International Journal of Science, Engineering and TechnologyAuthors: Dr. Pankaj Malik, Mishthi Patodia, Vedant soni, Deepika Kumari, Pragati Agrawal, Jaiswal Tanmay
Abstract: Litchi is a high-value fruit crop traditionally cultivated in regions with favorable climatic and soil conditions. Expanding litchi cultivation into non-traditional areas such as the Malwa region of Madhya Pradesh requires accurate identification of suitable locations to minimize cultivation risks and maximize productivity. This study proposes a village-level suitability assessment framework that integrates geospatial data, climatic variables, soil characteristics, groundwater availability, and satellite-derived vegetation indices with Explainable Machine Learning (XML) techniques. Environmental and agricultural data were collected from multiple sources, including Sentinel-2 imagery, Soil Health Card records, meteorological datasets, and groundwater databases. Several machine learning algorithms, namely Random Forest, XGBoost, LightGBM, CatBoost, and Support Vector Machine, were trained to predict the suitability of villages for litchi cultivation. To enhance transparency and interpretability, SHAP (SHapley Additive exPlanations) and LIME (Local Interpretable Model-Agnostic Explanations) were employed to identify the key factors influencing suitability predictions. The experimental results demonstrated that the XGBoost model achieved the highest classification performance with an accuracy of 94.2%, precision of 93.6%, recall of 92.8%, F1-score of 93.2%, and ROC-AUC of 0.96, outperforming the other evaluated models. SHAP analysis revealed that winter temperature, annual rainfall, soil organic carbon, groundwater depth, and irrigation availability were the most influential parameters affecting litchi suitability. The generated village-level suitability maps identified several highly suitable zones within the Indore, Dewas, and Ujjain districts, while areas experiencing higher temperatures and limited water resources were classified as marginally suitable or unsuitable. The proposed framework provides a transparent and data-driven decision support system for farmers, horticulture planners, and policymakers, facilitating climate-resilient expansion of litchi cultivation in the Malwa region.
DOI: http://doi.org/10.5281/zenodo.21504453
