Analysis of Tea Plantation Suitability Using Geostatistical and Machine Learning Techniques: A Case of Darjeeling Himalaya, India
Netrananda Sahu, Pritiranjan Das, Atul Saini, Ayush Varun, Suraj Kumar Mallick, Rajiv Nayan, S. P. Aggarwal, Balaram Pani, Ravi Kesharwani, Anil Kumar
Abstract
This study aimed to identify suitable sites for tea cultivation using both random forest and logistic regression models. The study utilized 2770 sample points to map the tea plantation suitability zones (TPSZs), considering 12 important conditioning factors, such as temperature, rainfall, elevation, slope, soil depth, soil drainability, soil electrical conductivity, base saturation, soil texture, soil pH, the normalized difference vegetation index (NDVI), and land use land cover (LULC). The data were normalized using ArcGIS 10.2 and the models were calibrated using 70% of the total data, while the remaining 30% of the data were used for validation. The final TPSZ map was classified into four different categories: highly suitable zones, moderately suitable zones, marginally suitable zones, and not-suitable zones. The study revealed that the random forest (RF) model was more precise than the logistic regression model, with areas under the curve (AUCs) of 85.2% and 83.3%, respectively. The results indicated that well-drained soil with a pH range between 5.6 and 6.0 is ideal for tea farming, highlighting the importance of climate and soil properties in tea cultivation. Furthermore, the study emphasized the need to balance economic and environmental considerations when considering tea plantation expansion. The findings of this study provide important insights into tea cultivation site selection and can aid tea farmers, policymakers, and other stakeholders in making informed decisions regarding tea plantation expansion.