Monitoring Arctic-Boreal Wetland Inundation using Temporal Sentinel-1 SAR Imagery

Abstract:

About half of the world’s wetlands are distributed in northern high latitudes (north of 45°N), with a large proportion within the Arctic-Boreal Region (ABR). Wetlands in this region contain abundant soil carbon due to low rates of decomposition under anoxic conditions, which makes transient zones between their inundated and dry areas a fluctuating land cover type important for methane emissions. These zones may contribute a disproportionately high share of overall methane fluxes, even though their areal coverage may be relatively small. Therefore, it is critical to map and monitor ephemeral inundation accurately, regularly, and frequently, for the purpose of better serving wetland methane emission modelling. Long-wavelength synthetic aperture radar (SAR), with its unique ability to detect inundation beneath vegetation canopies and to penetrate cloud cover, is well suited to wetland studies.

In this study, we propose a two-level decision tree method that incorporates spatial information to map and monitor temporally transient wetland inundation using time series of Sentinel-1 SAR data. Three sites with differing hydrological conditions, namely the Peace-Athabasca Delta (PAD), Yukon Flats and Yellowknife regions, were selected for study. The method takes advantage of the temporal characteristics of backscatter brightness to first generate a maximum possible inundation map, which is then employed as a constraint to derive further inundation maps. The method is implemented on the Google Earth Engine Platform to facilitate utilization of Sentinel-1 time series. Resultant inundation maps were evaluated using field samples collected from 2017-2019, as well as other independent wetland maps. Spatio-temporal dynamics of transient wetland inundation in these sites were analyzed to identify maximum inundation extent, annual inundation pattern, inundation frequency and hydroperiod of transient wetland zones. Limitations and uncertainties of the methodology are discussed, together with implications for Arctic-Boreal hydrological science and associated implications for ABR carbon.