Catchment classification and regionalisation with Self-Organising Maps (SOMs) and flexible model structures.

点击翻译按钮获取中文摘要

The major research objective of the project is catchment classification with a view to regionalization. Three different clustering methods serve as a basis for catchment classification: Self Organising Maps (SOMs), a benchmark conceptual model and the SUPERFLEX modelling framework. We will classify catchments by runoff behaviour and by hydroclimatic and physiographic properties. The project will use data from approximately 100 catchments in western Germany. Next, the classification methods will be compared and based on the best classification method we will regionalize runoff and model parameters to test its suitability for regionalization.Catchment classification as well as regionalization will be used to gain insight into the functioning of meso-scale river basins: to uncover resemblances and discrepancies, to interpret their hydrological behaviour and to provide meaningful data-backed-up hypothesis. Furthermore, we will develop auto-mated procedures of our methods and assemble them into a toolbox, which can be used by a wide group of users. This study will strive for a dual approach: firstly, to implement novel clustering methods with a view to regionalization and secondly, to stimulate the ongoing and highly relevant discussion on catchment classification and regionalisation.