Evaluation of Global Precipitation Databases with Observed Precipitation Values in Balkhlochay Basin.

Global databases provide complete source of required precipitation information for users. The aim of this study is to investigate the accuracy of the TRMM, NCEP and MERRA global precipitation databases in estimating the daily and monthly precipitation in the Balkhaluchai catchment area of Ardabil province. Precipitation data of Nir station the three mentioned databases were extracted from 2002 to 2018. Mann-Kendall, Mann-Whitney and standard normal tests were used to check the trend and homogeneity of precipitation data. For three global precipitation databases, amount of R², RMSE, NSE, RE and GMER were determined. The results of homogeneity test showed that all P-values were higher than 0.05 and the precipitation data were completely stochastic and homogeneous and no trends were observed in them. In both TRMM and NCEP databases, the R² coefficient is good and is above 0.6. The MERRA database has had a weaker performance in this case. For the monthly period and the two TRMM and NCEP databases, the NSE index was higher than 0.5, which indicates the reliability of the precipitation data produced in these two stations. At the daily and monthly level, for all three global databases the GMER parameter values were obtained smaller than unity, which indicates that all global databases have overestimates. The overestimation of precipitation for the MERRA global database is much more severe than other two databases. MERRA database has a much weaker performance than NCEP and TRMM databases. In general, the TRMM database has a better efficiency than other databases. [ABSTRACT FROM AUTHOR]