A Multi-variate Probabilistic Framework for Assessment of Heat Impact on Yields in India’s Wheat Bowl.

Abstract:

India accounts for more than a tenth of the global wheat production, with the wheat-growing belt along the Indo-Gangetic Plains (IGP) forming a significant contributor. Wheat yield from this region is reported to show high sensitivity to temperature variability and thus, is vulnerable to weather and climate change. Therefore, a quantitative appraisal of wheat response to the regional temperature variability is essential for measures on food security in India in a changing climate. This study first investigates observed changes in the intensity, frequency and spatial extent of heat stress events for wheat cultivation in Punjab, Haryana and Uttar Pradesh – the wheat bowl of India. Heat stress events for wheat cultivation are defined as episodes when maximum day-time temperatures exceed the reported senescence threshold of 34°C during the grain-filling stage (Feb-March). A Generalized Linear Model (GLM) based statistical framework is used to study the changes in these events. Maximum day-time temperatures, number of heat stress days, and area under heat stress are all found to significantly increase from 1967 to 2018.

Further, this study also proposes a copula-based probabilistic framework to quantify the relationship between wheat yield and (i) area under heat stress, and (ii) count of heat stress days in a year. This framework is used to derive the changing probability of yield exceeding target yields, for different observed climate conditions. The probability of yields in the region exceeding average yields is found to decrease from 61% to 39% with an increase in area under heat stress from 0 to 25%. These probabilities decrease from 61% to 36% for a similar change in the number of days under heat stress. The wider range in the probability estimates for the latter case is also suggestive of greater sensitivity of yields to the count of heat-stress days during the observed period. These estimates will be useful in projecting wheat response under future climate change with potential applications in decision making.