Markov Chain analysis of rainfall of Coimbatore

Abstract

Rainfall is considered one of the most important weather parameters which helps in deciding the time of sowing, pest and disease management and harvesting. Markov chain analysis deals with predicting future values based on past values. In the present study, Markov Chain analysis was used to predict the future probability of monthly rainfall and examine the pattern and distribution of rainfall using daily rainfall data from the year 1982 to 2016 (34 years) in the Coimbatore district. This study mainly analysed the probability of rainfall in the Coimbatore district of Tamil Nadu based on Markov chain process. Based on the National Center for Hydrology and Meteorology, the intensity of rainfall per day was categorized and a day is considered as no rain if rainfall was less than 0.1 mm, low rain if rainfall was between 0.1 mm to 10 mm, moderate rain if rainfall was between 10 mm to 20 mm and heavy rain if rainfall was above 20 mm. By calculating the transition probability matrices and steady-state probability matrices for each month based on the conditional probability of rain on a particular day given that rain on the previous day which is to predict the state of rainfall on the next day. This study reported that the availability of water for crop production is higher during the winter, pre-monsoon, the onset of the southwest monsoon, and at the end of the northeast monsoon. There may be a scarcity of water from August to November for agricultural activities. Based on this study, farmers can plan for a better cropping system in advance to get a better yield.