Assessment of Spatio-Temporal Variability and Probabilistic Prediction of Annual Rainfalls in a River Catchment of Udaipur, Rajasthan

In the present study, the spatial and temporal variability of annual rainfall is assessed and predicted at selected return periods by using probabilistic approach for the Wakal River catchment of Udaipur, Rajasthan, India. Annual rainfall data for five raingauge stations (i.e. Dewas, Jhadol, Ogna, Kotra and Gogunda) of the study area for a period of 15 years (1992–2006) were analysed by box plots and normal probability plots. Normality was tested by applying four normality tests. Spatial and temporal variability of the rainfall was evaluated by using Levene's analysis of variance (ANOVA) and Mann-Kendall tests. Furthermore, nine probability distributions were fitted to the annual rainfall series. The best-fit distribution was selected based on Chi-square and Kolmogorov-Smirnov tests. Box plots revealed few outliers in the rainfall series of Dewas, Jhadol and Ogna stations. Due to outliers, the annual rainfall of these three stations was found to be non-normal by normal probability plots, which was further confirmed by the normality tests. After removing outliers, the annual rainfalls were found to be normal. Moreover, the annual rainfall does not have significant spatial and temporal variability. The results of the goodness-of-fit tests indicated that log Pearson type-III is the best-fit distribution and rainfalls are predicted at selected return periods.