AMI is in the forefront of meteorological modeling  with extensive experience in real-time weather forecasting, regional and site-specific modeling, meteorological data analysis , and preparation of meteorological inputs for air quality modeling applications. Advanced models operational on AMI's in-house computers include the mesoscale, nonhydrostatic, 3-D models (MM5 and the newer WRF) and diagnostic wind models such as DWM. AMI has developed a PC-based weather forecasting system that can provide real-time forecasts worldwide for energy resource development and natural disaster management. These forecasts include accurate wind and temperature predictions in diverse settings (complex terrain, coastal, offshore).  In addition, AMI has developed and implemented several software packages that allow the efficient generation of modeling inputs, computer visualization  and Web-based presentation of modeling results.  Representative meteorological modeling projects performed by AMI include the following:
 

Forensic Meteorological Modeling of Pesticide Releases

AMI has recently used the state-of-the-art  Weather Research and Forecast  (WRF) model  to investigate the impacts of airborne releases of pesticide to nearby vineyards located in Oregon.  Three outer nested grids of 27, 9 and  3  km grid resolutions are used with the WRF model to resolve the regional airflow. Due to complex terrain, the innermost grid used a fine grid resolution of 1 km  to accurately simulate local winds and temperature. These winds are used to assess the transport of pesticide during spraying operations to nearby vineyards. 

Weather Forecast for Wind Energy Plant

Under a contract with Electric Power Research Institute (EPRI) and U.S. DOE National Renewal Energy Laboratory (NREL), AMI has performed weather and energy generation forecasts for the Southwest Mesa wind energy plant located in west Texas. AMI is using state of the science tools, i.e. the mesoscale model MM5 and the diagnostic wind model CALMET, to generate weather forecasts. Four nested grids of 45, 15, 5 and 1.67 km grid resolutions are used with the MM5 model. A fine grid resolution of 1 km is used in CALMET to resolve the complex topography of this site and accurately forecast winds and temperature. Forecast winds are used to predict the power production from the wind turbines onsite. Forecasts are generated twice daily (at 00 and 12 UTC) and accessed from a website.     
 

Regional Meteorological Modeling for Photochemical Modeling in the San Diego Air Basin

For San Diego County, AMI performed a comprehensive modeling study that used the mesoscale model MM5 for generating three-dimensional wind fields and other meteorological inputs for photochemical grid modeling with the CAMx model. The non-hydrostatic MM5 model was applied to the August 3-7, 1997 ozone episode that occurred during the Southern California Ozone Study (SCOS97).  Meteorological fields predicted by MM5 such as winds and temperature have been compared against actual field measurements, and the MM5 performance has been evaluated  using both graphical statistical techniques. Meteorological inputs generated by MM5 are used by the CAMx photochemical model for assessing the effectiveness of ozone control strategies proposed for the San Diego area. 

Real-time Tropical Storm Weather Forecasting

In recent years, several tropical storms have hit  the coastal areas  of Vietnam, especially in Central Vietnam. Torrential rains accompanying these storms have caused severe flooding that results in heavy losses of life and  widespread damages.  We  have successfully applied the AMI-developed PC-based version of the advanced, three-dimensional prognostic mesoscale model MM5 to simulate two of these storms: Elvis (1998) and Eve (1999) . Results of these simulations have shown that  MM5 can be used to  provide real-time forecast of future storms.

The MM5 72-hour forecasts, available from the Website http://www.vnbaolut.com ,  cover most of Southeast Asia, from south China in the north to north Borneo in the south, and Thailand in the west to the Philippines in the east. It uses two modeling domains: the outer grid (Domain D01) is comprised of 41x53 grid cells with a grid resolution of 72 km, and an inner domain (D02) focusing on Vietnam has a finer resolution of 24 km with  67x82 grid cells. Both modeling grids use 23 vertical levels.

Four MM5 forecasts, valid for 72 hours,  are available daily,  for 00 UTC (07 A.M. Vietnam time), 06 UTC (1 P.M. Vietnam), 12 UTC (7 P.M. Vietnam) and 18 UTC (1 A.M. next day Vietnam). All forecasts are completed within 30 minutes of their forecast time, e.g. the  00 UTC  forecast is available around 7:30 A.M. Vietnam time and the 12 UTC by  7:30 P.M. Vietnam time.  Initial and boundary conditions have been derived from the NCEP AVN global analyses. The forecasts are updated with the latest weather information and reports from satellites, radars, weather stations, airports and ships in the region. The AVN global forecasts employ a coarse grid resolution of over 100 km. With a finer grid resolution (24 km in the inner domain), the MM5 forecasts are expected to be more accurate than other Internet forecasts, such as those from CNN and Yahoo, that rely on these coarse global forecasts. Our MM5 forecasts also provide predicted rainfall amount and wind speed , information that are important in storm emergency planning. The rainfall amounts predicted by the MM5 forecasts compare favorably against the actual measurements reported in recent storms.

PC-Based Version of the Nonhydrostatic, 3-D Mesoscale Model Version 5 (MM5)

The mesoscale model MM5 has been developed by U.S. National Center of Atmospheric Research (NCAR) and is the most accurate and updated model available today. It is being used in mesoscale meteorological modeling, including real-time forecasting, by government agencies and well-known universities, both in the U.S. (e.g., National Weather Service and NASA) and abroad (e.g., Europe, Hong Kong and Taiwan). The MM5 model  requires extensive computer resources and it normally runs on powerful and expensive computers such as a Cray supercomputer or high-end Unix workstation. Recently, we have ported the complete MM5 package to run on a Intel/AMD PC under the Linux operating system. For the MM5 benchmark test from NCAR, a Pentium-III PC with a single 450 Mhz CPU (costing less than $1K) has been  shown in the following table to run 25% faster than the Cray YMP supercomputer that costs millions!

Daily Peak Temperature Predictions for Energy Load Forecast

For Southern California Edison, AMI processed and analyzed the ETA model output of temperature for several locations in Southern California. Some of these locations are in complex terrain. These ETA predictions were utilized to evaluate the performance of a statistical scheme for predicting daily peak temperature. These peak temperature predictions are used in forecasting the daily energy demand.

Comparison of Wind Field Models in the Santa Barbara Channel

Under a contract to Santa Barbara County APCD, AMI performed a comparative study of state-of-the-art wind models in a region dominated by complex terrain and land-sea breezes . The models used in this study include: the diagnostic wind model DWM, the prognostic CSU Mesoscale Model (CSUMM), and the MMFDDA model. Developed by AMI, the MMFDDA is based on CSUMM with the capability of four-dimensional data assimilation (FDDA) of actual field measurements. The wind models were applied to the SCCCAMP September 5-7, 1984  ozone episode which occurred in the South Central Coast Air Basin. Results of the UAM photochemical modeling using the generated wind fields are used to assess the accuracy of wind models.

Meteorological and Photochemical Modeling of Large-Scale Albedo Changes in the Los Angeles Air Basin

Under contract to Southern California Edison, AMI performed a modeling study to assess the effectiveness of large-scale albedo changes as an innovative ozone control strategy. AMI adapted the state-of-the-art mesoscale model MM5 for predicting the potential temperature changes and generating three-dimensional wind fields for photochemical grid modeling. The non-hydrostatic MM5 model was applied to the August 26-28, 1987 ozone episode which occurred during the Southern California Air Quality Study (SCAQS) in the South Coast Air Basin. AMI also developed software to interface between MM5 and the photochemical grid model UAM. The MM5 and UAM models were applied to simulate both the base case and the scenario with maximum surface albedo modifications derived from a satellite-based land use database.

Comparison of Wind Field Models in the Los Angeles Air Basin

AMI performed a comparative study of state-of-the-art models capable of generating three-dimensional wind fields used in photochemical grid modeling. The models used in this study include: the diagnostic DWM model, the prognostic CSUMM model, and the MMFDDA model. Developed by AMI, the MMFDDA is based on CSUMM with the capability of four-dimensional data assimilation of actual field measurements. The wind models were applied to the August 26-28, 1987 ozone episode which occurred during the Southern California Air Quality Study (SCAQS) in the South Coast Air Basin. Results of the UAM photochemical modeling using the generated wind fields are used to assess the accuracy of wind models.