Abstract The Hudson Bay Lowlands (HBL) are recognized as the second largest peatland complex in the world. Due to variability in peat thickness across a large and heterogeneous landscape, the existing carbon (C) storage estimates for the HBL may contain large uncertainty. Here, we use geospatial variables that are associated with HBL peat formation, age, accumulation, and occurrence to understand the driving factors for peat depth variability and map peat depth and C storage at 30 m spatial resolution. The estimated average peat depth of HBL is 184(±48) cm with 90% of values falling between 89 and 264 cm. Based on the spatially explicit peat depth, the HBL total C storage is estimated to be 30(±6) Pg. Distance to the coastline is the most important indicator of peat depth where the depth increases with distance further away from Hudson Bay coastline, confirming that the time since peat formation is closely related to peat depth.
Plain Language Summary The Hudson Bay Lowlands (HBL) contain the second largest peatland complex in the world. We used spatial data sourced from satellite observations and geospatial information that are associated with peat occurrence, age, formation, and accumulation to estimate peat depth and carbon storage at 30 × 30 m spatial details for the entire HBL. We combined several machine learning models together in a way that improves their ability to work well on new data with a technique called “stacking,” to improve the accuracy of peat depth estimation. The estimated average peat depth was 184 cm while the entire HBL stores 30 billion tonnes of carbon. The peat depth and carbon storage information presented in this study will help monitor and assess the vulnerability of carbon storage to anticipated changes in climate, resource development, land use, and disturbances that are intensifying in the region. They are also crucial for managing and protecting this vital ecosystem, quantifying the carbon cost of resource development, and for developing ecologically sound land management practices in the region.
Key Points We use stacking, an ensemble learning technique designed to mitigate overfitting, to estimate peat depth in the Hudson Bay Lowlands (HBL) The average peat depth of HBL is 184 (±48) cm with 90% of depths falling within 89–264 cm HBL stores 30 (±6) Pg carbon (C) with average value of 86 (±35) kg m −2