Title: Statistical and trend analyses of long-term rainfall data: a case study for Mauritius

Authors: Reena Hansa Seebocus; Michel Roddy Lollchund; Miloud Bessafi

Addresses: Department of Physics, Faculty of Science, University of Mauritius, Réduit 80837, Mauritius ' Department of Physics, Faculty of Science, University of Mauritius, Réduit 80837, Mauritius ' Laboratoire d'Energétique, d'Electronique et Procédés, University of La Réunion, France

Abstract: This paper focuses on the analysis of monthly rainfall data for the period 1950-2016 for Mauritius using statistical techniques and decomposition method. For the statistical analysis, the data are fitted to commonly used probability distribution function for which parameters are estimated using the method of maximum likelihood estimation (MLE). The Anderson-Darling (A-D), Kolmogorov-Smirnov (K-S) and chi-square (C-S) tests are then employed to determine which PDF best fits the data. The results obtained indicate that the log-normal, GEV and inverse Gaussian PDFs best fit the rainfall data at less than 5% significance level. The ensemble empirical mode decomposition (EEMD) method is employed to study the trends in the data. Results obtained are in terms of intrinsic mode functions (IMFs) and the trendline. The analysis reveals that there is a general linear decrease of 1 mm/year in the amount of rainfall.

Keywords: rainfall; statistical techniques; probability distribution function; goodness-of-fit tests; ensemble empirical mode decomposition; EEMD; Mauritius.

DOI: 10.1504/IJHST.2020.109945

International Journal of Hydrology Science and Technology, 2020 Vol.10 No.5, pp.413 - 428

Received: 10 Jul 2018
Accepted: 09 Nov 2018

Published online: 30 Sep 2020 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article