Hello, A change in allocation has been requested: Requester: raok (Rao Kotamarthi) Project: WRF-CHEM-GVAX Title: Weather and Aerosol forecast products for the Ganges Valley Aerosol Project Description: The Ganges Valley Aerosol eXperiment (GVAX), led by Dr. Rao Kotamarthi, will begin its one-year ground-based observation with the Atmospheric Radiation Monitoring (ARM) mobile facility at Nainital, India,starting June, 2011. This major field study involving multiple Department of Energy laboratories will focus on investigating the potential impact of aerosols on regional scale radiative forcing changes over the Ganges Valley. Measurements of radiative flux, cloud, convection, and aerosol physical, chemical, and optical properties will be made to develop core data sets for scientific analysis. The measurements will range from those made at surface locations that yield point and profile data sets to a few selected aircraft platform-based measurements. However, these measurements and the subsequent data sets must be placed in a larger regional context in order to relate the data to a particular dynamic state of the atmosphere and begin developing hypotheses tha t explain observed phenomena. The regional context will be developed within the conditions generated by a 3-D regional scale climate model. Baseline simulations of 3-D regional scale model will provide the necessary analytical support for many of the scientific studies done under GVAX. In addition, the 3-D model will be a test bed for high performance computing data storage and visualization. We have implemented and run the latest 3.3 beta version of the Weather Research and Forecasting (WRF) model with an online capability of simulating chemistry and aerosols (WRF-Chem) on the Argonne Fusion clusters. The WRF-Chem model has been optimized and uses domain decomposition to divide total work load on the distributed-memory systems. Without chemistry and aerosol modules, the standard WRF CONUS benchmark (48 hr and 12 km forecast over Continental US) completes in 15 minutes on a teraflop system (~250 processors). For the GVAX campaign, the principal investigator will simulate a model domain from 55E to 95E and 0 to 36N, divided into 367x331 12-km grid cells with 28 vertical layers from the surface to the top of the atmosphere. The model boundary conditions will be updated on a daily basis from the gridded NOAA/NCEP global forecasts, which are available online within about 36 hrs to 48 hrs after real time. Therefore, there is a need to perform at least 72 hr model simulations in order to provide 36 hr forecasts required for the field experiment. The test runs on the WRF-Chem model have been completed on a number of nodes. The regional model scales up to 250 cores and needs ~10 hours to complete a 72-hr forecast. The speedup achieved with 72 cores is about 13 relative to 4 cores. Memory used on each node decreases from 4.5 GB, 1.5GB, 1.1 GB, to 0.9 GB while the number of nodes increases from 4, 16, 36, to 72 cores. The ideal computing time to finish daily forecasts is around 3~5 hours. Thus, the daily forecasts is expected to require about 200-250 cores. Because of gas-phase and aerosol chemistry for pollutant forecast output, the WRF-Chem is much more expensive in computation time compared to running the WRF model only. However, forecasts of the geographic distribution and transport pattern of atmospheric gases and aerosols are required for the GVAX campaign. The above estimated time does not include pre-processing of the GFS meteorology fields, post-processing and visualization of the forecast results. In addition t o the model runs, these processes will be done in serial and cost an additional several hours on Fusion. We have setup a Live Access Server (LAS) node at Argonne to serve the model results to the user community. The site (atmos.anl.gov:8080/las) is now operational and we plan to regularly update the website with model results on a daily basis 3-D regional scale model simulations will provide the necessary analytical support for many of the scientific studies done under GVAX. These modeling results of air pollution and meteorological fields will become one of the central ‘data’ products for the experiments archived at the ARM websites. The expectation is that there will be a significant number of scientists from around the world who will be interested in GVAX data. In addition, the 3-D model will be a test bed for high performance computing data storage and visualization. The large volumes of model outputs generated would be used to test or assess the feasibility of using the cloud computing systems (such as Magellan) to perform/extract a pre-determined set of forecast products as and when needed by scientists distributed at disparate locations world-wide. Current: undetermined amount Justification: Requested: 270000 A specific reason has been given: we continue the calculations in a diagnostic mode for the project This needs to be approved and the final allocation amount decided upon. Thank You, The LCRC Accounts System