Dry and wet spell probability analysis using Markov chain model for planning jute-based cropping systems in eastern India

Abstract

Knowledge about the rainfall distribution, frequency of dry and wet spells, and the probable weeks of the onset and termination of monsoon season are the prerequisite for planning agricultural operations particularly for rainfed crops such as jute in eastern India. In this paper, therefore, the marginal, conditional, and consecutive dry and wet week probabilities were analysed using Markov chain model, and the forward and backward accumulation method was used for computing probabilities of onset and withdrawal of monsoon season for planning jute-based cropping systems in eastern India. Both the methods were applied for weekly rainfall data of the four stations situated in jute growing areas of eastern India such as Nagaon (period: 2011-2020), Pundibari (period: 1998-2016), Barrackpore (period: 1983-2020) and Bhubaneswar (period: 2008-2020). In jute season (13-29 weeks), the probability of dry weeks was maximum (3-81%) in Barrackpore and minimum (0-19%) in Nagaon. The probability of dry week preceded by another dry week was maximum in Bhubaneswar and minimum in Nagaon. The probability of two and three consecutive dry weeks was maximum (0-65% and 0-52%, respectively) in Barrackpore and minimum (0-9% and 0-6%, respectively) in Nagaon. The onset and withdrawal of monsoon season were also computed for all the four stations data. This analysis will be helpful for water management and determining or adjusting proper sowing time of jute or any other crops especially rainfed crops so that the critical water-sensitive phenological phases might not coincide any water stress situation and thus ensuring to reach potential yield of that particular crop.