Machine learning models for predicting vegetation conditions in Mahanadi River basin

Area of Science:

• Environmental Science
• Remote Sensing
• Climate Change Studies

Background:

• River basin vegetation is influenced by climate factors like precipitation and land surface temperature (LST).
• Predicting vegetation health, measured by the Normalized Difference Vegetation Index (NDVI), is crucial for environmental management.

Purpose of the Study:

• To identify the optimal machine learning model for predicting NDVI using LST and precipitation.
• To determine the correlation between NDVI, LST, and precipitation in the Mahanadi basin.

Main Methods:

• Utilized monthly precipitation data from CHRS and MODIS products for LST and NDVI via Google Earth Engine (GEE).
• Evaluated four machine learning models: Linear Regression (LR), Random Forest (RF), Support Vector Regression (SVR), and K-Nearest Neighbors (KNN).
• Assessed model performance using R², RMSE, MSE, MAE, and Explained Variance Score (EVS).

Main Results:

• The Random Forest (RF) model demonstrated the highest R² values for both training and testing datasets.
• The K-Nearest Neighbors (KNN) model achieved the lowest Root Mean Square Error (RMSE) in the testing set.
• Established a positive correlation between precipitation and NDVI, and negative correlations between precipitation and LST, and NDVI and LST.

Conclusions:

• The Random Forest model is the most effective for NDVI prediction in the Mahanadi basin.
• Findings offer insights into climate-vegetation dynamics and aid in river basin management and climate change impact assessment.