Models to predict net and gross sedimentation in lakes

Abstract

This study models net and gross sedimentation in lakes. The fluxes of material play an important role in most lake contexts. They influence transport, bio-uptake and ecological effects of most toxins and nutrients. The aim is to present a new type of 'mixed' dynamic/statistical model and to discuss advantages and disadvantages with this approach. Empirical data to validate the model emanate from sediment traps from 25 lakes. Rates of gross sedimentation in traps were compared to catchment and morphometric parameters determined from maps in an attempt to identify the processes that influence sedimentation. The most important variables are relative depth (linked to resuspension), forest and open land cover (of the near area of the catchment), relief of catchment, coverage of mires and lakes, and the lake water retention time. This study shows that it is easy to lose predictive power in dynamic models relative to empirical models, in which each parameter constant and x-parameter automatically accounts for the complexities in natural ecosystems. In dynamic models empirical knowledge is replaced by logical constructions, which may not be accurate. The mixed model would need to be extended with new large submodels to predict, for example mean distribution coefficient of total-P in tributaries, tributary suspended load, rate of sedimentation and resuspension rate. This would result in a huge 'prescriptive' model with wide uncertainty limits for the predicted mean values for net and gross sedimentation.