Validation of CHIRPS satellite-based precipitation dataset over Pakistan

Abstract

This study evaluates high resolution Climate Hazard Groups Infrared Precipitation with Stations (CHIRPS) with Surface Precipitation Gauge (SPG) of recent 38 years (1981–2018) through statistical metrics including Mean Error (ME), Mean Absolute Error (MAE), Root Mean Square Error (RMSE) and Correlation Coefficient (CC) on daily, monthly, annual and seasonal timescales over the entire Pakistan. Trends of both the datasets were also assessed using non-parametric Mann-Kendall (MK) and Modified MK (MMK) tests. CHIRPS (SPG) captured maximum precipitation in the range of 8.5–9.0 (8.0–8.5) mm/day in northeastern parts during the Monsoon season. CHIRPS slightly underestimated SPG in the range of −0.80 to −0.17 mm/day on area averaged precipitation of 46 stations for the period 1981–2018. Maximum ME, MAE, RMSE were observed as 2.0 mm/day, 1.8 mm/day and 1.8 mm/day respectively on annual timescale. Strong linear relationship (CC > 0.70) was noted between CHIRPS and SPG on monthly, Winter, Monsoon and Post-Monsoon season timescales. CHIRPS (SPG) showed increasing trends in 43 (26) stations using MK test and almost similar trend behavior was detected using MMK test on monthly timescale. CHIRPS showed increasing trends for all the six administrative regions for the period 1981–2018 using MK test and same trend behavior was observed using MMK test during the Monsoon season. Overall, good agreement between MK and MMK trend tests were observed with trivial variations in most of the administrative regions. Hence, it is concluded that CHIRPS can be a substitute of SPG in low and medium elevated and data scarce regions.