. Vertical Domain: Surface to about 0.27 hpa divided into 28 layers. For a surface pressure of 1000 hpa, the lowest atmospheric level is at a pressure of about 996 hpa. Vertical Representation: Sigma coordinate. Lorenz grid. Quadratic-conserving finite difference scheme by Arakawa and Mintz (1974)*. Vertical Resolution: 28 unequally-spaced sigma levels. For a surface pressure of 1000 hpa, 8 levels are below 800 hpa, and 7 levels are above 100 hpa. Computer/Operating System: Cray-C90 and Cray-Y/MP computer using 8-16 processors in a UNICOS environment. Computational Performance: About 15 minutes Cray-C90 computation time per one-day forecast at T126. Initialization: Initialization is not necessary because the statistical spectral interpolation analysis scheme eliminates the unbalanced initial state. Time Integration Scheme(s): The main time integration is leapfrog for nonlinear advection terms, and semi-implicit for gravity waves and for zonal advection of vorticity and moisture. An Asselin (1972)** time filter is used to reduce computational modes. The dynamics and physics are split. For physical processes, implicit integration with a special timefilter (Kalnay and Kanamitsu, 1988)*^ is used for vertical diffusion. In order to incorporate physical tendencies into the semi-implicit integration scheme, a special adjustment scheme is performed (Kanamitsu et al., 1991/\ Smoothing/Filling: Mean orographic heights on the Gaussian grid are used. The Medium Range Forecast (MRF) model was started to be developed in 1982 by Sela J. \ The model is based on the usual expressions of conservation of mass, momentum, energy and moisture. In order to take advantage of the spectral technique in the horizontal, the momentum equations are replaced by the vorticity and divergence equations, thus eliminating the difficulties associated with the spectral representation of vector quantities on a sphere. MRF and AVN are two

7 Computer Techniques in Environmental Studies 73 different modes of running the Global Spectral Model. MRF mode covers both hemispheres and runs at 00 UTC during 288 hours with a temporal resolution of 24 hours and output resolution of 191 km with 13 vertical levels. AVN mode has another additional two modes: AVN-191 and AVN-111. AVN-191 covers both hemispheres and runs at 00, 06, 12 and 18 UTC during 72 hours with a temporal resolution of 6 hours and output resolution of 191 km with 13 vertical levels and AVN-111 covers Northern hemisphere and runs at 00/12 UTC during 48 hours with a time resolution of 6 hours and output spatial resolution of 111 km with 23 vertical levels. Vertical numerical meteorological soundings are downloaded from the NOAA Web server following the scheme of Figure 3. The OP AN A model starts with the meteorological soundings for the whole period of simulation (120 or 168 hours). 3. Discussion and results The results show that the OPANA/Bilbao Air Quality Forecasting System obtains good results when comparing with observations in spite of that the calibration procedure should be completed and additional periods during the year should be checked. Ideally, one year period should be the recommended calibration period for a system such as OPANA/Bilbao. On the other hand, the user input has provided several suggestions to improve the functionality of the Web interface, which is considered to be quite useful not only for the final user (the citizen) but also for the Management Department Operators. Figures 5, 6, 7, 8 and 9 show several patterns when comparing observed results from different air quality monitoring stations in the Bilbao Area with the forecasted air concentrations by using the OPANA/Bilbao system. Ozone results compare well but an important difference is found for the first day of simulations. We believe that the emission module should be refined to take into account such a changes since the simulation period starts on Saturday and finishes on Wednesday. So that the weekend emission inventory seems to require further improvements when observing Figure 8. In Figures, scenario A is the forecasting data and Scenario B is the observed data. Further studies should include Ozone Source Apportionment Technology (OSAT). The Laboratory is actually working on incorporating faster chemical numerical solvers to allow incorporation of different scenarios for reducing or avoiding ozone episodes when known in advance by the OP ANA / Bilbao system.

8 74 Computer Techniques in Environmental Studies Figure 5.- CO observations vs. Forecasting values by using OP AN A/Bilbao. Figure 6.- NO observations vs. Forecasting values by using OPANA/Bilbao e Ihor*:. Jl i \i**j SpKtt: ME. **: mwm. StKlon; Ermio: II Kw IbtlT Un. A Figure 7.- NO2 observations vs. Forecasting values by using OPANA/Bilbao. Figure 8.- O3 observations vs. Forecasting values by using OPANA/Bilbao. \, : Figure 9.- SO2 observations vs. Forecasting values by using OPANA/Bilbao

9 Acknowledgements Computer Techniques in Environmental Studies 75 We would like to thank to DGXIII for funding the EQUAL project which has supported this contribution. The Municipality of Bilbao for providing initial data for EMIMA model. Prof.- Dr. R. Friedrich (Stuttgart, Germany) for providing the IER emission data base for Bilbao domain. Dr. Smiatek from FI References 1. San Jos6, R., Prieto, J.F., Castellanos N. And Arranz J.M. Sensitivity study of dry deposition fluxes in ANA air quality model over Madrid mesoscale area, Computational Mechanics Publications, pp , San Jos6, R., Rodriguez L., Moreno J., Sanz M. And Delgado M. Eulerian and Photochemical Model over Madrid area in a mesoscale context. Computational Mechanics Publications, pp , San Jos6, R., Marcelo L.M., Moreno B. and Ramfrez-Montesinos A. Ozone modelling over a large city by using a mesoscale eulerian meteorological and transport model: Madrid case study. Plenum Press New York and London, pp , Flassak T. And Moussiopoulos N. Simulation of the sea breeze in Athens with an efficient non-hydrostatic mesoscale model. Elsevier. Amsterdam, pp Grell G.A., Dudhia J. And Stauffer D.R. A description of the fifth-generation Penn State/NCAR Mesoscale Model (MM5). NCAR Technical Note/TN- 398-STR Jacobson M.Z. and Turco R.P. SMVGEAR: A sparse-matrix vectorized gear code for atmospheric models. Atmospheric Environment. 28, pp Gery M.W., Whitten G.Z., Killus J.P. and Dodge M.C. A photochemical kinetics mechanism for urban and regional scale computer modelling. Journal of Geophysical Research, 94, D10, pp

10 76 Computer Techniques in Environmental Studies 8. Wesely M.L. Parameterization of surface resistances to gaseous dry deposition in regional scale numerical models. Atmospheric Environment, 23, pp Sela J. The NMC spectral model. NOAA Technical reports NWS 30, U.S. Department of Commerce. NOAA, National Weather Service, 36pp Arakawa, A. and W. H. Shubert, 1974: Interaction of a Cumulus Ensemble with the Large-Scale Environment, Part I. J. Atmos. Sci., 31, Asselin, R., 1972: Frequencyfilterfor time integrations. Mon. Wea. Rev., 100, Kalnay, E. and M. Kanamitsu, 1988: Time Scheme for Stronglyt Nonlinear Damping Equations. Mon. Wea. Rev., 116, Kanamitsu, M., J.C. Alpert, K.A. Campana, P.M. Caplan, D.G. Deaven, M. Iredell, B. Katz, H.-L. Pan, J. Sela, and G.H. White, 1991: Recent changes implemented into the global forecast system at NMC. Wea. and Forecasting, 6,