Microtopography shapes soil pH in flysch regions across Switzerland

Abstract

As topography is a key factor controlling soil genesis and strongly influences physical and chemical soil properties, terrain attributes are routinely used in digital soil mapping to spatially predict soil properties. Forests on flysch sediments along the northern slopes of the Swiss Alps often have a strong microrelief. The dominant soil types are Gleysols in depressions and Cambisols on ridges, with large pH variation within short distances. Based on the theory of soil development we expected that soil-forming processes driven by micro-scale topographic variation shape similar micro-scale spatial patterns of soil properties at different sites within the flysch region. Therefore, the main objective of the study was to investigate model extrapolation within flysch regions, which has turned out to be difficult on many other geological substrates. At three sites, each of about 2 ha, we first built three local models to examine whether a relationship between microtopography and topsoil pH could be inferred from high-resolution terrain attributes and pH measurements. Using data from all three sites we then calibrated a joint model and examined model extrapolation by calibrating models with data from two sites and predicting pH at the third. All models were based on multiple linear regressions that used 0.5 m resolution terrain attributes derived by a multiscale approach as explanatory variables. The cross-validated R2 for the local pH models varied between 0.56 and 0.77, and the corresponding RMSE between 0.57 and 0.64 pH units. The R2 and RMSE for the joint model were 0.60 and 0.76, respectively. The results of the local models suggest that microtopography is a dominant soil-forming factor on flysch sediments that triggers soil genesis on a spatial scale from submetre to metres. Although the extrapolated models showed a reduced prediction ability with R2 values of 0.25, 0.46 and 0.53, the selected terrain attributes were relatively similar among the models, which may indicate the common driving processes. The results for the joint model suggest that using high-resolution terrain attributes yields a fairly accurate spatial prediction of the highly variable topsoil pH in forests on flysch sediments across Switzerland.