Deciphering rainfall patterns in Haryana, India : a comprehensive analysis of optimal probability distributions

Abstract

This research aimed to examine the trends of extreme rainfall events and determine the most suitable probability distribution for rainfall in Haryana state, which is divided into two agroclimatic zones: eastern and western. Five stations (Ambala, Karnal, Hisar, Sirsa and Bawal) were selected from across the state for this analysis. The study processed and analyzed 36 years (1985-2020) of daily rainfall data from each station to identify the maximum one-day and five-day rainfall, as well as monsoonal and total rainfall. The Man-Kandal test and Sen’s slope estimator were used to assess trends in these events. Ten probability distributions were chosen for the rainfall analysis, followed by goodness-of-fit tests such as the Kolmogorov-Smirnov test, Anderson Darling and the chi-square test at a 0.01 level of significance to select the best-fit probability distribution. The study found that one-day maximum rainfall in the state varied significantly, ranging from 23 mm (at Hisar) to 203 mm (at Sirsa). Notably, there was a significant increasing trend in one-day maximum rainfall at Bawal (southern Haryana) at a rate of 0.95mm/year. The maximum five-day rainfall was highest at Ambala (560.8 mm) and lowest at Hisar (29.6 mm). Monsoon rain accounted for 78% of the mean annual rainfall, which ranged from 373.6 mm (at Sirsa) to 843.5 mm (at Ambala). The probability analyses indicated that the General Extreme Value function was a good fit for most rainfall events at most locations. However, based on rainfall variables, the Lognormal (3P) function was the best fit for most locations when explaining the distribution of one-day maximum rainfall events in Haryana. For five-day maximum rainfall events as well as monsoonal rainfall, the Gen. Extreme Value function was the best fit; and for annual rainfall events, both the Gen. Extreme Value and Normal functions performed well.