Environmental variables improve remote sensing-based water table monitoring in peatlands

Water table (WT) is a key indicator of peatland ecosystem functioning, but its spatiotemporal monitoring is challenging. Optical remote sensing has been used in peatland WT monitoring with varying success, but few studies have tested whether environmental variables—particularly topographic and tree stand structure variables derived from LiDAR—improve modelling performance. We tested whether environmental variables improve (1) uncrewed aerial vehicle-derived spatial WT models in two northern boreal, partly drained aapa mires and (2) satellite image-derived spatiotemporal WT models in a southern boreal drained peatland forest in Finland. We employed random forest regression and variable selection techniques to model WT, using optical remote sensing and environmental variables as predictors. Our results showed that environmental variables related to topography and tree stand structure improve modelling performance, with R2 increasing by 0.01–0.19 compared to optical-only models. Our findings support the integration of optical and environmental data for spatial and spatiotemporal WT monitoring in boreal peatlands.