Automated Mapping of Arctic Floodplains to Improve Estimates of Sediment and Carbon Fluxes

Abstract:

Arctic floodplains are often underlain by permafrost deposits with high amounts of soil organic carbon (SOC) and nutrients. The spatial extent of these floodplains affects both carbon emissions to the atmosphere in response to permafrost thaw and fluxes of carbon and nutrients into the aquatic system via riverbank erosion. Despite their importance in regulating permafrost carbon and nutrients dynamics, the spatial extent of Arctic floodplains remains largely unquantified. In this effort, we developed an automated floodplain mapping algorithm based on gridded topographic data. The algorithm uses the slope threshold of a hypsometric curve within a circular kernel centered on channel points to determine potential flooding elevations above the nearest river (i.e., HAND elevation). We used a calibration procedure to identify the combination of parameters that optimize the agreement between the algorithm’s mappings and hand-mapped floodplains across the Arctic. We utilized these optimal values to map the extent and uncertainty of floodplains along major Arctic rivers. Our results demonstrate that the proposed algorithm delineates floodplains with a relatively high accuracy and is thus useful for quantifying floodplain-related sediment and carbon fluxes across the Arctic permafrost regions.