Data Quality and Evaluation¶
This page summarises the quality, uncertainty, evaluation information, known gaps and caveats, and version history for HydroLand. Because HydroLand is driven by the Climate DT models, evaluation is model- and experiment-specific and is collected in the model-specific pages linked below.
Evaluation approach¶
Scientific evaluation of HydroLand assesses the quality of hydrological states, fluxes, and indicators derived from the driving climate models ICON, IFS-FESOM, and IFS-NEMO. The evaluation currently covers three aspects:
Bias-adjusted meteorological forcing — because HydroLand is driven by bias-adjusted precipitation and temperature, the forcing is evaluated against ERA5-Land reanalysis to understand how input quality propagates into the hydrological outputs.
Land-surface fluxes — actual evapotranspiration is compared with the satellite-based GLEAM dataset.
Routed river discharge — simulated discharge is compared with GRDC gauge measurements.
The evaluation is conducted separately for individual continents, because hydrological model performance varies across regions.
Reference datasets¶
Domain |
Reference dataset |
|---|---|
Precipitation and temperature |
ERA5-Land (Muñoz-Sabater et al., 2021) [Muñoz-Sabater et al., 2021] |
Evapotranspiration |
GLEAM v3 (Martens et al., 2017) [Martens et al., 2017] |
River discharge |
GRDC — Global Runoff Data Centre (2023) [Global Runoff Data Centre, 2023] |
Current evaluation status¶
The available evaluation material currently focuses on the IFS-NEMO configuration. It covers historical bias-adjusted precipitation and temperature, actual evapotranspiration against GLEAM, and routed river discharge against GRDC gauges, reported per continent. The same model-specific page also includes a future-minus-historical projection analysis for the 2015–2049 period, using projected meteorological forcing, HydroLand evapotranspiration, and routed river discharge. The full IFS-NEMO assessment is available in IFS-NEMO. This page will be updated as further model configurations and periods become available.
Model-specific evaluation and status pages¶
Known gaps, anomalies, and caveats¶
The current documentation identifies the following known gaps and caveats:
Evaluation material for further model configurations and periods will be added as it becomes available.
Irrigation, reservoir operation, and water extraction are not represented. HydroLand therefore simulates river discharge as if these human water management processes were absent. When compared with discharge measurements such as GRDC gauge records, simulated discharge in heavily managed basins may therefore be higher than measured discharge.
Potential evapotranspiration is estimated from temperature and latitude using the empirical formulation of [Oudin et al., 2005]. This should be considered a limitation because it is less physically detailed than energy-balance PET methods, which are not used in the current configuration. More physically based PET approaches are being considered for future HydroLand versions.
Regions with poorer performance are generally associated with limited data availability for model development and calibration, which is often the case in regions of the Global South. The model-specific evaluation pages, starting with IFS-NEMO, provide the current regional results and help identify where these shortcomings affect interpretation.
HydroLand version¶
The current data release was produced with HydroLand version 1.2.6 (Zenodo record).
The authoritative change log is maintained in the HydroLand package repository.