GloFAS v4, the 0.05 degrees quasi-global (-180,180,90,-60) implementation of the LISFLOOD model, was calibrated using in-situ discharge gauge stations with a minimum drainage area of 500 km2 and at least 4-years-long time series of measurements more recent than 01 January 1980.
1996, quality checked, at least 4-years-long time series of discharge measurements more recent than 01 January 1980, were available for calibration. More specifically, 212 calibration points were in Europe, 250 in Asia, 61 in Oceania, 420 in Africa, 617 in Centre-North America, and 436 in South America; the total gauged area entailed 47.5 % of the quasi-global domain (Figure 5, yellow area).
Firstly, the Distributed Evolutionary Algorithm for Python (DEAP, Fortin et al. 20121) was used to optimize the parameters of catchments for which discharge data were available (gauged catchments). The calibration tool is available open source (ec-jrc/lisflood-calibration: Lisflood OS (Calibration tool) (github.com)) and it allowed to estimate the values of 14 LISFLOOD-OS parameters, with the purpose to optimize the modelling of snow melt, water infiltration into the soil, surface water flow, groundwater flow, lakes and reservoirs dynamics. Secondly, a pragmatic regionalization approach was implemented to transfer the parameters from the gauged catchments (donors) to the ungauged catchments. The donor catchments were identified according to the highest geographical proximity and climatic similarity to the ungauged catchments (e.g. Parajka et al 20053, Beck et al. 20164). The modified Kling Gupta Efficiency (Gupta et al., 20092) was selected as objective function and a minimum drainage area of 500 km2 was used for both the above explained steps of the calibration.
At the end of the calibration, 14 parameter maps with quasi-global extent were produced. These parameter maps were used to execute the long-term run (LTR), a continuous simulation with model forced with ERA5 reanalysis, for the period 01/01/1979-31/12/2019 (the first three years are generally excluded from evaluation, the exceptions to this rule are explained here XXX). Simulated daily discharge was then compared against observed discharge from the 1996 calibration stations. Because of the unequal length of observation across the domain especially for nested catchments, hydrological modelling performance is evaluated over all available discharge data rather than calibration and the validation periods separately. This page summarises GloFAS v4 hydrological skill.
Overview
The hydrological performance of GloFAS v4 is expressed by the modified Kling-Gupta Efficiency (KGE') (Knoben et al. 2019). A detailed explanation of the modified Kling-Gupta Efficiency (KGE') is available from EFAS hydrological model performance.
Figure 10 provides the cumulative distribution function of KGE' values and the KGE' distribution for the 1996 calibration stations. The median KGE for the gauged catchments was 0.70, with the calibrated parameters leading to higher accuracy than the mean flow benchmark for 92.9% of the gauged catchments (i.e. KGE’ > -0.41, Knoben et al. 2019)
Figure 10 – KGE' distribution and cumulative distribution function for all calibrated stations
Figure 11 presents GloFAS v4 results in terms of KGE' components that represent respectively: the linear correlation between observations and simulations (correlation), a bias term (mean bias) and a measure of the flow variability error (variability bias) (Knoben et al., 2019). Results are presented for all calibration stations.
Figure 12 – KGE' components (correlation, bias, variability) distribution and cumulative distribution function for all stations
Spatial analysis
Figure 13 shows the spatial distribution of the GloFAS v4 hydrological performance. KGE' values > 0.7 are shown in light blue and blue. KGE’ values < -0.41 are shown in black. KGE' is generally uniformly distributed across the domain, with higher performance (blue) in large parts of North and South America, Central Europe, and Asia. Calibrated catchments with high performances are also found in Africa and Oceania. The lowest performances (black) are often concentrated in catchments with strongly regulated rivers.
Figure 13 – KGE' spatial distribution
A low score during evaluation of LISFLOOD OS model calibration is not necessarily an indicator for decreased forecast performance of global flood awareness system. GloFAS forecasts are compared to model derived thresholds (Thielen et al., 2009; Bartholmes et al., 2009), which eliminates systematic bias that leads to an overall lower score in hydrological performance. However, correlation is a desired quality in hydrological performance as it represents the timing of flood peaks. Given the mathematical structure of KGE', all stations where KGE'>=0.7 have correlation >=0.7 (Gupta et al., 2009); but some of the stations where KGE'< 0.7 can still have correlation>0.7, associated to a large mean bias and/or variability bias. Calibration points with low KGE' but correlation >=0.7 won't decrease the forecast performance of the Global Flood Awareness System, even if forecast discharge will exhibit large bias. Figure 14 shows a combination of the spatial distribution of GloFAS KGE' and correlation. Stations with KGE'<0.7 and Correlation>=0.7 are highlighted in white. Compared to Figure 13, 336 calibration stations with KGE<0.7 (yellow to black) show a Correlation>0.7 (white).
Figure 14 – Spatial distribution of the hydrological performance (KGE') across the domain combined with correlation: stations with KGE'<0.7 and correlation>=0.7 are highlighted in white.
Figure 15, 16 and 17 present the spatial distribution of GloFAS v4 hydrological performance across quasi-global domain in terms of KGE' components: correlation, mean bias and variability bias.
Figure 15 – Spatial distribution of correlation
Figure 16 – Spatial distribution of bias
Figure 17 – Spatial distribution of variability
Figure 17 – Spatial distribution of variability
Comparison of GloFASv4 against GloFASv3: overview
GloFAS v4 is the first GloFAS version using 0.05 degrees resolution, therefore allowing the representation of hydrological processes with 4 times higher resolution than all the previous GloFAS versions. Moreover, compared to the former 0.1 degrees resolution set-up, the higher resolution set-up was developed by making use of the latest research findings, remote sensing and in-situ data collections. These significant differences in the model set ups hinder a quantitative comparison between GloFAS v4 and GloFAS v3. However, an attempt was made to show the improvements of the new GloFAS v4 compared to GloFAS v3. GloFASv3 implementation set-up and calibration is described into detail by Alfieri et al. 2020.
Out of the 1996 calibration stations used in GloFAS v4, only 1173 could be used for the comparison with GloFAS v3 calibration. GloFAS v3 used 1226 calibration stations (Alfieri et al. 2020), however, 53 of those stations were not included in the calibration of GloFAS v4 because they were replaced by near-by stations with longer and higher quality data.
It is important to note that calibration periods in the two versions could be different, therefore the KGE' were computed using observed daily discharge for the entire observation period available to GloFASv4 calibration.
Figure 18 shows the KGE' cumulative distribution functions for the 1173 shared stations: GloFAS v4 in red and GloFAS v3 in black. The entirely revamped, high resolution model set-up and new calibration shows an increase in the percentage of stations with KGE' > 0.7, from 38% to 58%. Moreover, the percentage of stations for which calibrated parameters lead to lower accuracy than the mean flow benchmark (KGE’<-0.41) decreased from 13.6% to 2.8%.
Figure 18 - KGE' Cumulative distribution function for GloFAS v3 (black) and GloFAS v4 (red)
Comparison of GloFASv4 against GloFASv3: spatial analysis
Figure 21 presents the spatial distribution of KGE' skill score between GloFAS v4 and GloFAS v3 (benchmark) for the 1173 common stations. Improvements are represented in green and cyan, substantially similar values with +- 0.05 in KGE' skill score have no colour (white), degradations are represented in orange and red. KGE' skill score is generally positive over the entire model domain.
Figure 21 - Spatial distribution of KGE' skill score between GloFAS v4 and GloFAS v3 (benchmark).
Figure 22 presents the spatial distribution of difference between GloFAS v4 KGE’ and GloFAS v3 KGE’(benchmark) for the 1173 common stations. Improvements are represented in green and blue, substantially similar values yellow (TO DO: make it white), degradations are represented in orange and red.
Figure 22 - Spatial distribution of KGE' difference between GloFAS v4 and GloFAS v3 (benchmark).