Missing agriculture directory?

[GeraldCNelson]

I was looking for a folder called

• agriculture: Logic for describing agricultural response functions.

but couldn't find it.

I currently live in Mesa County, Colorado. If I interpret the results in the csv file county_damages_by_sector, Mesa County agriculture will benefit (value of -4.17). But we have mostly irrigated crop agriculture, especially orchard crops, where the expected temperature increase by 2100 will stress crops and reduce water availability in the Colorado River. So gains seem unlikely. At any rate, I'd like to see the logic.

[delgadom]

We only examine impacts to maize, wheat, soy, and cotton. The statement of methods goes into the methodology for each impact category in much more detail. The % change in yields value is a weighted-average impact for those four crops.

If there is very little production of those crops in your county, then the agricultural impacts in your county will factor very little in our estimate of total damages in your county, but the agricultural impact we calculate will not be representative of the effect on total agricultural production for your county.

Furthermore, we don't model water resources explicitly; only indirectly through the historical statistical relationships between temperature, precipitation, and yields for those crops. So if fundamental changes in the water supply will play heavily in future yields, we will miss those changes.

[jrising]

I looked at Mesa County, and that -4.17% is actually the predicted change in yields; that is, we are predicting losses, not gains. Sorry for the confusion-- I am going to mention that this is not clear to the person who put that file together.

Thanks for pointing out the mis-match between the README and the actual directories! This ended up just being a single file, so we removed the directory and moved it to impacts/agriculture.py. That file has the the logic that calculates impacts for single crops, combinations of crops, and the calculation of "growing degree days". Those are the key functions, and then the "controller" function make_agriculture in controller/acra.py. The weighting that @delgadom mentioned is just the average land area for each crop, which are in iam/cropdata/*-planted.csv.

For Mesa, our records show only corn and wheat planted, amongst the four crops we looked at. So the result is an average of those two.

The models we have for the four crops that @delgadom mentioned include the effects of temperature, precipitation, CO2, and (for corn and soy) a east-west distinction to account for irrigation. Orchard crops are more complicated than these field crops, and the science on them is much less developed, which is why we did not include them in our analysis.

[delgadom]

Oh good catch @jrising - thanks for tracking that down! And @GeraldCNelson - here is the link to our statement of methods. There is a lot in here that didn't make it into the body of the paper!

http://science.sciencemag.org/content/suppl/2017/06/28/356.6345.1362.DC1