Seasonal Water Yield

Hello all,

I have recently completed a SWY model for a regional scale project in the U.S. and am considering some updates. I have a couple of questions:

  • I’ve been using actual ET as an input in the model. However, I think
    I should use PET. Does anyone know of reliable sources for PET data covering the periods from 2000-2020? I found the people mentioned they used a dataset on Global Aridity Index and Potential Evapotranspiration (ET0) Climate Database v2, but it includes only one layer, while for the SWY I need monthly for 20 years. Can I use AET instead of PET? If not, can you please inform me what sources I can the data?
  • Could anyone offer insights on defining Threshold Flow Accumulation, alpha, beta, and gamma variables based on DEM, etc. Any guidance or examples would be greatly appreciated.

Thank you so much!

Hi hkarimi, thanks for reaching out. Here’s a link to where you can download the monthly ETo. Hover over the first archive to see the name. You should also use WorldClim precipitation, which this dataset is calculated from. And try to match the 20 year periods, check the ETo documentation to determine those years. Global Aridity Index and Potential Evapotranspiration (ET0) Climate Database v3
worldclim: Historical monthly weather data — WorldClim 1 documentation

As for alpha, beta, and gamma, click the i button next to the inputs in InVEST and find the defaults, use these for your first run, then work with your team to refine as needed if/when you calibrate.

As for TFA, I answered this question before, see if this helps you…RouteDEM issues; DEM artifacts - #7 by Nadine

Please let me know if you need clarification or more information or if these responses address your questions.

Hi @hkarimi ,

Thanks for writing in to the forum!

Have you consulted the InVEST User Guide? You’ll read there that the expected model inputs are monthly reference ET (ET0) rasters. ET0 is different than actual ET (AET) in that the former is not limited by water availability. I don’t know of any sources for monthly ET0 for 2000-2020, but I do know of one for 1970-2000. You can find data sources in the User Guide here. We recommend CGIAR, which you can download from Figshare.

The Threshold Flow Accumulation (TFA) value is explained here. You’ll note that there’s no exactly “correct” TFA value. As stated in the User Guide, “The correct value for your application is the value that causes the model to create a stream layer that looks as close as possible to the real-world stream network in the watershed. Compare the stream network raster output with a known correct stream map, and adjust the TFA accordingly - larger values of TFA will create a stream network with fewer tributaries, smaller values of TFA will create a stream network with more tributaries. A good value to start with is 1000, but note that this can vary widely depending on the resolution of the DEM, local climate and topography. Note that generally streams delineated from a DEM do not exactly match the real world, so just try to come as close as possible. If the modeled streams are very different, then consider trying a different DEM.

As for the alpha, beta, and gamma, parameters, we recommend starting with, and sticking to, the default parameters unless you have additional hydrological expertise or insights. As defined in the User Guide, those defaults are alpha = 1/12; beta = 1; and gamma = 1. Read more details about each of these here.

I hope that helps.

Happy modeling,

Hello Jesse,

Thank you so much for your message and valuable information. Yes, I know that I should use Et0, but unfortunately, I could not find the ET0 for the periods 2000-2020. I hope anyone can help regarding this!


Hello Nadine,

Thank you so much for providing the links. But I wonder that Global Aridity Index and Potential Evapotranspiration (ET0) Climate Database v3 only contain 12 monthly layers, not for each year specifically. Do you have any idea how I can find for 2000-2020?

Oh, those rasters are average monthly values over multiple years. The range is 1970-2000 period. .So, unlike the precipitation data, you don’t need to calculate the averages, it’s done for you.

Since CGIAR doesn’t have more recent ETo, you could run the model with older precipitation to match. Or you can try to find matching CHIRP.S precipitation and ETo here…Data Downloads CHIRPS | Early Warning and Environmental Monitoring Program

Hi @hkarimi -

If you can’t find existing ET layers that match your precipitation data, you can always calculate ET from monthly precipitation using one of a variety of ET equations. The equation I usually use is Modified Hargreaves, since it is relatively easy to calculate. It requires monthly precipitation, minimum/maximum temperature and average monthly radiation. The User Guide provides a little more information about this and couple of other equation options. A web search will provide other options as well.

~ Stacie

Hi Stacie,

Thank you so much for your message. I am a bit confused. What do you mean by “ET layers that match your precipitation data”. Actually, the site people said used for ET0 just contains the average for a month, not each year separately.



Thank you so much for your help.


Hello Stacie,

Can you please tell me how I should match ET layers with precipitation data? I cannot understand what you mean by that. Thanks a lot.


I was referring to two related things:

  • The ET data should be based on the same precipitation data that you’re using as input to the model.
  • The time period that the ET data covers should be the same as the time period that the precipitation data covers.

~ Stacie