Air Quality Forecasting


From 2009 through spring of 2013, the Israeli Air Force (IAF) and the National Center for Atmospheric Research (NCAR), with oversight provided by Project Chief Scientist, Dr. Dorita Rostkier-Edelstein of the Israel Institute for Biological Research (IIBR), have jointly developed an advanced mesoscale numerical weather prediction (NWP) system. Core technologies of this system include the NCAR Weather Research and Forecasting (WRF) Real-Time Four-Dimensional Data Assimilation (RTFDDA) and forecasting system, and WRF DA (data assimilation) 3DVAR (3-D variational data assimilation). It was the first forecasting system within the Research Applications Laboratory (RAL) to employ such a hybrid data assimilation system, blending conventional meteorological observations with satellite radiance measurements. This system, called “MAGEN” – Model for Advanced Generation of 4D Weather, has been in production in real-time to provide accurate weather forecasts at high model resolution onsite at the IAF facility. Since then, continuing, concerted efforts between the IAF and NCAR have been made to investigate ways to enhance and expand the features of the MAGEN system. Of all inclement weather events commonly found in the Eastern Mediterranean Sea, dust storms are identified as the target component to add to the MAGEN system, since they are considered to be among the most severe environmental problems, especially causing degradation of visibility and air quality, in this region of the world. A follow-on project, MAGEN Dust and Visibility Initial Operating Capability (DVIOC) was funded by the Israel Ministry of Defense (IMOD) to support this effort and pursue the goal of enhancing the MAGEN system with a preliminary dust and visibility forecasting capability for real-time experimental forecast applications.

2017 Accomplishments

The DVIOC project marked the first time RAL’s in-house WRF enhancements, including detailed observation nudging, inline observation quality control, and improved adaptive time-step were adapted to be running alongside the chemistry module of WRF (WRF-Chem). The project included two case studies that served multiple purposes -- finding the balance or tradeoffs between the level of sophistication of the WRF-Chem schemes/physics and computational efficiency, and providing confidence in the preliminary dust and visibility forecasting capability through extensive verification. The two case studies focused on dust storms occurring during 1) April, 2015, and 2) March, 2016. Conventional meteorological observations are used to verify the weather aspect of the WRF-Chem runs.  Other observations from various platforms are used to verify the dust storms and visibility. These observations include Aerosol Optical Depth (AOD) from the Moderate Resolution Imaging Spectroradiometer (MODIS), 10 and 2.5 µm particulate matter (PM10 and PM2.5) concentrations, surface visibility measurements, and satellite RGB products, an example of which is shown in Figure 1 (dust storm in magenta) vs the model dust storm (vertically integrated dust concentrations) of the same time period in Figure 2. Model dust storms in general verify well against observations, although timings of storm onsets and peaks could be slightly off by a few hours. We found the timings of the onsets critically depend on a sufficient spin-up time period. This finding was folded into the design of the final dust/visibility forecasting component of the MAGEN system. The case studies also revealed lack of locally formed dust storms over the Arabian Peninsula. Further investigation identified high soil moisture contents in the global model in this particular region as a possible culprit that prevents the dust storms from materializing. The final deliverable system was installed on the computing hardware in late October, 2017 and is now serving in real-time production at the IAF base.

2018 Plans

Although strengths and weaknesses have been identified from the case studies, the discoveries represent only about two weeks in a year. Decision has been made to let the system continue to run in production mode for a year to further identify areas of improvement. However, several possible follow-on tasks to refine the dust/visibility forecast capability have been proposed by NCAR. These proposed activities include dust remote sensing and in-situ observation data assimilation and verification; tuning up dust-weather interaction processes; and simulation of man-made and sea salt aerosol.