Evapotranspiration raster for UCM

Science, Research & Data
1

Hi all,

I’m working on an urban heat island mitigation study using the InVEST Urban Cooling Model, and I’d appreciate some feedback or suggestions.

:white_check_mark: What I’ve done so far:

  • I created a biophysical table based on FAO56 crop evapotranspiration data.

  • For each LULC class (agriculture, urban green, forest, etc.), I calculated a seasonal average Kc based on growth stages and durations.

  • I used a 5-month average Penman-Monteith reference ET₀ = 3.334 mm/day, computed from station data for the growing season (May to September).

  • I multiplied each land use’s Kc × 3.334 to get a proxy ET raster in mm/day.

:pushpin: Where I need help:

  1. Is this a valid method to derive rasterized ET₀ input for the Urban Cooling model?

    • The model requires ET₀ as a raster layer (in mm/day), but my ET raster is Kc × average ET₀.

    • Would this be considered ETc instead? If yes, what’s the recommended way to get ET₀ as a raster?

  2. How can I generate a spatially distributed raster of ET₀ (reference evapotranspiration)?

    • I’m using ArcGIS and QGIS.

    • Are there trusted global or regional raster datasets for ET₀?

    • Or should I interpolate ET₀ from nearby stations (I have daily values)?

  3. Would it be better to use the Hargreaves method (less input data) instead of Penman-Monteith to create a monthly ET₀ raster?

NB

  • LULC map: 2m raster,

  • Reference ET data: May–Sept 2024 from station

2
  1. since you have the data for actual ET (=Kc * ETo), you might be able to get ETo if you have Kc
  2. There are several options to get ETo raster. (1), as you mentioned, you could interpolate point/station data to rasters; (2) This global data is available on Google Earth Engine if you use GEE → ee.ImageCollection(“IDAHO_EPSCOR/TERRACLIMATE”), please note they labeled ref ET as pet, but the data should be ETo according to their method description.
  3. One might be more accurate than the other but the approach choice is case dependent and mainly depend on on what data you have.
2 Likes
3

In addition to what @yingjie has mentioned above (all great options), there is also a global ET0 raster (CGIAR) that is commonly used as an InVEST input. You can find the annual or monthly data on our Hub here.

3 Likes
4

@yingjie Thank you for your valuable recommendations

5

@e_pavia thank you …my worries is the resolution of CGIAR is more coarser… is it ok to use it ?

6

Considering your LULC is high resolution (2m), it may be too coarse, especially if your AOI is small, as we have used it down to 10m LULC resolution. However, it might be helpful as a starting point or to generally compare your own calculations to (based on the guidelines provided by Yingjie above).

1 Like
7

@e_pavia Ok thank very much

8

This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.