Impact of assimilation of satellite, prepbufr and GPSro data on wind speed and dust concentration forecasts in WRF-Chem model

Dust storms are significant phenomenon in south west Asian countries like Iraq, Syria and Iran. Mineral dust is generated by wind erosion over arid and semiarid land surfaces and is transported locally and over vast distances, causing adverse environmental and weather problems. Recent draughts over dust sources in Iraq and Syria have remarkably increased dust events in the area particularly over west of Iran. Real time prediction of dust storms especially quantitative forecasting of dust concentration has become highly desirable to alleviate its damaging consequences. In this study the impact of the assimilation of satellite radiance, prepbufr and GPSro data in the wind speed and dust load forecasts of WRF-Chem model using WRFDA system are investigated. Prepbufr data are a collection of surface and upper air observations and GPSro data are GPS radio occultation data. These data are operationally collected by the National Centers for Environmental Prediction (NCEP). Data assimilation is applied to two dust events starting from Iraq and Syria borders on August 31st 2015 and June 15th 2016. For each case, two experiments are conducted. An experiment assimilating above mentioned data with three dimensional variational (3D-Var) intermittent assimilation method and a control simulation with no assimilation. The assimilation cycles in the intermittent method consist of three subsequent analyses at 00, 06 and 12 UTC. After the last assimilation cycle, the model is integrated for 48 hours in the future. In variational data assimilation a key element to get a qualified analysis is the accurate specification of error statistics for the background forecast. For the calculation of background error, the model with the same specification for the experiments is run for the whole January 2014 at 0000 and 1200 UTC and the 12- and 24-h forecasts are used to calculate the background error using the National Meteorological Center (NMC) method with CV5 option. The horizontal resolution of the domain is 21 km with 142×130 grid points covering Iran and western neighboring countries. The model has 41 vertical levels with the model top at 25 hPa. Initial and boundary conditions are taken from NCEP Global Forecast System (GFS) model with the horizontal resolution of 0.5º×0.5º.Results show that the agreement between spatial distribution of dust load prediction of the model and Meteosat-10 satellite RGB images is improvedusing data assimilation especially in first forecast hours. Quantitative comparison of 10 m, 850 hPa and 700 hPa model wind speed with surface observation data and ERA-Interim ECMWF reanalysis data show up to 11% improvement in RMSE especially in first forecast hour times. The positive impact of data assimilation is decreased as the forecast length increases