Assessment and Prediction of Forest Ecosystem Health using Machine Learning and Edaphic Factors

Forests are vital for biodiversity and climate regulation, yet they face threats from environmental changes and disturbances. Monitoring forest health traditionally requires extensive manual labor and time. This study aims to de-velop a predictive framework to assess forest health status: Healthy vs. Unhealthy by analyzing the correlation between tree structural attributes, such as DBH, Height, Crown Width, and soil nutrient profiles, such as Total Nitrogen, Total Phosphorus. The study utilizes a dataset comprising structural and environmental parameters from various forest plots. Exploratory Data Analysis (EDA) is em-ployed to identify key health indicators, and Machine Learning (ML) algorithms, XGBoost, Random Forest (RF), and Support Vector Machines (SVM), are used to perform predictive analysis on the health status of trees. The models are then evaluated using the error metrics: R-squared (R2), Mean Absolute Error (MAE), and Root Mean Squared Error (MSE). Additionally, the impact of biodiversity indices and fire risk on forest stability is evaluated. It is expected that specific soil nutrient thresholds promoting tree health will be identified, leading to the development of a high-accuracy model for predicting potential forest degrada-tion. The findings provide actionable insights for forest managers to implement targeted conservation strategies.