Additional datasets for model generators

[Huite]

Here are a number of datasets that would be useful to have available via HydroEngine. (Especially GLHYMPS2.0 is awfully unwieldy, as a 3.6 GB shapefile!)

I've provided an overview with a URL + which specific data are necessary, and licensing information if available.

Earth2Observe

I've had a look at the single (modelled) recharge dataset on E2O, it doesn't look very good to me. We have some data "locally" (P drive) from PCRGLOBWB (see last item), but there reasons you don't want to use that: I'm sure it has poor performance in some areas, and of course it doesn't have great resolution for our application.

There are of course plenty of alternative precipitation datasets available, and we preferably want one that's all tidied up and already available on EarthEngine; I suspect you guys have way more familiarity with that than I do. This one does have a license attached, so that's nice.

URL: https://wci.earth2observe.eu/data/group/earth2observe

Data: raster

Precipitation https://wci.earth2observe.eu/data/dataset/ecmwf_total_precipitation_monthly

To do

Possibly aggregate to mean mm/d, as input for steady-state recharge for groundwater models.

Licensing

1. Creative Commons Attribution Share-Alike: https://creativecommons.org/licenses/by-sa/2.5/
2. cc-by-nc: Naamsvermeling-NietCommercieel 2.0 Nederland,https://creativecommons.org/licenses/by-nc/2.0/nl/

SoilGrids250m

URL: http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0169748

Data

Site charateristics: raster

1. Absolute depth to bedrock (BDTICM)

Physical soil properties¹: raster

1. Bulk density (fine earth fraction) in t / cubic-meter (BLDFIE)
2. Soil texture fraction clay in percent (CLYPPT)
3. Coarse fragments in volumetric percent (CRFVOL)
4. Soil texture fraction silt in percentage (SLTPPT)
5. Soil texture fraction sand in percentage (SNDPPT)

¹ For these to be useful, I think we have to look into hydro-pedotransfer functions to generate hydrological parameters. @markhegnauer: I heard you mention this dataset, have people experimented yet with using with wflow?

Licensing

Data Availability: SoilGrids are available under the Open Database License (ODbl) v1.0 and can be downloaded from www.soilgrids.org and/or ftp.soilgrids.org without restrictions. https://opendatacommons.org/licenses/odbl/1-0/

NASA: Global 1-km Gridded Thickness of Soil, Regolith, and Sedimentary Deposit Layers

URL: https://daac.ornl.gov/cgi-bin/dsviewer.pl?ds_id=1304

Data: raster

Average sediment thickness: average_soil_and_sedimentary-deposit_thickness https://daac.ornl.gov/SOILS/guides/Global_Soil_Regolith_Sediment.html#datadescraccess

Licensing

Unfortunately, I'm unable to find a license on the webpage. They have a page on citation policy here, but again no mentions of licensing at all. I'm guessing we should send an e-mail to them.

For now: probably let the user download it, and set a local path in the config yaml.

GLobal HYdrogeology MaPS 2.0 (GLHYMPS 2.0)

URL: https://dvn.library.ubc.ca/dvn/dv/UVIC_RD/faces/study/StudyPage.xhtml;jsessionid=5bf3201e066631f7db3b027c18d0?globalId=hdl:11272/VINWU&studyListingIndex=0_5bf3201e066631f7db3b027c18d0

Data: shapefile

1. Permeability
2. Porosity

To do

A rasterized version is necessary for iMODFLOW, so having (clip + rasterize) functionality is the most convenient (see also issue #29).

Licensing

Creative Commons Attribution 4.0 International License. https://creativecommons.org/licenses/by/4.0/

PCRGLOBWB

Marta has sent me a link with the location one of the P drives.

Data: raster

1. Groundwater abstraction
2. Recharge

To do

iMODFLOW requires groundwater in abstraction form (even though it's transformed into a raster again for the groundwater model). The data is currently in (coarse) raster form. We can turn it into point data using cell midpoints, but this isn't without issues:

• Coarse cells means that all abstractions in a possibly relatively large area are summed into a single point abstraction.
• If the user chooses a relatively small cell size, this will result in one very deep pumping cone rather than many shallow ones.
• I'm not sure how the abstractions have been derived. This is one essentially pre-processed model input, so I wouldn't be surprised if there's a higher version of the data somewhere. We should probably enquire via Marta.

Possibility for now: resample to provided cell size, make sure total amount abstracted remains the same (does sum work for upsampling?). Then turn non-zero cells into point data using cell midpoints. This can easily be done locally, we might just want to check whether abstractions don't end up in the sea (but best not spend too much time).