Assessment of the Weather Research and Forecasting Model in Simulation of Rainfall for Khorasan Razavi Province, Iran

This paper aims to assess and simulate rainfall in the cities of Khorasan Razavi Province to warn and control floods using the Weather Research and Forecasting (WRF) model. For this purpose, the two rainfall events were selected as the representative of the most severe rains that caused a lot of human and financial losses to the province. Accordingly, the five WRF model schemes, i.e., Purdue-Lin (Lin), WRF Single-Moment class 3, 5, 6 (WSM3, WSM5, WSM6) and WRF Double-Moment class 5 (WDM5), were used to perform rainfall simulation. Also, the six verification methods, i.e., threat scores (TS), false alarm ratio (FAR), hit rate (H), false alarm rate (F), Peirce skill statistic (PSS), and R -squared ( R 2 ), were employed to determine the simulation accuracy. The findings showed that for the desert climates, the WDM5 scheme with R 2 of 78%; for the semi-desert climates, the Lin scheme with R 2 of 98%; for the temperate mountain climates, the WSM3 and WSM6 schemes with R 2 of 98% and 72%; and for the cold mountainous climates due to their high altitude and mountain, all schemes with R 2 of 1 were employed. The research results also showed that the TS verification method for the Lin and WDM5 schemes have more acceptable results with 66% and 63% rather than the other schemes. Also, the FAR for the Lin scheme with the lowest result (36%) is the best scheme in simulation and rainfall forecast for flood risk warnings in the province. Overally, the research results showed that foresting WRF model schemas according to the climate of each region could be suitable for flood warnings and control, and provide more time to protect people’s lives and property in the next 24 h.