Detecting drought events over Iran during 1983–2017 using satellite and ground-based precipitation observations

Abstract

Timely and accurate detection of drought events using reliable data with continuous spatial coverage is of great importance to adapt efficient mitigation strategies. Despite several studies that have been carried out to validate satellite precipitation products in Iran, the performance of satellite precipitation products in precipitation amount estimation and drought events detection has not been assessed yet simultaneously. It is unclear if good estimates of mean precipitation values by satellites can also provide good performance in detecting drought events. Here, we used a comprehensive framework to evaluate the accuracy of the satellite-based precipitation and their performance in detecting drought events against ground-based data. We evaluated three satellite precipitation products, namely, TRMM TMPA 3B43 v7, CHIRPS, and PERSIANN-CDR using multiple evaluation metrics, and three different drought indices (Standardized Precipitation Index-SPI, Standardized Precipitation Evapotranspiration Index-SPEI, and Multivariate Standardized Precipitation Index-MSPI) over different climate zones of Iran during 1983–2017. Our results indicated that TRMM and PERSIANN-CDR had mainly overestimation while CHIRPS had mostly underestimation in precipitation over the different climates of Iran. The precipitation overestimation (underestimation) in TRMM, PERSIANN-CDR, and CHIRPS were observed in 80.0% (20.0%), 71.4% (28.6%), and 45.7% (54.3%) of stations, respectively. The most overestimation in satellite products was found to be in the arid-related climate regions, while the most underestimation was detected in humid-related climate regions. The average correlation coefficient values of PERSIANN-CDR (0.6) and TRMM (0.6) products against ground observations were higher than CHIRPS (0.4) dataset. Further, we found a lower correlation coefficient (0.3) between monthly satellite-based and ground-based precipitation in arid-related regions (with Root Mean Square Error, RMSE, ~ 10 mm per month) of the country but a high correlation coefficient (0.6) in humid-related climates (with RMSE ~100 mm per month). The results of the drought assessment indicated that the most severe meteorological drought events identified by SPI, SPEI, and MSPI were in 1998–2001, 2008–2009, 2010–2012, and 2016–2017 in all climate zones. The most intense drought events were in arid and semi-arid climate zones. Overall, in terms of the statistical metrics at multiple timescales, TRMM and PERSIANN-CDR outperformed CHIRPS, when compared against ground-based observations, for drought events detection. Also, TRMM and PERSIANN-CDR with higher scores for Probability of Detection (POD) and Critical Success Index (CSI) and lower scores for False Alarm Ratio (FAR) showed a good ability to detect drought events in arid and semi-arid climates compared to other climate zones. Although the CHIRPS estimated precipitation with acceptable accuracy, it did not detect drought events as the TRMM and PERSIANN-CDR did.