Dynamic TOPMODEL

TL;DR: A new implementation of Dynamic TOPMODEL is presented that provides much improved usability and interoperability whilst retaining the advantages of the original version in terms of simplification of landscape complexity into response units.

Abstract: In 2001, Beven and Freer introduced a "dynamic" variant of TOPMODEL that addressed some of the limitations of the original model whilst retaining its computational and parametric efficiency. The original assumption of a quasi-steady water table was replaced by time-dependent kinematic routing within hydrological similar areas. The new formulation allows a more flexible discretisation, variable upslope drainage areas and spatially variable physical properties.There has, however, never been a freely distributable version of dynamic TOPMODEL. Here, we describe a new, open source, version developed in the R environment. It incorporates handling of geo-referenced spatial data that allows it to integrate with modern GIS. It makes use of data storage and vectorisation features of the language that will allow efficient scaling of the problem domain.The implementation is evaluated with data from a small catchment. The formulation of the model in terms of a flow distribution matrix is described and its use illustrated for treatment of surface and subsurface flow routing. The model uses an improved implicit solution for updating the subsurface storages and fluxes. The paper focuses on the robustness of the predicted output variables to changes in the time and space discretisations. A new implementation of Dynamic TOPMODEL is presented that provides much improved usability and interoperability whilst retaining the advantages of the original version in terms of simplification of landscape complexity into response units.An improved algorithm for subsurface routing has been implemented to address the shortcomings of the previous version when landscape units were not strictly ordered in their downslope positions within the meta-hillslope representation of the catchment.The new implementation is evaluated against data for an upland catchment and its response to varying time and space intervals evaluated.Some guidance is provided for selecting suitable time and space steps when running the model.