Verification of the Weather Research and Forecasting Model (WRF) for the Heavy Precipitation Forecasting in the Karun basin. A case study (8-9 February 2006)

Accurate weather forecasting plays a key role in people daily life and agriculture. By using of numerical weather forecasting models could predict meteorological parameters, but the model outputs have errors which should be verified. Model errors are due to the random and systematic errors. The eyeball and statistical verification methods help us to evaluate model outputs accuracy. In February 2006, a heavy precipitation occurred in Karun basin that had a lot of economic damages. The purpose of this study is evaluating WRF model for forecasting Heavy Precipitation in Karun Basin.
Material and In this study, the precipitation data from synoptic stations and TRMM satellite is used. The boundary and initial conditions of model are FNL data. The WRF model outputs for a case of rainfall, which has been caused the flooding in the Karun region, with observation precipitation data were verified. The model has been run for 24 and 48 hours predictions with two nested of network steps of 15 and 45 km. To evaluate the accuracy of WRF numerical prediction model, the model results, were investigated with the reality of the eyeball verification and statistical points of view.
The results of eyeball verification indicated that the precipitation pattern and its amount in the area are correctly predicted. The statistical verification results in three thresholds, the occurrence or absence of rainfall, average and heavy precipitation in the first and second 24 hours were evaluated. The results show that the model in the first and second 24 hours prediction of occurrence or absence of the rainfall is very precise and the accuracy of results is close to 100 percent. For the second threshold, the model accuracy in rainfall prediction is high but the accuracy was higher in the second 24 hours, so that in the second threshold and the second 24 hours, the number of cases correctly predicted is high. Also, in the second threshold in the first 24 hours, the model has been dry, in other words the number of days when precipitation has occurred, has been less than anticipated value. In the second 24 hours, the model has been wet, the number of days that the precipitation has not happened, but the model has predicted, is high. In the third threshold, the prediction accuracy is higher in the first 24 hours and in two cases, the model has been wet. In other words, for the third threshold, the model accuracy in predicting heavy rainfall has been in average value in the region. Finally, the results show that in the occurrence of the flood forecasting in Karun basin, the model is reliable.
In February 2006, a heavy and unprecedented precipitation occurred in Karun basin that had a lot of economic damages. In this study, the WRF model outputs for a case of rainfall, which has been caused the flooding in the Karun basin, with observation precipitation data were verified. The model has been run for 24 and 48 hours predictions with two nested of 15 and 45 km. To evaluate the accuracy of the WRF numerical prediction model, the model results, were investigated with the reality of the eyeball verification and statistical points of view. Finally, the results show that in the occurrence of the flood forecasting in Karun basin, the model is reliable. Iran has a complex topography and accurate predicting of meteorological parameters especially precipitation and 2 meter temperature is very difficult. Although the model output is reliable, but using of the best model configuration and ensemble forecasting could improve the model outputs.