question on freshwater use from groundwater in water demands

[laurenebouaziz]

We are looking at modelling water demands and are wondering about the approach used in the global Lisflood (GLOFAS) model regarding the input data.
One of the input data needed in order to model water abstractions is the fraction of freshwater use from groundwater (as opposed to surface water and non-conventional sources).
We downloaded the input map fracgwused_Global_03min.nc and we are wondering what this map actually contains.
I could not find precision in the documentation or linked paper but I read in the European application that the country or admin level statistics are masked for groundwater bodies and recomputed by water region area to try and fit the original statistics after mapping.
The lines in the code that do this operation are here: https://github.com/ec-jrc/lisflood-code/blob/master/src/lisflood/hydrological_modules/waterabstraction.py#L227-L233

What we are wondering is, if the fracgwused_Global_03min.nc map is the original statistics before masking the groundwater bodies and upscaling to the water region (maps also available in that same folder) or the map after these processes are done.

See below an extract of the map for the Alps.

Many thanks in advance for your help,

[StefaniaGrimaldi]

Dear @laurenebouaziz,

thank you for your interest in OS LISFLOOD and in the global implementation maps.

I am not sure that I understood your question; I will write some technical and background information: I hope that a step-wise approach will help us identifying the solution to your inquiry.

The input maps fracgwused and gwbodies were generated according to the approach explained in Gelati et al. (2020)

These maps were not updated during the work of the more recent publication Choulga et al. (2024), hence they are not described in the manuscript, as you correctly pointed out.

The work of Choulga et al. (2024) led to the generation of an entirely revamped and improved dataset. Nevertheless, further updates and improvements are always necessary! In fact, we are already working on the next release of the dataset (which will include, for instance, updates to the soil maps). Regarding the fracgwused, we will verify the existence of relevant updates in the raw datasets.

In the current map, some pixels are set to zero, these pixels largely correspond to pixels with gwbodies=0. This is consistent with https://github.com/ec-jrc/lisflood-code/blob/master/src/lisflood/hydrological_modules/waterabstraction.py#L233

The use and meaning of the water regions map is explained in https://ec-jrc.github.io/lisflood-model/2_18_stdLISFLOOD_water-use/, please note the section "Important techical note for the adequate definition of the water regions".
The map provided in https://jeodpp.jrc.ec.europa.eu/ftp/jrc-opendata/CEMS-GLOFAS/LISFLOOD_static_and_parameter_maps_for_GloFAS/Water_demand/ was generated using the OS LISFLOOD utility
https://github.com/ec-jrc/lisflood-utilities?tab=readme-ov-file#waterregions and it is consitent with GloFAS v4 inter-calibration catchments (https://confluence.ecmwf.int/display/CEMS/GloFAS+v4+calibration+methodology+and+parameters)

I hope that the pieces of information above are useful,
we are at your disposal for further discussions,
kind regards,
Stefania

[laurenebouaziz]

Dear Stefania,

Thank you very much for your prompt reply and all the details provided. I think the question we are after is:

Are the pixel values in the fracgwused map valid at the national or regional scale (average values) or are these values already corrected for the non-Groundwaterbodies. This correction takes place in the following code section https://github.com/ec-jrc/lisflood-code/blob/master/src/lisflood/hydrological_modules/waterabstraction.py#L227-L236. We think the values in the fracgwused map are not yet corrected for the non-Groundwaterbodies, as this happens in the code, but are not completely if this is indeed the case.

I hope this clarifies our question.

Thank you and best wishes,
Laurène

[StefaniaGrimaldi]

Dear @laurenebouaziz,

thank you for the clarification. I would like to perform a quick check before providing an answer. I am out of office for meetings and a conference until Friday night. I will get back to you early the next week.

Kind regards,
Stefania

[laurenebouaziz]

Dear Stefania, thank you very much for your reply, I will be away next week, but look forward to seeing your reply when I return.
Best wishes, Laurène

[StefaniaGrimaldi]

Dear @laurenebouaziz,
your hypothesis is correct: the code internally adjusts fracgwused to account for the non-groundwaterbodies. The partial masking of the fracgwused map highlighted by your screenshot is the outcome of subsequent developments. As mentioned in my previous message, that specific map was generated a few years ago and we aim to verify the existence of relevant updates in the raw data in view of the next release of the global dataset of static maps.
I hope that my answer helps,
kind regards,
Stefania

[laurenebouaziz]

Dear Stefania,

thank you for you reply. So just to be really sure:
Is the map in the screenshot the fracgwused before or after the following code section https://github.com/ec-jrc/lisflood-code/blob/master/src/lisflood/hydrological_modules/waterabstraction.py#L227-L236

thanks again and kind regards,
Laurène

[StefaniaGrimaldi]

Dear Laurène,
all maps in https://jeodpp.jrc.ec.europa.eu/ftp/jrc-opendata/CEMS-GLOFAS/LISFLOOD_static_and_parameter_maps_for_GloFAS/ are used as input to the OS LISFLOOD model.
I understand your perplexities regarding the map in the screenshot: as mentioned above, that map is the outcome of subsequent developments (hence, the partial masking), and it currently under revision in view of the next release. The input map is expected to show the data before the corrections implemented within the code.
Please do not hesitate to share your questions and comments - users' feedback is always appreciated!
Kind regards,
Stefania

[laurenebouaziz]

Dear Stefania, thank you very much for this clarification. We will make sure to also check the map again in the next release.
Best wishes, Laurène