Assessing the Accuracy of European Center for Medium Range Weather Forecasts (ECMWF) Reanalysis Datasets for Estimation of Daily and Monthly Precipitation

An accurate estimation of precipitation is important and necessary for flood simulation, drought monitoring and water resources management. Currently, most parts of the world are suffering from the lack of the rain gauge observations and the spatial coverage of ground observations aren’t enough and continues. One of the most important precipitation datasets is the model-based precipitation datasets, by which the satellite techniques, the general circulation models (GCMs) and the land surface models (LSMs) are integrated to provide high temporal and high resolution datasets for all parts of the world. This datasets can compensate the lack of adequate ground observation gauges or can be considered as an alternative for ground observations, especially in ungauged regions. In this research the accuracy of the most important reanalysis datasets, called ECMWF, for estimation of daily and monthly precipitation over the SefidRood watershed for the time period of 2000-2008 was investigated. In addition, for better assessment of the proposed precipitation datasets, TRMM dataset was used. Findings on daily and monthly time scales, show that the correlation coefficient (CC) between observed and ECMWF dataset is so remarkable, especially in south, central and west parts of the study area. For instance, the CC values of the average precipitation of ECMWF data versus gauge datasets in both daily and monthly time steps were estimated to be about 0.83, 0.94, respectively, while the CC values for TRMM dataset versus gauge datasets were estimated to be 0.32 and 0.57, respectively. In contrast to reanalysed datasets, one of the most important weakness of the precipitation datasets such as TRMM is that they estimate the rainfall only based on the cloud thickness and its available water. Moreover, according to the categorical verification statistics in both time spans, ECMWF due to having low value of false alarm ratio (FAR) and high values for accuracy and probability of detection (POD) yields acceptable results over the SefidRood watershed. SefidRood watershed is a large scale region and contains different climate and topographical conditions and hence the results of this research can be used as an appropriate guidance for other similar areas. Based on the findings in this study it’s highly recommended for using this rainfall dataset as one of the best alternatives for ground observations, especially in data sparse regions that accessing to ground datasets is so hard or almost impossible.