Assessing the long-term fluctuations in dry-wet spells over Indian region using Markov model in GEE cloud platform

Abstract

The long-term fluctuations in dry-wet spells were assessed at standard meteorological week (SMW) over India using Climate Hazards Group InfraRed Precipitation with Station (CHIRPS) rainfall data. The weekly sum of rainfall was embedded in Markov Chain Probability Model in Google Earth Engine (GEE) platform to compute initial and conditional probabilities of dry-wet spells during 2009-2020. An effective monsoon window (23rd SMW–39th SMW) was identified where initial probabilities (IPs) of dry (Pd) and wet (Pw) spells intersect at 50% probability level. Significant spatiotemporal variation of IPs was observed with initiation and withdrawal of monsoon over India. The analysis of co-efficient of variation (CV) showed low CV (<60%) in Pd and high CV (>60%) in Pw in semi-arid and arid regions whereas northern, central and eastern regions observed high CV (>60%) in Pd and low CV (<40%) in Pw. The drought prone and moisture sufficient zones were indentified based on the analysis of long-term frequency distribution of dry-wet spells and trend. Inter-comparison of IPs between CHIRPs with IMD (Indian Meteorological Department) and NOAA CPC (National Oceanic and Atmospheric Administration/Climate Prediction Centre) showed encouraging results. The study provides baseline reference for climate-resilient agricultural crop planning with respect to food security.