airGRmaps
  • Interface
  • Explanations
Selection of the outlet
Selection of the model
Metadata links

Why airGRmaps

This app allows users to retrieve regionalized parameters for use with the GR hydrological models (available in the airGR R package). GR parameters can be accessed by providing geographical coordinates, or by browsing the map.

Using the airGRmaps interface

The GR parameters have been pre-calculated on a regular 100 m square grid. Because of the unavoidable imperfections of any gridded representation of the hydrographic network, a visual check is required by the user to ensure that the selected parameters are relevant:

  1. Select one of the available tile maps
  2. Click on your point of interest
  3. Overlay the grid of contributing areas: each cell displays the upstream drainage area (in km²)
  4. Select the desired “river cell” before extracting the parameter set

The science behind airGRmaps

Parameterizing the GR hydrological models typically requires measured streamflow, in addition to the compulsory input data: Precipitation, Potential Evaporation and (where snow accumulation and melt play an important role) air temperature. Thanks to the measured streamflow data, an appropriate search algorithm (also called calibration algorithm) can identify the optimal parameter set for a given catchment and each of the GR hydrological models. However, sometimes, no measured streamflow is available to feed the calibration process. Consequently, the question of what we can do when the catchment we are interested in is truly ungauged arises. The airGRmaps application aims at providing a ready-to-use solution, based on an extensive regionalization work.

What do we call regionalization?

Hydrologists name regionalization the exercise that consists in guessing the most likely hydrological property at an ungauged point on a river. Usually, this is based on a “regional” analysis (hence regionalization) of relationships between the hydrological property and landscape or climatic characteristics. In our case, the hydrological property in question is a vector of 4 parameters for GR4J, 5 parameters for GR5J and 6 parameters for GR6J. Note that the two parameters of the CemaNeige snow accounting routine can be without fear set to their default value (see the evaluation of these default values in Valéry 2010) and therefore they are not regionalized in airGRmaps. Over the last 30 years, several authors have make regionalization attempts at INRAE (former CEMAGREF and IRSTEA): Edijatno (1991), Makhlouf (1994), Perrin (2000), Boldetti. (2012), Poncelet (2016). The airGRmaps are based on the regionalization relationships developed by Poncelet (2016) in her PhD. The method developed is known under the name TRUCAGE (for Transfert pour la Régionalisation Unifiée par CAlage GroupÉ) and is based on whole [1] parameter sets identified by regional [2] calibration.

References

Boldetti., G. 2012. “Estimation des paramètres des modèles hydrologiques sur des bassins versants non-jaugés : confrontation des approches directes et indirecte.” PhD thesis, AgroParisTech (Paris), IRSTEA (Antony). https://theses.hal.science/tel-02598146.

Edijatno. 1991. “Mise au point d’un modèle élémentaire pluie-débit au pas de temps journalier.” PhD thesis, Université Louis Pasteur (Strasbourg), CEMAGREF (Antony). https://theses.hal.science/tel-02575471.

Makhlouf, Z. 1994. “Compléments sur le modèle pluie-débit GR4J et essai d’estimation de ses paramètres.” PhD thesis, université Paris-Sud (Orsay), CEMAGREF (Antony). https://theses.hal.science/tel-02574920.

Perrin, C. 2000. “Vers une amélioration d’un modèle global pluie-débit au travers d’une approche comparative.” PhD thesis, INPG (Grenoble), CEMAGREF (Antony). https://theses.hal.science/tel-00006216.

Poncelet, Carine. 2016. “Du bassin au paramètre : jusqu’où peut-on régionaliser un modèle hydrologique conceptuel ?” PhD thesis, UPMC (Paris), IRSTEA (Antony). https://theses.hal.science/tel-01529196.

Valéry, A. 2010. “Modélisation précipitations–débit sous influence nivale. Élaboration d’un module neige et évaluation sur 380 bassins versants.” PhD thesis, AgroParisTech (Paris), CEMAGREF (Antony). https://theses.hal.science/tel-02594605.

[1] By “whole” parameter sets, we mean that our regionalisation is based on the transfer of optimal parameter sets, and not parameter values. Respecting the coherence of the parameter set (and the unavoidable interactions between parameters) helps increase the robustness of regionalized parameters.

[2] Regional calibration means that the optimal parameter sets have not been identified classically catchment by catchment but for groups of catchments.