Enhancing coconut yield potential: A climate-smart land suitability analysis using machine learning

Abstract

Coconuts (Cocos nucifera L.) play a critical role in Kerala's agricultural landscape, serving as a cornerstone of agricultural income and significantly contributing to the state's economy. Despite their economic importance, variations in land and climate conditions across the region lead to inconsistencies in coconut yield and productivity, limiting the full potential of coconut farming. This study aims to enhance coconut cultivation in Kerala by i) comparing various machine learning (ML) and deep learning (DL) models to identify the optimal model for soil suitability prediction; ii) developing a climate model to assess climate suitability; and iii) integrating both soil suitability and climate suitability models to classify the study regions into suitability categories—highly suitable, moderately suitable, less suitable, and not suitable, for coconut farming. Using a dataset from the Soil Survey Department, the XGBoost algorithm was applied to classify soil suitability in the study area (Thiruvananthapuram, Kerala, India). Climate suitability was assessed using the MaxEnt model. Finally, GIS tools were used to combine these results into a comprehensive suitability map. For soil suitability prediction, we tested various machine learning and deep learning models, ultimately selecting XGBoost as the optimal model due to its near-perfect accuracy of 100%. The MaxEnt model enhanced the assessment of climate suitability with an accuracy of 67.7%, providing insights into optimal farming conditions. This study presents an integrated land and climate suitability model for coconut farming, demonstrating the effectiveness of ML and DL models for soil suitability analysis. This approach offers a robust framework for improving coconut cultivation and can be applied to other regions and crops.