Characterization and return period analysis of meteorological drought under the humid subtropical climate of Manipur, northeast India

Abstract

Monitoring drought characteristics is crucial for understanding drought behaviour and developing effective mitigation plans. In this study, we analyze the characteristics of meteorological droughts in the eastern Himalayan region by utilizing both the Standardized Precipitation Index (SPI) and Copula functions. In this study, we utilized monthly rainfall data spanning 35 years to estimate three critical characteristics of droughts: duration (D), severity (S), and Intensity (I). To determine the best fit marginal distribution of each univariate drought characteristic, we employed five commonly used probability distribution functions (PDFs). We conducted Kolmogorov-Smirnov (K–S) and Anderson-Darling (A-D) tests.

The bivariate modelling for the joint D-S, S–I, and I-D datasets involves fitting Archimedean families such as Frank, Clayton, Gumbel, and Joe copulas. To perform the trivariate modelling, two meta elliptical copulas, including Normal and Frank, and two Archimedean families, namely Clayton and Gumbel, are fitted using the test statistics BIC (Bayesian Information Criterion) and AIC (Akaike Information Criterion). The cross-validation process using Maximum Likelihood Estimation (MLE) is employed to identify the most appropriate Copula model based on its goodness of fit. This step is crucial for selecting the best model to accurately describe the joint behaviour of drought characteristics. Once the best-fit Copula model is determined, it is utilized to estimate the return period of various drought characteristics, thereby facilitating the investigation of their joint return period. Furthermore, the distribution of S, D, and I classes is categorized into different return periods (T) to facilitate drought management planning.

The findings revealed moderate drought conditions were recorded for SPI 1 and SPI 3 with a 2–5 years return period. For SPI 1, this drought class remains seasonal even for higher return periods. Further, the drought class transitions from seasonal to quarter for SPI 3 and a return period of 10–50 years. Regarding SPI 6 and SPI 12, the drought class is seasonal for a return period of 2 years, but it later progresses into the quarter to long-term drought class.