Influence of precipitation forcing uncertainty on drought monitoring in South Asia

Abstract:

Drought is one of the major natural disasters in South Asia (SA), affecting the agriculture, water supply, ecosystem, economy and livelihood. To mitigate the impact of drought on agriculture in SA, we develop a state-of-the-art agricultural drought monitoring tool [1]. The South Asia Drought Monitor (SADM) provides simulated soil moisture at high resolution (0.1°) in near-real-time using an advanced hydrological model. Model-based soil moisture inherently comes with uncertainty that could be associated with selection of hydrological model, forcing dataset, calibration strategy among others. In this study, we evaluate the implication of using different precipitation forcing datasets on simulated soil moisture and the resulting drought characteristics. The mesoscale hydrologic model(mHM, https://www.ufz.de/mhm) is used to reconstruct soil moisture using three precipitation forcing datasets. These are CHIRPSv2, PGFv3 and ERA5 from 1982 to 2018 at 0.25° resolution. All the other model inputs such as temperature (PGFv3 at 0.25° resolution), terrain elevation, soil, geology, land cover, and LAI are kept constant. Based on the historic reconstruction of soil moisture, Soil Moisture Index (SMI) is estimated using a non-parametric kernel-based cumulative distribution function [2]. Besides, the drought statistics such as areal extent, duration and magnitude are calculated using a spatio-temporal clustering algorithm [2]. The estimated SMIs using three different precipitation datasets (CHIRPSv2-SMI, PGFv3-SMI and ERA5-SMI) show similar dynamics, though there are some variations. PGFv3-SMI, ERA5-SMI have high correlation (r = 0.90) compared to ERA5-SMI, CHIRPSv2-SMI (0.63) and PGFv3-SMI, CHIRPSv2-SMI (0.69). Comparison of areal extent (A), duration (D) and magnitude (M) show all three precipitation products can identify the major drought events, but there are differences in the actual drought area, duration and magnitude. One of the major droughts identified by all three precipitation products is the drought of 2015-16. ERA5 based cluster analysis shows the event of 2015-16 is the fourth largest drought (M = 2598.7, A = 16%) in the study domain. However, the same event ranks as the eighth largest drought with CHIRPSv2 (M = 1750, A= 12.8%) and the eleventh largest drought with PGFv3 based output (M= 580, A= 8%). The study highlights the importance of addressing uncertainty due to precipitation forcings prior to making conclusions on drought characteristics.

References:

[1] Saha, T. R., Samaniego, L., Shrestha, P. K., Thober, S., and Rakovec, O., (2020). Development of a drought monitor for South-Asia, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-20389. https://doi.org/10.5194/egusphere-egu2020-20389.

[2] Samaniego, L., Kumar, R. and Zink, M., (2013). Implications of Parameter Uncertainty on Soil Moisture Drought Analysis in Germany, Journal of Hydrometeorology. DOI: 10.1175/JHM-D-12-075.1.