SWIR-Based Moisture Index to Detect the Temporal Fluctuation of Water Table Depth in Northern Peatlands

Abstract:

Northern peatlands are a huge terrestrial carbon stock and ongoing sink. Shallow water table depth (WTD) and high soil moisture conditions are among the main factors that enable the negative net radiative forcing effect of peatlands. The OPtical TRApezoid Model (OPTRAM) is an approach for the surface and root zone soil moisture estimation from satellite-based SWIR and NDVI data. Through the strong capillary connection between soil moisture and WTD under shallow water table conditions, we hypothesize that OPTRAM also indicates WTD dynamics in peatlands. In this study, we provide evidence of the usefulness of OPTRAM for monitoring WTD dynamics in northern peatlands. We propose a method to identify OPTRAM pixels that are the most representative for monitoring WTD (in the following referred as ‘best pixels’), which is based either on WTD that is measured in-situ or simulated by a land surface model. WTD simulations were taken from the PEATland-specific adaptation of the NASA Catchment Land Surface Model (PEATCLSM). With the assumption that temporal fluctuations in WTD are very coherent within a peatland, the ‘best pixels’ were identified by the highest temporal Pearson correlation between time series of OPTRAM and WTD (either in situ or simulated with PEATCLSM). We calculated OPTRAM from MODIS, Landsat and upscaled Landsat at MODIS resolution images for five northern peatlands (three bogs, two fens) with long-term in-situ WTD measurements. The performance of OPTRAM for WTD monitoring varied between the remote sensing data; the highest temporal Pearson correlation (mean of 0.7 across the ‘best pixels’ in five peatlands) was obtained for OPTRAM with Landsat data at 30 m resolution. Further, we found that for the 30 m ‘best pixels’ the vegetation cover was presented by mosses and grasses with little or no bushes or trees. Overall, correlation coefficients were spatially highly variable and decreased in regions with higher shrub or tree abundance. Our study demonstrates that ‘best’ OPTRAM pixels can be localized by the use of in-situ (if such records are available) or, as an alternative, simulated WTD. The opportunity to use simulated WTD from a global land surface model to localize ‘best pixels’ suggests the application of OPTRAM at the global scale for the detection of WTD changes in northern peatlands.