Urban Cooling: valuation with building footprints

[davemfish]

I'm looking at the spreadsheet of invest_model_results.csv (see #80 ) and at https://github.com/chrisnootenboom/urban-workflow/blob/master/naturban/valuation/ucm.py

And the valuation seems to hinge on counting heating & cooling degree-days. Though I can't quite figure out how those could be calculated when the model does not return a timeseries of temperature.

[chrisnootenboom]

The methods are as follows:

• Run InVEST monthly based on the average monthly baseline air temperature extracted from gridded global temperature datasets (http://climate.geog.udel.edu/~climate/html_pages/Global2017/README.GlobalTsT2017.html)
• Convert spatialized monthly air temp (from InVEST) into spatialize Cooling or Heating Degree Days (and then kWh) using regression equations from Roxom et al. 2020 (https://doi.org/10.1016/j.uclim.2019.100546)
• Calculate mortality risk based on methods from Guo et al. 2014 (https://doi.org/10.1097%2FEDE.0000000000000165) and the “comfort” temperature of the city in one of the tables in this paper

[davemfish]

Ahhh, monthly runs, cool! Thanks for the data sources!

[davemfish]

After discussing, we've decided that the very first step for displaying results can just include reporting the estimated air temp differences across scenarios, just modeling the daytime mean air temperature for the month of August.

There is a valuation method, including calculating heating & cooling degree-days, just with a single-month model. We can implement that later on, not for the immediate milestone. It involves a linear regression-based estimate of degree-days for the month that was modeled.

[davemfish]

On hold for now, though it does seem likely we want to go down the path of valuation described above.

[davemfish]

#120 includes the valuation methods

[davemfish]

On hold for a decision about #126

[davemfish]

I have extracted the building footprint data from Microsoft's data store. https://planetarycomputer.microsoft.com/dataset/ms-buildings#Example-Notebook

For the valuation, we also need energy-consumption parameters. Chris and I discussed this on Slack and he shared a table with those parameters, keyed by lucode. In theory, these parameters could vary by "building type" instead of by LULC, but our buildings database does not have a "type" variable. And right now the valuation script Chris prepared expects these energy parameters to vary by 'lucode'. So we'll run with that for now. (Also right now the energy table has constant values for all parameters across all lucodes. But in theory we could improve that parameterization later).

I still need to figure out the appropriate way to use the building footprint data in the context of this valuation workflow. And the building data also includes a mean-height field, so really we have footprints and heights of all(?) buildings.

[davemfish]

Why aren't we using the valuation component built-in to the invest urban cooling model?

Chris's custom valuation method calculates cooling-degree-days, and then applies the energy consumption & cost parameters. That is more sophisticated than the invest method, which does not consider degree-days and might require some different parameters than we have.

[davemfish]

It's not clear to me if/how to incorporate building footprint data into this valuation. The valuation parameter kwh_per_cdd is in a table where it can vary by building_type (though all the values are constant across building_type right now).

How can building footprint/height data influence the energy cost parameters? Should kwh_per_cdd vary by building volume? If so, how will we parameterize that? What size building is assumed for the 1.8 kwh_per_cdd value that we are using right now?