Blue Carbon Mapping Using Temporally Optimized Satellite Remote Sensing Imagery: A Regional Study of Northeast US Salt Marshes

Abstract Coastal wetlands store three to five times more carbon per unit area than tropical rainforests in continually accreting peat soils, collectively referred to as “Blue Carbon.” However, variability in soil carbon density within and between sites leads to large uncertainty when estimating carbon stocks and sequestration rates. Salt marsh carbon sequestration is mainly driven by nonlinear ecogeomorphic feedback between tidal inundation, bioproductivity, and sediment supply—all of which can be observed by satellites. In this study, we used soil bulk density and soil organic content from 410 soil samples collected across 15 sites in the Northeast US to relate soil properties to remotely sensed spectral observations. We tested model fits using Landsat 5, 7, 8, and Sentinel 2 images from 1984 to 2022 to determine the optimal season and tidal conditions for relating remote sensing indices to soil properties. We explored the roles of sediment supply and tidal range in regional prediction models. The study found that (a) spatial patterns of remote sensing indices correlate well with soil properties; (b) at the marsh scale, remote sensing indices capture the spatial variability of soil properties with image acquired at high tide and vegetation phenology specific to geomorphic setting; (c) at the regional scale, tidal range improves the prediction model in barrier marshes, while sediment supply improves the prediction model in fluvial marshes. The considerable spatial variation of SOC within marshes and across regional gradients highlights the need for high resolution maps of salt marsh soil properties.