Predicting soil moisture conditions across a heterogeneous boreal catchment using terrain indices

Abstract. Soil moisture has important implications for drought and flooding forecasting, forest fire prediction water supply management. However, mapping soil remained a scientific challenge due to canopy cover small-scale variations in conditions. When accurately scaled, terrain indices constitute good candidate modelling the spatial variation of conditions many landscapes. In this study, we evaluated seven different at varying digital elevation model (DEM) resolutions user-defined thresholds as well two available maps, using an extensive field dataset (398 plots) registered five classes from survey covering (68 km2) boreal landscape. We found that could be explained by indices, best predictors within studied landscape were depth index (DTW) machine-learning-generated map. Furthermore, study showed large difference between effects changing DEM resolution thresholds, which severely affected performance predictions. For example, commonly used topographic wetness (TWI) performed on 16 m, while TWI calculated higher than 4 m gave inaccurate results. contrast, above stream (EAS) more stable 1–2 resolution. None highest 0.5 m. addition, highlights challenges caused heterogeneous types area shows need local knowledge when interpreting modelled The results clearly demonstrate represent conditions, validated, selecting unsuitable or threshold can give ambiguous even incorrect