Soil erosion assessment in alpine grasslands using fallout radionuclides: critical points, solutions and applications

Soil erosion processes are one of the main threats to the Alps. They affect slope stability, water budgets, vegetation productivity and the overall biodiversity of the alpine ecosystem. In particular, recent land use and climate changes exacerbated the impact that sheet erosion, a dominant but scarcely visible process, has on alpine grasslands. Yet, the quantitative estimation of the effects of sheet erosion is constrained by the topographic and climatic conditions of the Alps, which hinder the application of conventional assessment techniques. Recently, the use of fallout radionuclides (FRN) as soil erosion tracers showed very promising results in deriving integrated estimates of soil degradation processes affecting alpine soils, over a range of different time scales. Nonetheless, for a correct application of the FRN method, special attention should be paid to three main critical points that are extensively discussed in this thesis, namely: (i) the selection of suitable reference sites; (ii) the selection of the approach (i.e. the traditional approach, the resampling approach, or the repeated sampling approach); and (iii) the selection of the appropriate conversion model.
First, we investigated the suitability of undisturbed reference sites in an alpine valley (Urseren Valley, Canton Uri, Switzerland) for the application of 137Cs, the most commonly used FRN for soil erosion studies. In alpine regions, which are heavily affected by the heterogeneous Chernobyl 137Cs fallout and by high geomorphic and anthropogenic activity, the choice of reference sites is a great challenge. To avoid the uncertainties associated with a wrong selection of reference sites, we have developed and proposed the decision support tool CheSS, which allows Checking the Suitability of reference Sites using a repeated sampling strategy and a decision tree. Comparing the 137Cs inventories of reference sites, which have been sampled in two different periods, enables identifying the sites where no soil disturbance processes have occurred and that can be further used as a stable and reliable basis for the application of the method. Chess also directs particular attention to the analysis of the spatial variability of the 137Cs distribution at the sites. The results of the Chess application to our study area imply that no suitable reference sites could be found.
As a further step, we have tested the application of a 137Cs repeated sampling approach in the Piora Valley (Canton Ticino, Switzerland), where previous studies have failed to identify undisturbed and homogeneous reference sites. The repeated sampling approach facilitates the derivation of short-term soil redistribution rates by comparing the FRN inventories measured at sampling sites in different times, thus without the need of reference sites. Twelve points located along four transects have been sampled in 2010 and in 2014, and their 137Cs inventory has been compared. The results indicate high soil degradation dynamics, which correspond to a range of yearly soil redistribution rates of 3-36 t ha-1.
At both study areas, the high difficulties associated with the use of 137Cs as tracers (i.e. the extremely high small-scale variability of 137Cs distribution) led us to examine the applicability of 239+240Pu (also an artificial FRN), whose presence in the Alps is not connected to the Chernobyl fallout, but mainly to the atmospheric nuclear weapon tests. As a result, its distribution is much more homogeneous compared to 137Cs. 239+240Pu is also preferable to 137Cs, because it has a longer half-life and its measurements are more cost- and time-effective. However, the conversion of 239+240Pu inventories into soil redistribution rates has been impeded by the fact that the available models are not able to describe the specific behavior of Pu isotopes in the soil, as they are mainly designed for 137Cs.
Consequently, our energy has been directed towards developing a new conversion model, called MODERN (Modelling Deposition and Erosion rates with fallout RadioNuclides). MODERN is an innovative model based on a single formula that derives both soil erosion and deposition rates. MODERN accurately depicts the soil profile shape of any selected FRN at reference sites and allows the adaptation of the depth profile to simulate the behavior of the FRN under different agro-environmental conditions. A first application of MODERN has been performed on a 239+240Pu dataset collected in the Urseren valley. Thanks to its characteristics and its adaptability, MODERN describes the specific depth distribution of Pu isotopes in the soil better than other models. The MODERN code has been developed in Matlab™ and is publically released on the website of our research group. In order to expand its accessibility, the new package modeRn has been recently developed using the free and open-source system R. modeRn also includes new features that enhance its potential and usability.
This thesis offers a detailed overview of the difficulties associated with the application of FRN in alpine areas. It also presents new, effective, and useful tools that help reduce the sources of uncertainty of the FRN method (CheSS) and promote its application to derive soil redistribution rates at different land use conditions (MODERN). Future studies should focus on using precise and accurate FRN-based estimates to validate large-scale modelling techniques, in order to improve the monitoring and identification of soil erosion risk areas in alpine regions.