Publication Details |
| Category | Text Publication |
| Reference Category | Journals |
| DOI | 10.1007/s44378-025-00077-w |
Licence  |
|
| Title (Primary) | Integrated framework for assessment and spatial prediction of humus layer properties of forest soils |
| Author | Thomas, F.; Becker, C.; Petzold, R.; Schmidt, K.; Scholten, T.; Werban, U.
|
| Source Titel | Discover Soil |
| Year | 2025 |
| Department | MET |
| Volume | 2 |
| Page From | art. 49 |
| Language | englisch |
| Topic | T5 Future Landscapes |
| Data and Software links | https://doi.org/10.23728/b2share.bc8730f4b4724169baee96a21c1e894e |
| Keywords | Digital soil mapping; Machine learning; Predictive modelling; Forest soils; Humus layer properties; Vis-NIR spectroscopy |
| Abstract | Data about the availability of nutrients in the humus layer of forest soils is vital information for sustainable forest management. For well-informed decision making, changes in humus layer conditions have to be monitored. The currently used schemes based on indicator plants are more and more overprinted by element input and forest management measures. Large and repeated field campaigns and laboratory analysis can provide the required information, but are expensive and time consuming. Techniques of Digital Soil Mapping (DSM) offer an alternative to gain the necessary information by spatially predicting selected soil properties, but their usage for mapping humus layer properties of forest soils is rare. Further, vis-NIR spectroscopy is a non-destructive approach that can provide physical and chemical soil properties via regression models. In this study, we present the application of an integrated framework as measuring system using DSM with the aim of providing spatial information about properties of the humus layer of forest soils. In order to reduce the amount of laboratory work, we applied the use of vis-NIR spectroscopy to determine the humus layer properties values within the framework. As input variables, we used available information based on soil forming factors and already existing information on mineral soils (soil form class, climate data, satellite data as proxy for vegetation, relief data and parent material and geographic location). Therefore, generally available data like satellite imagery and climate data can be used in combination with information that is already available in forest site mapping, e.g. on parent material, soil texture and pedogenetic soil type as well as hydromorphic features. Conditioned Latin Hypercube sampling was used to determine the locations of the field sampling points to collect the soil material, ensuring a valid representation of the humus layer properties in the test area. Probed layers were the Oh horizon and 0–5 cm depth. We tested the developed framework in a case study on a forest site in Saxony, investigating C/N ratio, pH value, cation exchange capacity and base saturation. Random Forest model calibration for spatial prediction achieved R2 > 0.9 for all investigated humus layer properties. Using the developed framework, we were able to create high resolution maps of humus layer properties on forest soils in the case study area. Especially for C/N ratio and pH value, the derived maps showed high spatial variation within the study area. For our test site, the framework revealed depleted humus conditions, which should be addressed by forest management measures. Vis-NIR predictions of humus layer properties were calculated using partial least squares regression. In Oh horizons, model results achieved R2 values between 0.17 and 0.69. In 0–5 cm, R2 values ranged from 0.43–0.62. RMSD values between produced maps based on chemical values and vis-NIR predictions were 0.7 and 0.96 for C/N, 0.06 for pH, 6.8 μeq/g for CEC and 3.25% and 3.3% for BS. We conclude that the framework produced maps that can be used to assess humus conditions and thus support decision making in forest management. The use of vis-NIR spectroscopy offers the possibility to reduce the amount of laboratory work, but there is a trade-off in the accuracy of the results. In general, the framework can be used to fill the data gap by providing spatial information on humus layer properties at the forest stand scale. |
| Persistent UFZ Identifier | https://www.ufz.de/index.php?en=20939&ufzPublicationIdentifier=30952 |
| Thomas, F., Becker, C., Petzold, R., Schmidt, K., Scholten, T., Werban, U. (2025): Integrated framework for assessment and spatial prediction of humus layer properties of forest soils Discover Soil 2 , art. 49 10.1007/s44378-025-00077-w |
|