Long-term predictions – can we make predictions about landscape form and function?

Abstract:

Landscape Evolution Models are now at the stage of development where if calibrated, they can provide meaningful and useful results. However, these models are largely calibrated with parameter values determined from present, or comparatively recent surface conditions which are themselves product of much longer-term geology-soil-climate-vegetation interactions. A significant issue for numerically based Landscape Evolution Models (LEMs) is their calibration/parameterisation and validation of any results. Before use, each LEM requires a set of data and parameter values for it to run reliably and most importantly produce results with some measure of precision and accuracy. Here we examine the reliability of a LEM to predict catchment form over geological time (here 500 000 years) for a potential rehabilitated mine landform using defensible parameters derived from field plots. The findings here demonstrate that there is no equifinality in landscape form. The employment of different parameter sets produce geomorphically and hydrologically unique landscapes throughout their entire evolution. Therefore parameterisation does matter at geological time scales. While at relatively short (<10 000 years) time scales the differences in hillslope form are minimal (as described by the hypsometric curve, area-slope and cumulative area distribution) there are large differences in sediment output. Therefore obtaining reliable and defensible parameters for input to LEMs is essential.