Key points are not available for this paper at this time.
Introduction Pakistan’s heterogeneous climate exposes its vegetation to alternating drought, heat, and flood episodes, yet province-scale responses remain poorly resolved. Methods This study assessed climate-driven vegetation vulnerability across Pakistan from 2001 to 2023 by combining Landsat-based kernel NDVI (kNDVI), the Vegetation Health Index (VHI), Mann–Kendall trend analysis, and interpretable machine learning. Annual vegetation dynamics were evaluated across Balochistan (BLC), Khyber Pakhtunkhwa–Gilgit–Baltistan–AJK (KGA), Punjab (PJB), and Sindh (SND), and related to precipitation, temperature, SPEI, and PDSI. Random Forest with SHAP was used to quantify the relative contribution of climatic predictors to kNDVI. Results The record showed a persistent spatial contrast: BLC and southern SND remained the most moisture- and heat-constrained, with up to 77% and 59% of their areas, respectively, falling into decline classes, whereas PJB retained the most stable vegetation, supported by irrigation and stronger monsoon recharge. kNDVI captured cumulative stresses in BLC and KGA, where negative trends persisted despite wetter years, while VHI registered rapid post-flood greening after the 2010 and 2022 events, particularly in SND, reflecting short-term improvement in vegetation condition rather than sustained recovery. SHAP analysis identified moisture-related variables (precipitation, PDSI, and SPEI) as the dominant positive controls in KGA and PJB, whereas high temperature exerted the strongest negative influence in BLC and SND. Discussion The dual-index framework distinguishes areas where vegetation improvement is consistent with better hydroclimatic conditions from areas where decline persists despite climatic relief, providing a basis for drought-resilient afforestation, rangeland rehabilitation, and climate-informed land-use planning in Pakistan.
Mehmood et al. (Tue,) studied this question.