Eberswalde, 1 March: Can peat soils within forested areas be reliably detected using open geospatial and remote-sensing data?
IT4Forest enthusiast Evelyn Wallor in cooperation with Hainner Aparicio, Nicole Voss, Jan-Peter Mund and Winfried Riek answered this question in her recent article published on pre-print.org

This article covers a study that evaluates the detectability of organic soils within forests by combining open geospatial and remote-sensing data with mapped soil and water level information in two Random Forest (RF) approaches. Whether enclosed by or overlaid with forest, the organic soils in the study area show high soil water content, reaching saturation during the hydrological winter. Accordingly, terrain indices derived from Digital Elevation Models (DEMs) and soil moisture estimates from L-band ALOS PALSAR signals are used as predictors in the RF algorithms. CORINE Land Cover data are used to evaluate how different forest cover types (FCT) affect the RF models. Strong agreement in the target classification is achieved when FCT is included and when the higher-resolution DEM is used. The Boolean approach is less influenced by variations in predictor composition but is more sensitive to imbalance in the reference data, which is reflected in comparisons of the “event error” and “no event error.” In all RF models, the soil moisture pixel values retrieved from L-band ALOS PALSAR rank highest in variable importance when FCT is excluded.

The method explored in this study is relevant and shows potential in the still exploratory research path to detect the presence of organic wet soils in temperate forests based on L-Band SAR σo. Further assessments, as the ones previously mentioned, could improve this method and offer appropriate benefits. The detection of wet organic soils in temperate forests with global coverage could be of potential exploration and development, helping to reduce resource-intensive campaigns to know the exact spatial location of these valuable ecosystems

To access the artilce visit: Preprints.org (www.preprints.org) | NOT PEER-REVIEWED |

doi:10.20944/preprints202602.1553.v1
© 2025 by the author(s). Distributed under a Creative Commons CC BY license.

could be of potential exploration and development, helping to reduce resource-intensive campaigns
to know the exact spatial location of these valuable ecosystems