A new approach for large scale simulation of complex spatial processes: The 3D soil evolution model mARM

Modelling complex spatially distributed processes over large areas is a challenge facing many environmental disciplines. In some cases even when the physics and relationships are known they are too complex for large scale simulations. In the field of pedogenesis researchers have a good understanding of the physical processes affecting soil evolution but struggle with predicting their spatial and temporal variability. Spatial and temporal relationships in soil properties are known to be a significant factor in many hydrological and geomorphological processes. Modelling detailed distribution of soil properties is therefore important but yet to be fully achieved over large scales. In order to predict the spatial distribution of soil properties a new approach is needed in the form of detailed large spatial and temporal scale simulation of soil evolution. These simulations could not be done using conventional physically-based model due to their computational complexity and limited dimensionality. A more comprehensive and computationally efficient algorithm is needed. Our solution is a new pedogenesis model called mARM. mARM (matrices-ARMOUR) uses transition matrices in conjunction with physically-based equations to achieve a highly modular and computationally efficient modelling platform. This new approach and the modularity of the mARM platform allow us to simulate a variety of soil processes over long periods and large areas (millions of years and 1000's pixels). In its latest version (presented here) mARM has been extended to three spatial dimensions (mARM3D) by adding a soil profile component. This 3D model is a major breakthrough in pedogenesis modeling since this is the first time profile as well as surface processes are simulated over large scales. The modularity of mARM3D allows us to use it as a virtual laboratory to examine a variety of soil processes and relationships such as: • Depth-dependent weathering rate equations (presented here) • Runoff fluctuations (as a result of climate or environmental changes) • Weathering geometry (proportion of particle split) • Weathering-Erosion relationships and rates In addition to the above capabilities, mARM is intended to be integrated as a component in a landform evolution model (i.e. TelluSim) which will allow such a model, for the first time, to account for time and space variation in soil properties. In this paper we describe the mARM3D model framework which can potentially be used in other complex spatial models. We also present selected results from the mARM3D simulations at hillslope and landscape scale which show the effect of profile weathering physics on surface soil distribution.