Predictive modelling of soil organic carbon, carbon stocks, and fractions across elevation gradients in the dry-temperate region of North-West Himalaya.

Soil organic carbon (SOC) is fundamental to climate regulation and ecosystem resilience, yet how elevation-driven environmental variations influence its spatial distribution and predictability remains insufficiently quantified in topographically complex mountain regions. The present study addressed this gap by integrating geospatial interpolation and machine learning algorithms to predict and map the spatial distribution of SOC pools across an elevation gradient in Kinnaur district of Himachal Pradesh. Surface soil samples were collected from 146 georeferenced locations, stratified across four elevation zones, and modelled using 24 environmental covariates representing topography, environment, climate, and soil. SOC content increased consistently with elevation, from 10.60 g kg-1 at lower elevations to 19.72 g kg-1 at higher elevations, while SOC stock exhibited a two-fold increase along the elevation gradient. The study confirmed that the spatial variability of SOC was associated with topography and climatic variables, with elevation primarily representing an indirect surrogate, operating through these covariates. Among the tested models, Random Forest (RF) and Random Forest Regression Kriging (RF-RK) consistently outperformed Ordinary Kriging (OK), while elevation-stratified modelling reduced large-scale environmental heterogeneity and yielded only marginal improvement with RF-RK. High-resolution (30 m) SOC maps identified carbon accumulation hot spots at higher elevations. Model validation and uncertainty analysis further showed that, despite comparable point prediction, RF and RF-RK differed in their uncertainty calibration. Fuzzy clustering delineated two carbon management zones (CMZs), with lower-elevation zones characterized by greater carbon lability, whereas higher-elevation regions dominated by passive carbon pools, providing a spatially explicit framework for high-resolution carbon monitoring and site-specific carbon management in high-altitude regions of the North-West Himalaya (NWH).