Towards hydrological model calibration using river level measurements

Due to the limited availability of discharge data in many catchments globally, it is important to develop a calibration method that does not rely solely on discharge data. Motivated by this limitation, two calibration approaches using water level data directly in hydrological calibrations are proposed in the study. The first is a Spearman Rank correlation (SRC) based scheme, which calibrates modelled streamflow against observed water level using Spearman Rank correlation. The second is an Inverse Rating Curve (IRC) function based scheme, which introduces three more parameters to simulate water level from an inverse rating curve. The new approaches are tested in 11 catchments in Australia and the resulting discharge predictions show good correlation with observations. However, the results present large biases between observations and estimated discharge data due to the inherent limitation of the approach: absence of information on the true discharge range in the calibration process. To mitigate the biases, the magnitude-sensitive SRC/IRC-based schemes that incorporate a small number of observations are developed in this study. The bias issue is then mitigated significantly, but the improvement is not consistent throughout the examined catchments. One of the most critical challenges of the bias correction is that the whole dynamic range of discharge is constrained by a few observed discharge data, but overall, the new calibration approaches using only water level data prove to be promising.