Determining Spatially Variable Peat Depths of an Alpine Peatland in the Qinghai‐Tibet Plateau

Abstract Understanding peat depths is crucial for assessing carbon dynamics within peatlands. However, accurately estimating peat depths across large areas remains challenging. In this study, we developed a novel approach integrating a machine‐learning based regression modeling with an optimized field sampling strategy to predict spatially variable peat depths across the Black River (BR) watershed with an area exceeding 4,600 km 2 within the Zoige basin of the Qinghai‐Tibet Plateau. Using the key controlling factors (elevation, slope, soil organic carbon, and soil bulk density), we created a high resolution (13 m) digital map depicting the spatial distribution of peat depths throughout the BR watershed. We validated the accuracy of predicted peat depths through comparison with field measurements and revealed that slope is the dominant control in determining peat depths. The resulting map highlights that deep peat deposits typically cluster in upslope sections and areas of higher elevations. The total peat carbon stock within the study area amounts to approximately 4.195 Pg C, following a spatial pattern similar to peat depths but with greater concentration in three specific counties. These high‐resolution peat depth and peat carbon maps provide valuable insights for effective peatland management in these vulnerable alpine ecosystems.