The Sloping Mire Soil-Landscape of Southern Ecuador: Influence of Predictor Resolution and Model Tuning on Random Forest Predictions

Abstract

The sloping mire landscape of the investigation area, in the southern Andes of Ecuador, is dominated by stagnic soils with thick organic layers. The recursive partitioning algorithm Random Forest was used to predict the spatial water stagnation pattern and the thickness of the organic layer from terrain attributes. Terrain smoothing from 10 to 30 m raster resolution was applied in order to obtain the best possible model. For the same purpose, several model tuning parameters were tested and a prepredictor selection with the R-package Boruta was applied. Model versions were evaluated and compared by 100 repetitions of the calculation of the residual mean square error of a five-fold cross-validation. Position specific density functions of the predicted soil parameters were then used to display prediction uncertainty. Prepredictor selection and tuning of the Random Forest algorithm in some cases resulted in an improved model performance. We therefore recommend testing prepredictor selection and tuning to make sure that the best possible model is chosen. This needs particular emphasis in complex tropical mountain soil-landscapes which provide a real challenge to any soil mapping approach but where Random Forest has proven to be successful due to the testing of model tuning and prepredictor selection.

Improving the Spatial Prediction of Soil Organic Carbon Stocks in a Complex Tropical Mountain Landscape by Methodological Specifications in Machine Learning Approaches

“…Litherland et al [ 42 ] describe the underlying bedrock as part of the Chiguinda Unit of the litho-tectonic section Loja Terrane, consisting of pelite, schist, phyllite, meta-siltstone, sandstone and quartzite. The area is known for its immense organic layers leading to its description as sloping mire soil-landscape [ 10 ]. Its formation is partly related to the dominance of soil hydromorphic properties in the topsoil caused by the silty soil texture, heavy rainfall and shallow slope parallel subsurface flow [ 43 , 44 ].…”

“…Finally, spatial estimates of SOC are increasingly important to acknowledge the soils’ carbon storage potential in the context of climate change. However, it is particluarly the tropical mountain areas with their thick organic layers which are highly complex and difficult to access [ 10 ]. SOC stock data of tropical mountain forest soils are scarce, SOC stock data of the organic layer hardly exist.…”

“…According to Hastie et al [ 24 ] random forests do remarkably well with little tuning. In our experience [ 10 , 39 ] tuning random forest and boosted regression tree models can improve prediction results, but tuning needs to be tested carefully for it can also cause overfitting.…”

Associative Regressive Decision Rule Mining for Predicting Customer Satisfactory Patterns

“…In [20], random forest predictions were made using random forest algorithm to display prediction uncertainty. However, the true positive rate was not addressed.…”

Comparison of Three Supervised Learning Methods for Digital Soil Mapping: Application to a Complex Terrain in the Ecuadorian Andes

“…Parent material is highly weathered and comprises metasiltstones, siltstones, and quartzites, intermixed with layers of phyllite and clay schists [24]. The area is rich in sloping mires [17], thus leading to a prevalence of soils with stagnic properties and thick organic horizons. These were described as, for example, Humaquepts [25] (according to the USDA Soil Taxonomy [26]) and as Histosols and Stagnosols, associated with Umbrisols, Cambisols, Leptosols, and Regosols [27] (according to the World Reference Base for Soil Resources [28]).…”