Evaluation and modification of Aphrodite precipitation network in estimating monthly and annual precipitation in central parts of Iran

Precipitation estimation and preparing of accurate rainfall maps in regions with no recorded climatic data is critical in hydrology and meteorology studies. This is more important especially in desert areas that do not have dense meteorological stations. The dense network of Aphrodite daily precipitation provides daily rainfall data at 0.25 * 0.25 degrees in Iran. But before using it must be evaluated and corrected.

The aim of this study was to investigate the accuracy of Aphrodite network precipitation data in four desert catchments of Iran from the Central Plateau catchment of Iran in monthly and annual time scales. For this purpose, long-term monthly and annual average data of 9 synoptic and 2 rain gauge weather stations of the Meteorological Organization were used. The climate of the stations varies from extra arid, arid and semi-arid. To validate the data at each time scale, the statistics of correlation coefficient, root mean square error and mean bias error were used. Then the network data were modified by coefficient of means ratio. Then, the accuracy of the modified data was compared with the two usual methods of kriging interpolation and weighted inverse distance.

The MBE showed that Aphrodite network data have a good correlation with station data, but it seems that rainfall network estimation is more than the actual data, which some correction coefficients is used to bias correction of rainfall data. The correction factors vary between 0.60 in low rainfall months to 0.94 in high rainfall months. Evaluation of the modified network in different months showed that in addition of solving network over estimation problem, the error was greatly reduced. as well as in the case of annual rainfall, the root mean square error was reduced from 71.25 to 15.19 mm, that indicating, increase network performance by five times after applying correction coefficients.

  Comparison of modified network with the interpolation methods showed that, the modified network is much more efficient than the two interpolation methods. Therefore, the use of obtained correction coefficients for Aphrodite network modification is recommended to improve the accuracy of the Aphrodite precipitation network.