Spatio-temporal variability and trend analysis of rainfall in Wainganga river basin, Central India, and forecasting using state-space models.

Analysis of rainfall distribution and its changing pattern plays a vital role in managing water resources in a region. This work examined the spatio-temporal variability and trend of rainfall on yearly and season-wise scales during 1971–2013 in the Wainganga river basin situated in middle India. The Mann–Kendall (MK) test was implemented for identifying the temporal variation in rainfall trends. The magnitude of this changing trend was estimated by applying Sen's slope (SS) method. A paired sample t-test was also employed to appraise the statistical significance of changes in rainfall data. The results of MK and SS tests show both upward and downward trends. The results show positive and negative trends in the south-eastern and northwestern parts of the study area for annual rainfall. The monsoon rainfall also shows very close proximity with annual rainfall data trends. The t-test states that the observed changes in precipitation are statistically significant. After trend and pattern analysis, state-space models (SSMs) were used to predict rainfall for future scenarios. Four SSMs (single, double, and triple exponential, autoregressive integrated moving average (ARIMA)) were employed for rainfall prediction. The analysis shows that the ARIMA is best for rainfall prediction with architecture of (0,0,0) (2,1,0) and can be used in Wainganga River basin. [ABSTRACT FROM AUTHOR]