Introduction

The GloFAS v4.0 hydrological model performance was evaluated in the model calibration context in GloFAS v4 calibration hydrological model performance, using only the stations involved in the calibration and only over the specific periods related to the calibration excercise.

On this page, the model performance is analysed over the whole period of 1979-2021, with any available observations that met the quality criteria, specified in the verification methodology page (place holder GloFAS hydrological performance verification methodology). In total, 1987 stations were used for the v4.0 verificaiton and 1949 for the v4.0 vs v3.1 model comparison, with at least 1 year of quality checked data and minimum reservoir or lake influence. For the verification methodology and used metrics please see the page place holder GloFAS hydrological performance verification methodology.

General v4.0 performance

For this comparison, we used all stations with good quality river discharge observations and minimal human or lake influence that could be mapped (find the corresponding model river network location) onto the higher resolution v4.0 river network. In total 1987 stations could be considered as shown below with the available observation length (gaps are removed to compute the length). 

Figure 1. Number of years of available river discharge observations in the 1979-2021 reanalysis period.

The generic GloFAS v4.0 model performance is measured by the modified Kling Gupta efficiency in Figure 2. High skill (above 0.7) is shown over much of the higher latitude areas and also some southest Asian and central south American areas. The lowest KGE, including even some catchments with no skill at all (below -0.41), are spread across some tropical areas, often in central southern USA and Mexico and some areas in Africa, often in the drier climate.

Figure 2. KGE of the GloFAS v4 simulation.

The KGE's component scores highlight that much of the lower KGE skill comes from the often high and mainly positive bias, and larger variability errors. The correlation is more homogeneous, even though many of the low KGE areas also show low correlation, with exceptions, such as the 

Figure 3. Bias and variability ratio and Pearson correlation of the GloFAS v4 simulation.