Jensen and Lindholm Abstract

In this project, the rainfall-runoff for two different catchments in the Limpopo province in South Africa, have been modelled using MIKE 11 NAM model. The NAM model is used to analyse the water balance ratio and the hydrological fluxes for different environments.

To achieve this there have been used data from measuring stations in South Africa, which have been compared to data from a satellite (TRMM) and data based on re-analysis (Era-5). But because of the significant variations in the re-analysis data compared to the other sources, the NAM-model has not been analysed for this source.

The catchments have two seasons, one that is dry, with nearly no form for precipitation, and a wet season where most of the precipitation falls. The NAM-model have trouble calibrating with the irregular rainfalls, not achieving optimal R2-values (HoutSand R2=0,702, Letsitele R2=0,55).

The NAM-model also had a hard time calibrating the validation, because of the recent years being much drier than the years in the calibration period, which resulted in much lower R2-values for the validations (Hout/Sand R2=0,314, Letsitele R2=0,36).

The Nash-Sutcliffe values achieved for the TRMM data was similarly not optimal with either the calibration (Hout/Sand R2=0,670, Letsitele R2=0,44) or validation (Hout/Sand R2=0,429, Letsitele R2=0,23)

Through the use of the NAM-model, the groundwater has been found to only increase after events of heavy rainfalls, whereupon the amount of groundwater will slowly decrease until the next heavy rainfall.