We all agree that numbers and models are an important service of science for a various set of reasons. For instance, they help decision-makers in our society understand current and future trends and therefore contribute to a differentiated planning of development regarding numerous issues in the environmental, societal and economic sphere. One of the most pressing environmental issues, that was recognized by the United Nation’s food and agriculture organisation (FAO), is “soil degradation”. This process is a global phenomenon and is caused by non-sustainable land management and land use as well as natural processes, such as droughts. Soil is a «non-renewable natural resource»: The time it takes to develop new soil exceeds the time of many human generations and land degradation obviously contributes to an increase in development time. Thus, the natural base for sustaining human and other life decreases faster than it can be recovered therefore threatening important ecosystem functions of soils such as providing food and fuel as well as other ecological services. According to the FAO, the «current state of soil degradation threatens the capacity of future generation[s] to meet their needs». But how can we know about the «current» state of soil degradation? This article aims to address this question by looking at how global soil degradation can be studied, show uncertainties in the resulting models and briefly address what kind of information on different spatial scales the models can actually provide.
In 2003 for example, a group of researchers around Dawen Yang from the University of Tokyo modelled the global soil erosion potential with the so-called RUSLE approach (Revised universal soil loss equation). Soil erosion is a process where soil gets allocated from its initial location, most often leaving the place without fertile topsoil. Drivers of this process are mostly water and wind that hit unprotected soils (with little to now vegetation cover) under certain natural- or land management conditions.
An approach to modelling the global soil erosion potential
The RUSLE-approach as used in the study of Dawen Yang and others can be summarized as follows: it is assumed that the soil erosion potential (A) has different contributing factors. The rainfall erosivity (R-factor) describes to what degree soil is eroded by being directly hit with rain drops. The topography (LS-factor) describes the influence of length and steepness of slopes in an area on the potential soil erosion as soil is assumed to be more vulnerable to erosion with increasing length and steepness of a slope. Additionally, the soil erodibility (K-factor) is determined by looking at the soil properties (for example soil texture and water permeability). More focused on the human impact on soil erosion are the land cover and management (C-factor) and the soil conservation practices (P-factor). The importance of land cover and management relates to the linkage between vegetation cover and soil erosion. For instance, soils under intensive agriculture have a higher soil erosion potential than soils under grassland or forest. Land conversion processes, as for example deforestation with the aim to transform an area into arable land, are known to increase the soil erosion potential. However, there are also soil conservation practices of land users that can decrease soil erosion (for example agro-forestry). Altogether, the soil erosion potential can be expressed with the following simple equation:
The “sense” of modelled soil erosion
The research group around Dawen Yang used this RUSLE-approach and calculated the soil erosion potential on a 55 km2 grid globally. They estimate the accumulated annual soil erosion potential to be 133 billion tons, or an annual average of 10.2 tons per hectare globally. However, the soil erosion potential is not equally distributed over the globe. They identified two zones where soil erosion is most pronounced. Zone one is located on the west coast from North- to South America. The second zone goes from South Europe via the Middle East to Southeast Asia. The authors outline that the modelled soil erosion is most pronounced for mountainous areas, intensive croplands and highly populated regions therefore indicating natural and anthropogenic drivers. However, it is questionable if mountainous areas in high altitudes even feature developed soils, leave let alone the question about if whether they are used. Whereas it is obvious that the results of the presented study show tremendously high values of soil erosion, they also have to be looked at in the context of other current research. A study from Gerard Govers and others from Belgium in 2014 compared different studies from 1991 to 2007 that, like the study by the researchers from Tokyo University, estimated global soil erosion rates. They showed that the modelled global soil erosion rates differ greatly in a range between 50 billion up to 172 billion tons per year. This sheds a different light on the Yang’s study using the RUSLE-approach in 2003. It has severe consequences on the planning of international efforts to combat soil degradation if soil erosion rate estimates are at 50 billion tones per year or versus a rate three times higher are assumed as a baseline quantity of the environmental problem. This leaves us with the question of if and how this wide range of estimations can be explained.
The “non-sense” of modelled soil erosion
As outlined, estimates of global soil erosion differ widely. A reason for that can be located for example in the RUSLE-approach itself. In order to estimate a factor globally, researchers often have to extrapolate data from one or more places over the globe because there is no global data set available. In other words, it is assumed that a relationship between, say, rainfall amount/intensity and the rainfall erosivity (R-factor) that is measured in North America, holds true for every region of the world. Obviously, this assumption is problematic since rainfall amounts and intensities differ greatly between places on the global scale. Additionally, many of the RUSLE-factors are calculated based on global datasets that are incomplete or widely averaged. For example, a data set about soil conservation practices will never contain every single location where those practices were implemented by the land users, let alone that the practices may differ a lot. In other words, some RUSLE-factors are estimated based on global estimates of sub-factors which therefore show a very distorted picture of reality. Another issue is the selection of the spatial resolution of a model: researchers must deal with the trade-off between accurate information and global coverage. With a lower spatial resolution (e.g. 100 km2 grid cells), different functional landscapes are averaged to one potential soil erosion value. This can be misleading because for instance mountainous areas react differently to soil erosion triggers than agricultural areas, yet they are averaged in the model. With higher resolutions (e.g. 10 km2 grid cells) the researchers can use more precise datasets of the regions (where available!) with specific calculations. However, this leads to a decrease of comparability between locations and thus pronounces difficulties in interpreting global models. Using a global modelling approach to answer a research question thus has intrinsic shortcomings. The outcome of the model is determined by how a variable is calculated and estimated, what reference data was used, etc. Also, different modelling-approaches use different variables or calculations. This partly explains the large differences of global soil erosion rates modelled by different studies which was summarized by Govers and others in 2014. Their study also outlined that most soil erosion estimates don’t investigate what happens with eroded soil. Soil can get transported multiple times in a series of erosion events before being deposited at the «final» location. So, most models estimate the loss of soil but not where it ends up. In terms of land management this appears problematic since for instance terraces on hillslopes can be used to catch the eroded topsoil and re-use it at the deposition site. This shows that there are impacts of land management practices which current global soil erosion models cannot claim to cover.
Are global models enough to support combating soil erosion?
Using models offers a way to estimate the global soil erosion potential which can be identified as a global environmental issue with multiple drivers. Thus, this approach provides information about spatial patterns of severity of soil erosion. Hence, models are surely important tools to concentrate international efforts to combat soil degradation in a meaningful way. However, models cannot capture the whole reality of current soil erosion processes and extents and that the numerical results heavily depend on the dataset, calculation methods and spatial resolution that is selected for the model. Additionally, one has to ask how the people actually managing the world’s soils (foremost farmers and pastoralists) benefit from information about global erosion. Small-scale land management surely requires more detailed information about processes and leading to – or conservation practices preventing – soil erosion, other than can be provided by a 55 km2 resolution model. Also, the question about practices that are meaningful to combat soil erosion in individual socio-ecological contexts cannot be addressed by such methods of research, thus leaving local (and relevant!) actors of land management out of the focus. In conclusion, models provide soil erosion estimates on a large spatial scale, but they include uncertainties and cannot provide context-driven information on soil conservation practices that is needed by local actors to combat soil erosion.
- General information about soil and soil degradation are provided by the FAO.
- The study from Dawen Yang and others (2003) was published in Hydrological Processes.
- The findings of Gerard Govers and others (2014) were published in Procedia Earth and Planetary Science.
*The heading picture of this post was found on GeoLog (EGU). Credits go to Matthias Vanmaercke.
*The original version of this essay was written by the author in “Challenges in Geography II” at the University of Berne in the spring semester 2018.