Keywords
rainfall-runoff database, flood risk management, random rainfall generator, parsimonious
Start Date
1-7-2008 12:00 AM
Abstract
A comprehensive rainfall-runoff database for the Mulde catchment in Central Germany was developed to support flood risk assessment and flood management. A large number of randomly generated 20-day rainfall scenarios and several combinations of model initialization states are the basis for the simulation of basin response by applying a metric conceptual rainfall-runoff model. The rainfall scenarios are divided into three periods; a five day initialization period to represent pre-conditions, a two day storm period representing a rainfall extreme event, and a final period characterized by low rainfall volumes to study streamflow recession behaviour. The rainfall-runoff database can be used as an effective tool to easily assess possible streamflow situations assuming different rainfall volumes for the following days. The benefit of the database approach is that it can be easily used by persons who are not necessarily familiar with hydrologic modelling, because the modelling step has already been accomplished. The study area was strongly affected by the extreme flood event in August 2002. In order to study the rainfall-runoff models' capabilities to capture this extreme event it was calibrated to training sets of input data including and not including this event. The model was calibrated using different objective functions, the Nash- Sutcliffe efficiency and a modified Nash-Sutcliffe efficiency adapted to high flow conditions. Only simulations with model settings calibrated to time series including the extreme event were capturing the flood event sufficiently. Due to the growing complexity in rainfallrunoff modelling during seasons affected by snowmelt processes, the applicability of the database in the current state is limited in snow affected catchments to the warm season.
A rainfall-runoff database to support flood risk assessment
A comprehensive rainfall-runoff database for the Mulde catchment in Central Germany was developed to support flood risk assessment and flood management. A large number of randomly generated 20-day rainfall scenarios and several combinations of model initialization states are the basis for the simulation of basin response by applying a metric conceptual rainfall-runoff model. The rainfall scenarios are divided into three periods; a five day initialization period to represent pre-conditions, a two day storm period representing a rainfall extreme event, and a final period characterized by low rainfall volumes to study streamflow recession behaviour. The rainfall-runoff database can be used as an effective tool to easily assess possible streamflow situations assuming different rainfall volumes for the following days. The benefit of the database approach is that it can be easily used by persons who are not necessarily familiar with hydrologic modelling, because the modelling step has already been accomplished. The study area was strongly affected by the extreme flood event in August 2002. In order to study the rainfall-runoff models' capabilities to capture this extreme event it was calibrated to training sets of input data including and not including this event. The model was calibrated using different objective functions, the Nash- Sutcliffe efficiency and a modified Nash-Sutcliffe efficiency adapted to high flow conditions. Only simulations with model settings calibrated to time series including the extreme event were capturing the flood event sufficiently. Due to the growing complexity in rainfallrunoff modelling during seasons affected by snowmelt processes, the applicability of the database in the current state is limited in snow affected catchments to the warm season.
