This is a series of emails with Rich and Perrine about water yield model and the discussion is still active...any comments is welcome:
I have a quick question:
I'm a really new InVEST user and I'm trying to develop inputs for "Water yield" model to be applied in both current map (2013) and in four future scenarios maps (year 2040) and then compare their results. I already have evapotranspiration data (1950-2000) from cgiar (http://www.cgiar-csi.org/data/global-aridity-and-pet-database) and precipitation data as well to be used in current conditions 2013. However, I'm wondering if I have to create rasters for future evapotranspiration and precipitation to be able to calculate future water yield or how that works?
Sorry for email you directly but it seems no body answer my questions in the discussion list,
The "potential" evapotranspiration data from CGIAR is the same as "reference". There is often a confusion between these terms in the literature: the way we use them at NatCap is consistent with the definition from the FAO 56 document, where reference is the evaporative demand of well-irrigated grass, and "potential" evapotranspiration is the evaporative demand of another plant (i.e. reference evapotranspiration times a crop coefficient).
So the short answer is that you can use the CGIAR data for the InVEST model.
For future climate, I don't believe the CGIAR dataset provides reference ET. You can compute it based on projections of monthly temperature (e.g. WorldClim) and an approximation of ET as a function of min and max temperature (see User's guide).
Ok, I can use CGIAR in InVEST. However, do you think I can use CGIAR data for current conditions in 2013 EVEN THOUGHT this data is an average from 1950 to 2000?. In the user guide it says "Climate data (total precipitation and potential evapotranspiration)should also match the date of the land use map". I'm a little concern about the dates.
You are right about the date mismatch, it's definitely something to take it into consideration.
Fortunately, reference ET is less variable than precipitation,so using time series from a different period won't influence the model as much as it may for precipitation for example.
You could use the CGIAR data and perform a simple sensitivity analysis: e.g. increasing reference ET by 5% (you can estimate what this would mean for temperature increase with the formula for reference ET in the user's guide), and quantifying the difference it makes to model outputs.
Otherwise, you can use some temperature time series for, say, the 2000-2010 period and use the formula for reference ET in the user's guide to translate temperature into reference ET (you can interpolate data spatially if your temperature data are given for points in space, i.e. from weather stations).
Ok, I already have the CGIAR data (1950- 2000) which I'll use for the current conditions. Now, I have few more questions that I hope you can help me with:
A) To develop the future evapotranspiration for 2040 I have data on projections of monthly temperature from worldclim (average years 2041-2060) AND precipitation from Government (year 2040). However, I have three questions here:
1) It is OK to use data from these different sources and years?
2) Is this the equation you mention above (not sure since you didn't mention precipitation or extraterrestrial radiationin the last emails):
ETo=0.00013 x 0.408 x RA x (Tav +17) x (TD - 0.0123P)0.76
3) If the equation above is the one that I have to use, the problem is that I cannot find data on future extraterrestrial radiation (RA). Do you know a source for future RA and if I cannot find it what other equation can I used?
B) Biophysical table include information basically about Kc and root depth per each LULC. Since I have to evaluate current and future conditions, do you think that I have to generate two tables for current and future conditions?
C) Biophysical table I'm trying to calculate Kc for urban and roads classes. In the user guide says "built areas can be set to f*0.1 +(1-f)*0.6 where f is the fraction of impervious cover in the area...". Since my research area is in the Amazon, towns are pretty rural (mostly pervious areas), there is not information or papers about this percentage of impervious areas in this region....how do you think should I calculate this %?. Could be done based on my knowledge on the regions, like 20% of impervious cover?.
I really appreciate you comments, they are helping me a lot!