Modeling and Forecasting of Rainfall Trends based on Historical Data in Bungoma County, Western Kenya using Holt Winters Method

Abstract

Rainfall is termed as a meteorological phenomenon which is very useful for daily human activities. Most of the population depend on it for domestic purpose and agriculture, hence it is vital for farmers to know rainfall trends and patterns prevailing in their locality. In Kenya, rainfall variability has caused hunger to many people. In agriculture section, water problem is the most critical constraint in food productions. Extreme weather conditions which is occasioned by climatic change and weather variability makes a small scale farmer in Bungoma to face very high risks of reduced productivity. Scarcity of water is a severe environmental constraint to plant productivity hence drought causes loss in crop yield. Majority of the population in Bungoma do depend on agriculture either directly or indirectly therefore analysis of rainfall data for long periods will provide a lot of information about rainfall variability and this will help to better the agricultural activities of Bungoma farmers. The main aim of the study was to model rainfall patterns and hence predict the future rainfall trends in Bungoma region using Holt winters method in the context of the time series. This is because time series analysis plays an important role in modelling, predicting and forecasting meteorological data such as humidity, temperature, rainfall and other environmental variable. Therefore, data for Bungoma monthly and yearly rainfall patterns for the period 1988-2021 was obtained from the Kenya meteorological department. Collected data was analyzed using Holt winters method by R software. Rainfall data was found to be seasonal implying that most of the rainfall occurred in a specific period each year. Forecasted rainfall had increasing and decreasing prediction intervals and this implied that rainfall could either start decreasing or increasing. The data was found to be non-stationary due to presence of seasonality and rainfall trends. Findings of the study will make it possible to facilitate planning and management of water for both domestic and agricultural use in Bungoma region.