An alternative method for deriving a USLE nomograph K factor equation

Abstract

Artificial landscapes are often used in landscape ecological models to investigate questions around movement of individuals, the spread of invasive plants and diseases or the stability of meta-populations. Using these artificial landscapes has some strong benefits such as simplification of landscape complexity, a means to replicate landscape scenarios or the potential to study systematic gradients. Communicating the results to non-theoreticians, however, is often difficult because there is often no quantitative comparison made to link them with reality. Other models that directly use digitized real landscapes as input do not have this barrier, but they are often lacking generalizability or the potential to forecast the effect of changes in the environment.In this study, we propose a method to generate artificial landscapes with key parameters derived a priori from real landscapes through analyses based on Geographical Information Systems (GIS) data: surface cover, spatial aggregation and patch size. Existing methods of generating artificial landscape often do only a posteriori comparisons to prove that their landscapes are 'realistic'. We show that, by estimating the parameters beforehand, we can generate artificial landscapes incorporating multiple land use types that can be directly compared to existing landscapes in a quantitative manner. This allows for more targeted landscape generation and vastly reduces the parameter space that needs to be covered. At the same time, this method does not reduce the potential of the models in terms of being reproducible and transferable.We show an application of this method with two contrasting, complex agricultural landscapes from eastern Sub-Saharan Africa and South-east Australia. We artificially generated landscapes using key parameters from an analysis of digitized real landscapes and found that the algorithm allows flexibility in single target parameters while retaining 'realism'. Realism is assessed in different way. For the Australian data, we compare ranges in land use cover and aggregation between real and generated landscapes. For the African data, we found a log-linear relationship between these two variables in the empirical data that we then used to generate the realistic artificial landscapes. By creating a measurable link between real and artificial landscapes this method will help reduce communication barriers between theoretical scientists and the general public, increasing the impact of our science.