Comparison of seven precipitation datasets for drought monitoring over the United States/global scale

Drought is a recurrent natural hazard and has impacts on agriculture, ecosystems, economies, and society. The Standardized Precipitation Index (SPI) is one of the most widely used indices for monitoring drought. The SPI can reflect the deficit or surplus of precipitation at different time scales, thus it can be used for drought monitoring at these time scales. At present, more than 30 precipitation datasets are available with different resolutions and scales. This study compares seven of the precipitation datasets with relatively long records and high spatial resolutions to determine which is most appropriate for drought monitoring. These datasets include CRU (Climatic Research Unit) time series precipitation dataset, GPCC (Global Precipitation Climatology Centre) dataset, PREC/L (Precipitation Reconstruction over Land) dataset, University of Delaware Precipitation dataset, PRISM dataset, NLDAS (North American Land Data Assimilation System) dataset, and CPC (CPC US Unified Precipitation) dataset. Monthly SPI were calculated using these different precipitation datasets for the period from 1980 to 2010 over the United States (PRISM, NLDAS and CPC) or at the global scale (CRU, GPCC, PREC/L, U. Del). GLDAS Noah soil moisture data were used for evaluating the SPI datasets. Our monthly and seasonally analyses indicate that over the United States, PRISM is most representative of soil moisture conditions, and at a global scale, CRU dataset is most representative of soil moisture conditions.