Tidal marshes serve as important “blue carbon” ecosystems that sequester large amounts of carbon with limited area. While much attention has been paid to the spatial variability of sedimentation within salt marshes, less work has been done to characterize spatial variability in marsh soil carbon density. Soil properties in marshes vary spatially with several parameters, including marsh platform elevation, which controls inundation depth, and proximity to the marsh edge and tidal creek network, which control variability in relative sediment supply. We used lidar to extract these morphometric parameters from tidal marshes to map soil organic carbon at the meter scale. Fixed volume soil samples were collected in 2021 at four northeast U.S. tidal marshes with distinctive morphologies to aid in building predictive models. Tidal creeks were delineated from 1-m resolution topobathy lidar data using a semi-automated workflow in GIS. Log-linear multivariate regression models were developed to predict soil organic content, bulk density, and carbon density as a function of predictive metrics at each site and across sites. Results show that modeling salt marsh soil characteristics with morphometric inputs works best in marshes with single connected creek network morphologies. Distance from tidal creeks was the most significant model predictor. Addition of distance to the inlet and tidal range as regional metrics significantly improves cross-site modeling. Our mechanistic approach results in predicted total marsh carbon stocks comparable to previous studies but captures important meter level variation. Further, we provide motivation to continue rigorous mapping of soil carbon at fine spatial resolutions.