Soil erosion modelling as a tool for future land management and conservation planning

Abstract

Maintaining future agricultural productivity and ensuring soil security is of global concern and requires evidence-based management practices. Moreover, understanding where and when land is at risk of erosion is a fundamental step to combatting future soil loss and reach Land Degradation Neutrality (LDN). However, this is a difficult task because of the high spatial and temporal variability of the controlling factors involved. Therefore, tools investigating the impact and frequency of extreme erosive events are crucial for land managers and policymakers to apply corrective measures for better erosion management in the future. While the utility of using wind and water erosion models for management is well established, there is a paucity of work on the impact of climate change and extreme environmental conditions (e.g. wildfires) on soil erosion by wind and water simultaneously. Both erosion types are controlled by different environmental variable that vary highly in space and time. Therefore, the overarching aim of this study was to develop a joint wind-water erosion modelling method and demonstrate the utility of this approach to identify (1) the spatio-temporal variability of extreme erosion events in the South Australian agricultural zone (Australia) and (2) assess the likely increase of this variability in the face of climate change and the recurrence of wildfires. To fulfil the aim of the research project, we adapted two state-of-the-art wind and water (hillslope) erosion models to integrate modern high-resolution datasets for spatial and temporal analysis of erosion. The adaptation of these models to local conditions and the use of high-resolution datasets was essential to ensure reliable erosion assessment. First, we applied these models separately in the Eyre Peninsula and Mid-North agricultural regions. We evaluated the spatio-temporal variability of extreme erosion events between 2001 and 2017 and described the complex interactions between each erosional process and their influencing factors (e.g. soil types, climate conditions, and vegetation cover). Hillslope erosion was very low for most of the Eyre Peninsula; however, a large proportion of the central Mid-North region frequently recorded severe erosion (> 0.022 t ha-1) two to three months per year, for most of the years in the time-series. The most severe erosion events were primarily driven by topography, low ground cover (< 50%) and extreme rainfall erosivity (> 500 MJ mm ha-1 h-1). Average annual wind erosion was very low and comparable in the two regions. Nonetheless, most of the west coast of the Eyre Peninsula frequently registered severe erosion (> 0.000945 t ha-1 or 0.945 kg ha-1) two to three months per year, for most of the years. The most severe erosion events were largely driven by the soil type (sandy soils), recurring low ground cover (< 50 %) and extreme wind gusts (> 68 km h-1). We identified that erosion severity was low for the vast majority of the study area, while 4% and 9% of the total area suffered severe erosion by water and wind respectively, demonstrating an extreme spatial and temporal skewness of soil erosion processes. Then we combined the modelling outputs from the wind and water erosion models and tested the models’ response to major wildfire events. This research demonstrated how erosion modelling could be used to predict the impact of severe wildfire events on soil erosion. The two models satisfactorily captured the spatial and temporal variability of post-fire erosion. However, a very small fraction of the region (0.7%) was severely impacted by both wind and water erosion. We observed that soil erosion increased immediately after the wildfires or within the first six months for the ten fire-affected regions. For three of the wildfire events, the models showed an increase in wind and water erosion in consecutive months or at the same time. These results highlighted the importance to consider wind and water erosion simultaneously for post-fire erosion assessment in dryland agricultural regions. Finally, we had the rare opportunity to assess the impact of a catastrophic wildfire event on wind erosion in an agricultural landscape by examining the influence of unburnt stubble patches on adjacent burnt or bare plots using a spatio-temporal sampling design. The field study allowed a quantitative assessment of spatial and temporal patterns of wind erosion and sediment transport after a catastrophic wildfire event. It showed very high levels of spatial variability of erosion processes between burnt and bare patches and demonstrated how measuring field-scale sediment transport could complement fine-scale experimental studies to assess environmental processes at the field scale. This research highlights the utility of erosion models to inform corrective measures for future land management. We have implemented tools that allow a realistic assessment of the influence of climate change and extreme environmental conditions scenarios on soil erosion for a wide range of land cover over large regions. Here, the models enabled the identification of the relative post-fire wind or water erosion risk in dryland agricultural landscapes, making them particularly useful for land management under future uncertainty. Spatial patterns compared well with previous modelling approaches and underpinned the benefit of erosion models to assess spatial differences in erosion risk and evaluate corrective measures at the regional scale. However, modelled soil erosion magnitudes strongly depend on how the influence of soils is implemented in the models, making it difficult to set absolute quantitative soil loss targets for land management. The thesis has provided a proof of concept of the approach for South Australia. However, all input data can be freely sourced Australia-wide and similar dataset are available globally.