REFS
From EGEE-see WIki
Contents |
REFS On Line
You can find the daily weather forecast produced by refs in the following link: http://refs-meteo.see-grid.sci.ei
Application summary
One common problem in weather forecasting derives from the uncertainty related to the chaotic behaviour of the atmosphere that is conveyed to the issued weather forecasts. Therefore, a well-accepted innovative approach is to use a multitude of weather models and base the final forecast not only on the predictions of one model (deterministic forecast) but on an ensemble of weather model outputs (multi model ensemble forecasting). Another ensemble forecasting technique is based on perturbing the initial conditions provided to individual models, in order to generate inter-forecast variability depending on a realistic spectrum of initial errors.
Advances in numerical weather prediction (NWP) and related applications have been always very closely related with advances in computer sciences as NWP requires a large number of numerical calculations that are also parallelizable. So the porting of any NWP application to the grid is a natural choice. Indeed, in the frame of SEE-GRID-SCI the regional infrastructure in the area of South Eastern Europe will provide the environment for the development and deployment of a Regional scale Multi-model, Multi-analysis ensemble forecasting system. This system will comprise the use of four different weather prediction models (multi-model system). Namely the state-of-the-art numerical weather prediction models BOLAM, MM5, ETA, and NMM will be ported on the Grid infrastructure. The above models will be run for the same region many times, each initialized with various initial conditions (multi-analysis). With this procedure not only one deterministic forecast but a multitude of forecasts will be produced. The Grid infrastructure will be also used to coordinate, collect and analyze the outputs from all models for the generation of probabilistic forecasts over the area of central and eastern Mediterranean.
The application uses input from the National Center for Environmental Prediction (NCEP) in USA, where a global ensemble forecasting system is operated. Indeed on a daily basis the global ensemble members from NCEP are made available on an ftp server and these fields are ftp’ed and used as input to our regional ensemble prediction system. Upon arrival of these data, a set of respective jobs will be prepared and submitted to the Grid with the abovementioned requirements. Apparently prompt response from the Grid middleware for allocating and successfully completing the jobs is essential for both models.
Weather models used for the ensemble
BOLAM model
The first model used in the regional ensemble forecasting system is the BOLAM hydrostatic model. The version of BOLAM used in this study is based on previous versions of the model described in detail by [6] and [5]. It uses Arakawa C grid rotated lat.-lon. coordinates and ? vertical coordinate As for the physical parameterisations it uses vertical diffusion (surface layer and planetary boundary layer parameterization) depending on Richardson number[29][30]; the microphysical scheme proposed by [43] and the Kain-Fritsch convective parameterisation scheme [21]. In the version of Kain-Fritsch scheme implemented in BOLAM, an additional modification, regarding the delaying of downdraft occurrence has been introduced.
BOLAM model is used for operational weather forecasting at the National Observatory of Athens (NOA) since 1999. An evaluation of these operational forecasts in the Mediterranean region is given in [26] with very encouraging results concerning mainly precipitation forecasts. For the scope of this study one domain is defined consisting of 135x110 points with a 0.21 deg horizontal grid interval (~23 km) centered at 41*N latitude and 15*E longitude, covering the area of the Eastern Mediterranean. In the vertical, 30 levels are used while model top has been set at about 10 hPa. The vertical resolution is higher in the boundary layer and, to a lesser extent, at the average tropopause level. This domain coincides with the outer domain used for the deterministic operational model chain running at NOA since 1999. BOLAM is a serial application written in Fortran and poses modest requirements in terms of CPU power and memory consumption.
MM5 model
MM5 model (Version 3) is a non-hydrostatic, primitive equation model using terrain-following coordinates [10]. Several physical parameterization schemes are available in the model for the boundary layer, the radiative transfer, the microphysics and the cumulus convection. In order to select a combination of microphysical and convective parameterization schemes that better reproduce wet processes, Kotroni and Lagouvardos [24] performed a comparison of various combinations of schemes for cases with important precipitation amounts over E. Mediterranean. This comparison showed that the combination of Kain-Fritsch [21] parameterization scheme with the highly efficient and simplified microphysical scheme proposed by [43] provides the most skilful forecasts of accumulate precipitation for a grid spacing of 24 km. For that reason, both the operational chain of MM5 at NOA but also the operational chain of the regional scale multi-analysis ensemble system based on MM5 model, use the combination of these two schemes. Concerning the choice of the planetary boundary layer (PBL) scheme, the scheme proposed by [14] is used.
One grid has been defined and used at the regional ensemble system, with 24-km horizontal grid increment covering the major part of Europe, the Mediterranean and the northern African coast. In the vertical twenty-three unevenly spaced full sigma levels are selected. This grid coincides with the outer grid of the operational MM5 model chain used at NOA since 2001. MM5 is a parallel application, also developed with Fortran, that uses MPICH v1.2. This model is sensitive to various factors related to the execution environment and the model may not work properly for specific combinations of library versions, OS distributions and compiler options.
ETA model
The Eta Model is a state-of-the-art atmospheric model used for research and operational purposes. The model is a descendent of the earlier HIBU (Hydrometeorological Institute and Belgrade University) model, developed in the seventies in the former Yugoslavia [39]. In the eighties, the code has been upgraded to the Arakawa-style horizontal advection scheme of [16], then rewritten to use the eta vertical coordinate [40], and subsequently, at NCEP, supplied with an advanced physics package[17] [38]. It became officially operational at NCEP on 8 June 1993 [4], and was operational until 2003. In its various versions, the model has been and/or is widely used in numerous countries, including Algeria, Argentina, Belgium, Brazil, Cameroon, China, Costa Rica, Cyprus, Czech Republic, Denmark, Egypt, Finland, Germany, Greece, Iceland, India, Israel, Italy, Malta, Tunisia, Turkey, Peru, Philippines, Serbia, Montenegro, South Africa, Spain, Sweden, and the United States.
The name of the model derives from the Greek letter eta which denotes the vertical coordinate [34], one of the model features. The model orography is formed of steps. The steps can have slopes in the current version [37]. Model variables are distributed on the Arakawa E-grid. Major features of the Eta dynamical core are: the eta vertical coordinate [34], resulting in quasi-horizontal coordinate surfaces, and thus prevention of pressure-gradient force errors due to steep topography than can occur with terrain-following coordinates; forward-backward scheme for time differencing of the gravity-wave terms, modified so as to suppress separation of solutions on two C-subgrids of the model's E-grid [32][15]; the Arakawa approach in space differencing, with conservation of enstrophy and energy [16], energy conservation in transformations between the potential and the kinetic energy in space differencing [40]; option to run the model in a nonhydrostatic mode [19]. The model physics package comprises: convection schemes [3][18][20]; cloud mycrophysics [12]; radiation scheme [25][11]; land surface scheme [8] with 12 types of vegetation and 7 types of soil texture, 4 soil layers; turbulence and PBL: Mellor-Yamada 2.5, and Monin-Obukhov similarity theory in the surface layer, with Paulson stability functions.
While the primary use of the model has been for regional weather prediction and NWP type applications (for a review, see [35]), the model has been very successful also in regional climate and seasonal prediction applications (e.g., [1][9][22]). One grid centered at 15E longitude and 39N latitude has been defined and used at the regional ensemble system. The horizontal grid increment is 0.2 deg, while 45 levels are used in the vertical and the model top is set at 50 hPa. ETA is also a parallel application developed with Fortran, that uses MPICH v1.2.
NMM model
NMM is a non-hydrostatic mesoscale grid point model on semi-staggered Arakawa E grid [2]. The basic model characteristics are : hybrid sigma pressure vertical coordinate; controlled nonlinear energy cascade through energy and enstrophy conservation [16]; explicit time difference scheme with splitting [33][15]; Melor-Yamada level 2.5 parametrization scheme for planetary boundary layer [17][18]; Monin-Obukhov similarity theory for atmospheric surface layer with shallow dynamic turbulent layer at the bottom [18] and viscous sublayer; 2nd order horizontal diffusion, diffusion coefficient depends on deformation and turbuelnt kinetic energy; surface processes, evaporation, snow, snow melting, hidrology [17][18] in NOAH-LSM (Land Surface Model); large scale precipitation and modified Bets-Miler-Janji? convection scheme for deep and shallow convection [3]; NMC radiation scheme, GFDL (Geophysical Fluid Dinamic Laboratory).
One grid centered at 15E longitude and 39N latitude has been defined and used at the regional ensemble system (Fig. 1). The horizontal grid increment is 0.2 deg, while 38 levels are used in the vertical and the model top is set at 50 hPa. The NMM is a parallel application, developed with Fortran, that uses MPICH v1.2.
The Regional Ensemble Forecasting System (REFS)
The Regional-scale Ensemble Forecasting System (REFS hereafter), is based on the use of four numerical weather prediction limited-area models: BOLAM, MM5, NMM, and ETA. For the REFS system deployment the initial and boundary conditions are provided from the Global Forecast System (GFS, NCEP, USA). At NCEP 20 perturbed forecasts are run 4 times daily out to 16 days at a horizontal resolution of ~105 km. For the current application 10 members of the global GFS ensemble system including gridded fields of the 10 perturbed initial conditions and the resulting forecasts at 6-hour intervals are used to initialize each model and to nudge the boundary conditions. All simulations are initialised at 0000 UTC and the duration of the regional ensemble forecast simulations is 72 hours. So in total the REFS consists of 10x4=40 model forecasts. This multitude of weather forecasts should be averaged in order to provide the REFS products.
General workflow
The procedure of running each one of the 40 model forecasts or ensemble members hereafter is briefly described in the following:
• Initial and boundary conditions downloading. This is a common step for all 4 models. As mentioned, the initial and boundary conditions are provided by the National Centers for Environmental Protection (NCEP) of the US National Oceanic and Atmospheric Administration (NOAA). These data are made available to the public through the National Operational Model Archive and Distribution System (NOMADS). NOMADS is a Web-based application that facilitates the preparation and delivery of atmospheric and oceanic data based on a set of user defined criteria. For the purpose of REFS application the atmospheric conditions for a domain extending from 30W to 60E in longitude and from 20N to 70N in latitude are retrieved for a 72-hour forecast period at 6-h intervals. So in total 13 files x 10 different initial/boundary conditions = 130 files in total should be downloaded daily. These data are made available at about 1200 UTC each day but the NOMADS servers offering these data frequently suffer from high system loads. Thus, in many cases the attempts to retrieve the data might fail.
• Initial and boundary conditions pre-processing. This step although necessary for each model, it is model dependent. Indeed, once the initial and boundary conditions are successfully retrieved they pass through a pre-processing step in order to be transformed into the correct format required by each weather model. This pre-processing is done by one or more utility applications specifically developed for this purpose. So for REFS there are four pre-processing chains developed, for the four different models participating in the application.
• Weather model execution. After the data have been pre-processed they are ready to be used as input for the weather model. The execution of the model itself is the core of the workflow and typically the most time-consuming part of the application. It is also typically the most CPU and memory demanding part of the workflow, especially in the case that a parallel weather model is used like the MPI-enabled version of MM5, NMM and ETA.
• Model output post-processing. The final step of the workflow is the post-processing of the data produced by all the model ensemble members. This process includes the storage of the model output, extraction of the part of the output needed for the combined probabilistic forecasts, transferring of part of the output to another host for storage and further processing. The full output of each of the ensemble members can be as large as 200MB depending on the weather model that was used. So in total about 8GB of storage is required every day for storing intermediate and final results. Furthermore, if archiving of results is required storage space becomes a considerable issue for the application.
It is evident that this application consists of a kind of parametric jobs that are well suited for the GRID, since each ensemble member can be carried out in parallel. The results of the integrated application are then made available as a uniform data set in the Logical File Catalogue (LFC), ready to be analyzed. The LFC concept for the REFS application will be discussed later in the text.
Application requirements
The specific technical requirements for each model participating in REFS in terms of infrastructure are summarized in the following table. It is quite evident that the REFS application is highly demanding in terms of processing power and storage capacity and these characteristics motivated the porting of the application to the Grid. The Grid offers vast amounts of processing power, storage capacity and also capabilities for data sharing and collaboration. On the other hand, most scientific grids are offering only open-source tools, supporting software and libraries. Commercial and licensed software usually requires special arrangements in order to be made available to grid users. In our case, all four models require commercial compilers (either Intel Fortran or Portland Group (PGI) Fortran) in order to compile and run properly.
Table 1: Requirements for BOLAM / MM5 / NMM / ETA models
Cores per ensemble: 10 / 60-120 / 88 / 88
Archived Storage: 100MB / 2GB / 605MB / 1155MB
Storage per job on running node: 20MB / 250MB / 290MB / 120MB
Minimum Memory per job on running node: 512MB / 1 GB / 1 GB / 1 GB
Implementation issues
In this section some of the implementation issues will be discussed. Two groups have participated in the development and porting of REFS application in the Grid infrastructure. Namely the National observatory of Athens (Greece) have worked on the porting of BOLAM and MM5 regional model ensemble chains and the South Environment and Weather Agency (Serbia) have worked on the porting of NMM and ETA regional model ensemble chains, while both are working on the integration of all ensemble members towards a regional scale super-ensemble. For that reason there are some common but also some model specific implementation issues. The downloading of initial and boundary data from the NOMADS servers at NCEP as already mentioned is a source for many failures of the REFS application. To overcome this problem a specialized Python application was developed that tries many times and from different servers to retrieve the required data. The application is also multithreaded enabling the fast retrieval of initial data in parallel streams reducing this way the overall time required to transfer all required data.
As it concerns the model execution, three out of the four models used in REFS application are parallelized and need MPI support (MM5, NMM, ETA). Insufficient MPI support in current large scale grid infrastructures is a well known issue that relevant Grid projects are still trying to overcome. MPI support was one of the main reasons for some MM5 job failures. Further the models need specific compilers that are mainly commercial. So the model codes were statically compiled on a User Interface (UI) gLite node. This way we overcame the requirement to have licensed libraries on the target Worker Nodes (WNs). Nevertheless this way we’ve sacrificed on performance since the generated binaries are optimized for the target running machines.
In addition there are some specificities to each model implementation. For example MM5 model presented instabilities as reported previously, and the binary code crashed often with no apparent reason in many grid sites. This problem was addressed after experimentation with different compilers and compilation parameters to adopt Intel Fortran compiler for MM5 and to compile the code in a UI which was custom build to provide an identical environment to that offered by the HellasGrid infrastructure (http://www.hellasgrid.gr) which is part of the SEE-GRID-SCI infrastructure. By doing this we managed to eliminate the model crashes and optimize the performance by minimizing the model’s execution time. On the downside this means that we have to run the code only in a subset of the SEE-GRID-SCI’s infrastructure (the 6 sites comprising HellasGrid).
Job scheduling is an important issue in order to address the failures that would occur either due to the model itself, to the infrastructure or the NOMADS servers. In such cases recovery and resubmission of jobs was required. In order to develop a scheduling monitoring logic on the user side NOA adopted the Ganga framework [41]. Ganga is a Python library which provides an object-oriented programmatic abstraction of the command line tools available by gLite. Using Ganga we are able to have better control on the job management aspects of the application. It also gives us the opportunity to exploit the power of Python for issues related to user-interaction, statistics and log-keeping etc. SEWA developed a shell scipt based system and used crontab functionalities for scheduling the data acquisition and job submission. The script based system is highly flexible, portable and easily maintained. For the purpose of the calculating job success rates Event Logger application, developed by RCUB, is used to some extent, but will be fully implemented in the system in the future.
Currently the integrated analysis of all ensemble members is developed. Indeed after all the members finish the following post-processing procedure is followed:
• from each ensemble member model output predefined variables are written out for further analysis (temperature at 850 hPa and 500 hPa, geopotential height at 500 hPa, 6-hour rainfall, 10-m wind and mean sea-level pressure). As each model uses a different (but very close) horizontal analysis and domain projection the data are interpolated a lat-lon projection on the common domain with 0.25x0.25 degrees resolution. A common format has been set for all processed output.
• the super-ensemble is created if a predefined number finished ensemble member is available (50% that is 20 ensemble members). The analysis includes the calculation of parameters such as probability of exceedance of various precipitation thresholds, spaghettis of temperature at 850 hPa and geopotential at 500 hPa as well as meteograms of temperature at 850 hPa for the capital cities of the countries included in the common model domain.
REFS storage scheme
The jobs related to REFS are parametric and each simulation is carried out in different nodes, but the results of all ensemble members should be available to be analysed in a Logical File Catalogue (LFC). So for REFS application a LFC has been created and used for storing of:
• Various data files produced by the models. Such files typically contain the raw output (output produced daily by each ensemble member of a model), post-processed output and intermediate results from failed model executions.
• Software packages. Packages containing the model binary executables, additional required scripts and auxiliary data. These are fetched from the LFC to the worker node before the model execution.
Below the storage scheme that has been developed for REFS application is presented. Data are stored under /data folder and executables reside under /software folder. Data are organized per region. For the time being only one region exists (SEEurope) which is dedicated for the data produced currently by the SEE-GRDI-SCI REFS applications. Daily produced output (either raw data, failed info or final artifacts) are stored in separate directories depending on the production date. These directories are created everyday (early in the morning), before the execution of the models, by a cron job running in NOA UI node. The directory names follow the YYYYMMDD pattern. The file names are split in five parts: (a) producer Code <PRD>, that can be either NOA or SEW, (b) model name <MOD> that can be BOL, MM5, NMM and ETA, (c) production date <YYYYMMDD> in UTC, (d) ensemble member identifier <NN> which is a two digit code identifying a unique ensemble member that produced the raw data. For the processed output instead of the producer code <PRD> the parameter <PAR> is included in the name. Although some of the above fields can be considered redundant since the LFC folder hierarchy is enough to identify both the production date and the type of data, this is considered useful in order for the above information to be easily identifiable when a file has been retrieved from the LFC to the UI.
/grid
/meteo.see-grid-sci.eu
/REFS
/data
/REGION
/rawoutput
/<YYYYMMDD>
<PRD>-<MOD>-RAW-<YYYYMMMDD>-<NN>.tgz
/processedoutput
/<YYYYMMDD>
<PAR>-<MOD>-RAW-<YYYYMMMDD>-<NN>.tgz
/failed
/<YYYYMMDD>
<PRD>-<MOD>-FAIL-<YYYYMMMDD>-<NN>.tgz
/artifacts
/<YYYYMMDD>
<PRD>-<MOD>-<YYYYMMMDD>.tgz
/shared
/software/<PRD>
REFS performance Statistics
In this section some statistics on the performance of REFS application from the computing point of view are given. Namely we will report on the performance of each ensemble model chain. The periods referred to varying depending on the model and the developing group.
The performance statics of BOLAM and MM5, models refer to a four-month period (mid-April 2009 until mid-August 2009). MM5 was ran with 60 processors (6 cores allocated per member) on the three reserved sites in the HellasGrid infrastructure. BOLAM ensembles, in the same time, were submitted through WMS to all sites in SEE-GRID-SCI supporting the Meteo VO (a Virtual Organisation setup in SEE-GRID-SCI to accommodate all meteorology applications of the project). In both cases the deadline set for the ensemble to complete was 400 minutes (6:40hrs).
It should be noted that despite the fact that advanced reservation techniques have been applied in order to ensure that the required amount of CPU cores will be immediately available for the MM5 ensemble, temporary failures of other services (notable LFC, VOMS and WMS for BOLAM) some days still impede the successful completion of the ensemble.
Statistics for ETA and NMM models refer to job submissions made since the beginning of May 2009 until the end of October 2009. Both ETA and NMM jobs required 8 CPU cores per ensemble member. During the development and test phase ensembles were submitted to all sites. This resulted in large number of failures in execution. Therefore, in the operational run phase, the ensembles were submitted to three sites which provided necessary administrative and technical support (Institute of Physics Belgrade, RCUB, Bioengineering Research and Development Center Kragujevac.
References
[1] Altshuler, E., M. Fennessy, J. Shukla, H. Juang, E. Rogers, K. Mitchell, and M. Kanamitsu, 2002: Seasonal simulations over North America with a GCM and three regional models. COLA Tech. Rpt. 115, 66 pp.
[2] Arakawa, A., and V. R. Lamb, 1977: Computational design of the basic dynamical process of the UCLA general circulation model. Methods in Computational Physics, 17, Academic Press, 173-265.
[3] Betts, A. K., and M. J. Miller, 1986: A new convective adjustment scheme. Part II: Single column tests using GATE wave, BOMEX and arctic air-mass data sets. Quart. J. Roy. Meteor. Soc., 112, 693-709.
[4] Black, T. L., 1994: The new NMC mesoscale Eta Model: Description and forecast examples. Weather Forecasting, 9, 265-278.
[5] Buzzi, A., and L. Foschini, 2000: Mesoscale meteorological features associated with heavy precipitation in the southern Alpine region. Meteorol. Atmos. Phys., 72, 131-146.
[6] Buzzi, A., M. Fantini, P. Malguzzi and F. Nerozzi, 1994: Validation of a limited area model in cases of Mediterranean cyclogenesis: surface fields and precipitation scores. Meteorol. Atmos. Phys., 53, 137-153.
[7] Charney, J., 1951: Dynamical forecasting by numerical process. Compedium of meteorology. Americal Meteorological Society, Boston, MA. [8] Chen, F., Z. Janjic, and K. Mitchell, 1997: Impact of atmospheric surface-layer parameterizations in the new land-surface scheme of the NCEP mesoscale Eta Model. Bound.-Layer Meteor., 85, 391-421.
[9] Chou, S. C., J. F. Bustamante, and J. L. Gomes, 2005: Evaluation of seasonal precipitation forecasts over South America using Eta model. Nonlin. Proc. Geophys., 12, 537-555;
[10] Dudhia, J., 1993: A non-hydrostatic version of the Penn State/NCAR nesoscale model: validation tests and simulation of an Atlantic cyclone and cold front. Mon.Wea. Rev., 121, 1493-1513.
[11] Fels, S. B., and M. D. Schwarzkopf, 1975: The simplified exchange approximation: A new method for radiative transfer calculations. J. Atmos. Sci., 32, 1475-1488.
[12] Ferrier, B. S., Y. Jin, Y. Lin, T. Black, E. Rogers, and G. DiMego, 2002: Implementation of a new grid-scale cloud and precipitation scheme in the NCEP Eta Model. 19th Conf. on Weather Analysis and Forecasting/15th Conf. on Numerical Weather Prediction, San Antonio, TX, Amer. Meteor. Soc., 280-283.
[13] Foster, I., Kesselman, C., 2003: The Grid: Blueprint for a New Computing Infrastructure, 2nd edn., Morgan Kaufman, San Francisco.
[14] Hong, S-Y, and H-L Pan, 1996: Nonlocal boundary layer vertical diffusion in a medium-range forecast model. Mon Wea. Rev., 124, 2322-2339.
[15] Janjic, Z. I., 1979: Forward-backward scheme modified to prevent two-grid-interval noise and its application in sigma coordinate models. Contrib. Atmos. Phys., 52, 69-84.
[16] Janjic, Z. I., 1984: Nonlinear advection schemes and energy cascade on semi-staggered grids. Mon. Wea. Rev., 112, 1234-1245. [17] Janjic, Z. I., 1990: The step-mountain coordinate: Physical package. Mon. Wea. Rev., 118, 1429-1443.
[18] Janjic, Z. I., 1994: The step-mountain eta coordinate model: Further developments of the convection, viscous sublayer, and turbulence closure schemes. Mon. Wea. Rev., 122, 927-945.
[19] Janjic, Z. I., J. P. Gerrity Jr., and S. Nickovic, 2001: An alternative approach to nonhydrostatic modeling. Mon. Wea. Rev., 129, 1164-1178.
[20] Kain, J. S., 2004: The Kain?Fritsch convective parameterization: An update. J. Appl. Meteor., 43, 170?181.
[21] Kain, J.S., and J.M. Fritsch, 1993: Convective parameterization for mesoscale models: The Kain-Fritsch scheme. The Representation of Cumulus in numerical models, Meteor. Monogr., No 46, Amer. Met., Soc., 165-177.
[22] Katsafados P., A. Papadopoulos, and G. Kallos, 2005: Regional atmospheric response to tropical Pacific SST perturbations. Geophys. Res. Lett., 32, L04806, doi:10.1029/2004GL021828.
[23] Kotroni V. and K. Lagouvardos, 2004: Evaluation of MM5 high-resolution real-time forecasts over the urban area of Athens, Greece. J. Appl. Meteor (in press).
[24] Kotroni, V. and K. Lagouvardos, 2001: Precipitation forecast skill of different convective parameterization and microphysical schemes: application for the cold season over Greece. Geoph. Res. Let., Vol. 108, No 10, 1977-1980.
[25] Lacis, A. A., and J. E. Hansen, 1974: A parameterization of the absorption of solar radiation in the earth's atmosphere. J. Atmos. Sci., 31, 118-133.
[26] Lagouvardos K., V. Kotroni, A. Koussis, C. Feidas, ?. Buzzi, P. Malguzzi, 2003: The meteorological model BOLAM at the National Observatory of Athens: assessment of two-year operational use. J. Appl. Meteor., 42, 1667-1678.
[27] Lorenz, E. N., 1965: A study of the predictability of a 28-variable atmospheric model. Tellus, 17,. 321-333.
[28] Lorenz, E. N., 1968: The predictability of a flow which possesses many scales of motion. Tellus, 21, 289-307.
[29] Louis, J.F., 1979: A parametric model of vertical eddy fluxes in the atmosphere. Boundary Layer Met., 17, 187-202.
[30] Louis, J.F., M.Tiedtke and J.F.Geleyn, 1982. A short history of the PBL parameterization at ECMWF. Proceedings of ECMWF workshop on PBL parameterization. Reading 25-27 Nov. 81., 59-80.
[31] Mesinger F. and L. Lazic, 2004: Water vapor sources and sinks, and hydrometeor loading in the Eta Model. Research Activities in Atmospheric and Oceanic Modelling, 34, 5.21-5.22 (Available online at http://collaboration.cmc.ec.gc.ca/science/wgne/).
[32] Mesinger, F., 1974: An economical explicit scheme which inherently prevents the false two-grid-interval wave in the forecast fields. Proc. Symp. "Difference and Spectral Methods for Atmosphere and Ocean Dynamics Problems", Academy of Sciences, Novosibirsk 1973; Part II, 18-34.
[33] Mesinger, F., 1977: Forward-backward scheme, and its use in a limited area model. Contrib. Atmos. Phys., 50, 200-210.
[34] Mesinger, F., 1984: A blocking technique for representation of mountains in atmospheric models. Riv. Meteor. Aeronautica, 44, 195-202.
[35] Mesinger, F., 2000: Numerical Methods: The Arakawa approach, horizontal grid, global, and limited-area modeling. General Circulation Model Development: Past, Present and Future. International Geophysics Series, Vol. 70, D. A. Randall, Ed., Academic Press, 373-419.
[36] Mesinger, F., 2005: Eta Model at NCEP: Challenges overcome and lessons learned. Lecture notes, Workshop on "Design and Use of Regional Weather Prediction Models", The Abdus Salam International Centre for Theoretical Physics, Miramare, Trieste, Italy, 11-19 April 2005, 42 pp. Available online at http://agenda.ictp.trieste.it/agenda/current/askArchive.php? base=agenda&categ=a04186&id=a04186s262t1/lecture_notes.
[37] Mesinger, F., and D. Jovic, 2004: Vertical coordinate, QPF, and resolution. The 2004 Workshop on the Solution of Partial Differential Equations on the Sphere, Frontier Res. Center for Global Change (FRCGC), Yokohama, Japan, 20-23 July 2004. ppt in CD-ROM, Vol. 2.
[38] Mesinger, F., and L. Lobocki, 1991: Sensitivity to the parameterization of surface fluxes in NMC's eta model. 9th Conf. on Numerical Weather Prediction, Denver, CO, Amer. Meteor. Soc., 213-216.
[39] Mesinger, F., and Z. I. Janjic, 1974: Noise due to time-dependent boundary conditions in limited area models. The GARP Programme on Numerical Experimentation, Rep. No. 4, WMO, Geneva, 31-32.
[40] Mesinger, F., Z. I. Janjic, S. Nickovic, D. Gavrilov, and D. G. Deaven, 1988: The step-mountain coordinate: model description and performance for cases of Alpine lee cyclogenesis and for a case of an Appalachian redevelopment. Mon. Wea. Rev., 116, 1493-1518.
[41] Mo?cicki J.T., et al., 2009: Ganga: a tool for computational-task management and easy access to Grid resources. Computer Physics Communications, abs/0902.2685.
[42] Palmer, T. N., Molteni, F., Mureau, R., Buizza, R., Chapelet, P., & Tribbia, J., 1993: Ensemble prediction. Proceedings of the ECMWF Seminar on Validation of models over Europe: vol. I, ECMWF, Shinfield Park, Reading, RG2 9AX, UK.
[43] Schultz, P., 1995: An explicit cloud physics parameterization for operational numerical weather prediction. Mon. Wea. Rev., 123, 3331-3343.
[44] Spencer, P. L. and D. J. Stensrud, 1998: Flash flood events: importance of the subgrid representation of convection. Mon. Wea. Rev., 126, 2884-2912.
[45] Tracton, MS, and E. Kalnay, 1993: Ensemble forecasting at NMC: Practical aspects. Weather and Forecasting, 8, 379-398.
Downloads
Source Code
Application source is available for download from svn. Use the following command to check-out the latest stable release:
svn export http://svn.hellasgrid.gr/svn/refs.meteo.see-grid-sci.eu/REFS_0_2_0
You may also be interested in the development release which is also quite stable. Use the following command in this case:
svn export http://svn.hellasgrid.gr/svn/refs.meteo.see-grid-sci.eu/REFS
Documentation
The REFS Users Manual provides step-by-step how to install, configure and run meteo models through the REFS job management framework.
