A UCM Approach for Forecasting the Seasonal Rainfall Patterns in Coastal Andhra Pradesh, India 1901–2017

The changes in amount and pattern of seasonal rainfall have a significant impact on agriculture and water resources management in Coastal Andhra Pradesh (CAP) of India. This paper presents modeling and forecasting of the seasonal rainfall patterns in CAP using an unobserved components model (UCM) with the hidden components trend, seasonal, cyclical and irregular. The seasonal rainfall data for CAP were provided by the India Meteorological Department and the analysis for the four rainfall seasons, namely, winter, pre-monsoon, monsoon and post-monsoon. The UCM with deterministic level, deterministic trigonometric seasonal, deterministic cycle and stochastic irregular components is selected from the parsimonious models for forecasting the seasonal rainfall patterns in CAP based on the Bayesian information criterion (BIC), significance tests and statistical fit. The model parameters are obtained by using the maximum likelihood method, and the validity of the selected UCM is determined using normal correlation diagnostics and LJung–Box test statistics for residuals. The forecasting of seasonal rainfall patterns for the years 2018–2020 was carried out with the help of the selected UCM. Further, the UCM forecast reveals that winter rainfall will be around 36.4 mm in 2018, 11.3 mm in 2019 and 54.4 mm in 2020; pre-monsoon rainfall will be 91.3 mm in 2018, 135.1 mm in 2019 and 106.4 mm in 2020; monsoon rainfall will be 686.4 mm in 2018, 663.9 mm in 2019 and 652.9 mm in 2020; post-monsoon rainfall will be 286.2 mm in 2018, 267.9 mm in 2019 and 309.1 mm in 2020.