Calibrating a rainfall-runoff model in a data scarce catchment in Mozambique

More Info
expand_more

Abstract

Rainfall runoff modeling is a major task of many hydrologist. Therefore always a lot of research is done to modeling. One of the problems of modeling is often the lack of data. Most of the world is data scarce, in the sense that no discharge is being measured. Although this is the case, it is often preferable to make a model of a catchment in such an area. Therefore a lot of research is carried out in calibrating a rainfall runoff model without discharge data. This research is also focusing in this. With a fieldwork of 3 weeks and knowledge of the local fishermen, the catchment is being investigated. The catchment is in the south of Mozambique, and is the catchment of the Lumane river. This is the last tributary of the Limpopo river. Upstream of the Lumane is a fresh water lake, which feeds the river. The yearly fluctuating of this lake is used for the calibration of the rainfall runoff model. With information of the local fishermen, a hydraulic model of the river and the yearly precipitation pattern; a Fourier series is made, describing the lake’s water level. This Fourier series is used for the calibration of the rainfall runoff model. Next to this, also GIS data of actual evaporation is compared with the outcome of the model. This is also used for the calibration. In order to validate the model is looked at how well the model is following the Fourier series of the lake’s water level. Furthermore is checked if the fluctuating of the reservoirs is reliable, as well as how reliable the calibrated parameters are. Since the rainfall runoff model is a lumped model, but based on an hydrological classification, the parameters really mean something. Therefore can be seen if the model is acting as suspected. From the results it becomes clear that the model is mimicking the level of the lake. Also the discharge data seems to be in the right order of magnitude compared to historical data and local measurements. Only the evaporation outcome differs a lot from the GIS data during summer time. The reliability of this GIS data will therefore be discussed as well as the shortcomings of the model structure. After all one can conclude that it is possible to calibrate a rainfall runoff model without discharge data. However, to validate the model, discharge data is necessary. Else, it is hard to know how good the model really is.