Remote sensing of seasonal variation of LAI and fAPAR in Pakistan under changing climate using machine learning

Climate change represents an intense threat to terrestrial ecosystems, influencing vegetation dynamics that are essential for ecosystem functionality and agricultural yields. Hence, an accurate monitoring of such phenomenon is quite essential for an agricultural country like Pakistan. Our study evaluates the effects of climatic variability on vegetation in Pakistan through the utilization of two critical biophysical indicators of vegetation vitality and productivity, Leaf Area Index (LAI) and the Fraction of Absorbed Photosynthetically Active Radiation (fAPAR). Using remote sensing data from the Copernicus(GLS)SPOT/PROBA-V satellite and employing machine learning method, Random Forest (RF) algorithm, we analyzed the spatial and temporal distributions of LAI and fAPAR across the diverse climatic regions of Pakistan.LAI and fAPAR spatial plots illuminated significant regional discrepancies. Higher values were predominantly recorded in the fertile Indus River basin and northern mountainous areas, suggested a moderate to strong vegetation productivity. On the other hand, the arid regions of Balochistan and southern Sindh expressed low values, which indicates low vegetation and heightened vulnerabilities to desertification, diminished agricultural productivity, and ecological deterioration. Temporal analysis specified clear seasonal variabilities, with peaks during the monsoon interval and downplay throughout the dry months. The RF model demonstrated substantial predictive efficacy, attaining R² equals to 0.92 and 0.91 for LAI and fAPAR, respectively. We believe that these outcomes will provide a robust foundation for policymakers to address climate-induced stresses on vegetation and boost resilience in vulnerable areas of Pakistan