The performance assessment of PERSIANN-CCS-CDR rainfall estimates for hydrometeorological applications

Abstract:

Long-term precipitation estimation at fine spatiotemporal resolution is essential for many hydrometeorological studies. However, most available precipitation estimation datasets lack either the fine spatiotemporal resolution or long-term data record, which hinder characterization of annual, seasonal, and diurnal trends. The Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks-Cloud Classification System-Climate Data Record (PERSIANN-CCS-CDR) is a new addition to the Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks (PERSIANN) global satellite precipitation data family which is developed to address these shortcomings. This dataset is a climate data record precipitation product which provides 3-hourly rainfall estimates at 4km resolution from 1983 to the present for 60S-60N latitude band. The high spatiotemporal resolution of the dataset makes it suitable for studying the variability of precipitation pattern from diurnal to seasonal cycle at local to continental scale. Furthermore, the PERSIANN-CCS-CDR dataset has superior performance in capturing the spatial and temporal pattern of extreme events in comparison to the Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks-Climate Data Record (PERSIANN-CDR). Comparing the performance of PERSIANN-CCS-CDR and PERSIANN-CDR with Stage IV data for the 2005 Hurricane Katrina supports the superior performance of the new dataset for extreme events. In this study, we employ PERSIANN-CCS-CDR to characterize the annual, seasonal, and diurnal pattern and variability of precipitation and extreme events over different parts of the Contiguous US. The lengthened data record of PERSIANN-CCS-CDR along with its fine resolution allows a detailed investigation of the precipitation pattern, especially for extreme events in remote and data scarce regions.