APPLICATION OF DATA-DRIVEN MACHINE LEARNING MODELS FOR RAINFALL PREDICTION: A CASE STUDY OF SUB-HUMID KONKAN REGION OF MAHARASHTRA
Loading...
Date
2023-03
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Rainfall is one of the most influential hydrologic variables required for number of applications in water resource management, specifically in the agriculture sector. Rainfall prediction has gained utmost importance in recent times due to its association with natural disasters such as floods, landslides, drought, etc. Rainfall prediction can help decision makers of a variety of fields in making decisions regarding important activities like crop planting, agricultural operations, sewer system operations, and managing natural disasters like floods and droughts. This study presents a comparative analysis of four data-driven machine learning models, namely, Multiple Linear Regression (MLR), Random Forest (RF), Categorical Boosting (CatBoost), and Extreme Gradient Boosting (XGBoost) for predicting daily rainfall of Dapoli station, located in the Ratnagiri district of Maharashtra. Historical daily meteorological observations starting from 2005 to 2021, for seventeen years, were collected for the analysis from Department of Agronomy, College of Agriculture, Dapoli. The meteorological parameters data include the parameters such as rainfall (R), minimum temperature (Tmin), maximum temperature (Tmax), relative humidity in the morning (RH1), relative humidity in the afternoon (RH2), wind speed (WS), sunshine hours (SS), vapor pressure in the morning (VP1), vapor pressure in the afternoon (VP2), and evaporation (E). The whole dataset was split into two parts, the training dataset and the testing dataset. The data were in the proportion of 80% and 20% for the training and testing phase, respectively for the prediction of rainfall. The qualitative and quantitative performance of the aforementioned models was assessed using four statistical properties, viz. coefficient of determination (R2), Kling Gupta efficiency (KGE), root mean square error (RMSE), and index of agreement (d). After a detailed analysis, it was concluded that the RF model performed consistently well for predicting the daily rainfall at Dapoli station.
Description
Keywords
Citation
Theses of M. Tech