Four-decade hydro-meteorological variability and machine-learning based weather projections (1983-2050) for Nalphu village, Jajarkot, Nepal using XGboost

This study provides the first comprehensive, multi-parameter hydro-meteorological assessment and machine-learning based climate projection for Chhedagad Municipality Ward no. 10, Nalphu Village, a data-scarce mid-mountain region of Jajarkot, Nepal. Using four decades (1983–2024) of NASA POWER reanalysis data, we analyzed long-term variability in temperature, precipitation, cloud amount, humidity, windspeed, surface pressure, and solar radiation to characterize historical climate behavior and emerging shifts. Results reveal strong monsoon-driven seasonality across all parameters, a statistically significant cooling tendency in temperature after the early 2000s (Sen’s slope = − 0.052°C/year), and a major hydrological regime shift marked by a transition from dry to wetter conditions in 2004. Precipitation increased significantly (Sen’s slope = + 15.5 mm/year), driven primarily by enhanced monsoon rainfall. Cloud amount exhibited pronounced seasonal and interannual fluctuations and exerted clear climatic influence, strengthening rainfall during monsoon months (r = 0.60–0.72) and suppressing temperature through cloud-induced cooling. Humidity and specific humidity showed strong long-term increases, consistent with a wetter, moisture-laden atmosphere, while solar radiation demonstrated a significant decline after 2000. Windspeed displayed a multi-decadal weakening trend, and surface pressure remained exceptionally stable. XGBoost-based forecasts (2025–2050) project a continuation of current climatic patterns, including moderate warming, stable-to-wet precipitation cycles, steady cloud amounts, and no major shifts in radiation or pressure regimes. Overall, the study offers a rigorous, multi-method climate diagnosis and future outlook for one of Nepal’s most climate-vulnerable highland communities, providing essential scientific evidence for agricultural planning, water-resource management, solar power installation and local climate adaptation.