Evaluating homogeneity of monsoon rainfall in Saraswati river basin of Gujarat, India

Abstract

This study investigates presence/absence of homogeneity in 32-yr (1981–2012) rainfall record of four individual monsoon months (June–September) and monsoon season (JJAS) for 16 stations in Saraswati River basin, Gujarat, India. Temporal homogeneity is examined by Hartley, Link-Wallace, Bartlett, and Tukey tests, and spatial homogeneity is tested by Levene’s and Tukey tests. Coefficient of variation for rainfall in June (72–163%), July (48–100%), August (78–114%) and September (93–127%) indicate a large variability in comparison to that in JJAS period (45–60%). Correlation coefficient (r) finds moderately significant (r ≥ 0.7) to highly significant (0.7 > r ≥ 0.3) relationships in rainfall for 68 (57%), 120 (100%), 120 (100%), 109 (91%), and 120 (100%) pairs of stations in June, July, August, September, and JJAS, respectively. Distribution of rainfall is uniform and stable in July, and hence, kharif crops may be sown in July to mitigate impact of uncertain rainfall on agriculture. Dissimilar results of four homogeneity tests justify approach of adopting multiple statistical tests. Considering the likely findings of Link–Wallace and Tukey tests, both are recommended for testing homogeneity. Non-homogeneity is found at Paswadal (June, September, and JJAS), Navawas (August and JJAS), Palanpur (JJAS), and Pilucha (September) stations. The Levene’s test reveals spatial homogeneity in July, August, September, and JJAS; and non-homogeneity in June. Hierarchical cluster analysis delineates four clusters of rainfall stations in four months and JJAS with their geographically distinct locations and a remarkable difference in inter-annual rainfall dynamics.