Understanding Urban Cooling - Energy Consumption Input

Hello there!

I do have trouble understanding how the urban cooling valuation model concerning the Energy_Consumption works and what exactly the input should be.

The input needed is energy consumption per building type, in kW/degC.

I get that in order to come to a value that has the unit of kW/deg_C , you first go look into the Santamouris & al. paper in order to find out the plus in electricity need per plus of degree celsius ambient temperature for your region of interest.

Let’s say that would be a plus of 10%.

Now that we know the plus in energy need, we need the energy consumption per building type.

So first of all, I am wondering why the documentation says kW and not kWh? Considering kW is power while kWh is energy?

Now, altough it is possible to find data about average energy need per building type at least for some types (such as residential), it is quite imprecise and really hard to find when it comes to industry, where floor space varies greatly, so how can we just have one single number for all the buildings of a giant category such as f.ex. industry? So I suppose kWh/m^2/building type is the most sensible input? But then again, the calculation would be missing every floor space that is not ground floor, considering it only works with the footprint…

Anyway, let’s suppose now that residential buildings use 200 kWh/m^2 and industry uses 300 kWh/m^2.

Would this be what the type and consumption columns should look like?

Type____________ Consumption

Residential_______ 0.1x200 kWh/deg_C

Industry_________0.1x300 kWh/deg_C

Or am I totally miss understanding something?
As I do understand that the model calculates energy use per building on a pixel basis (Therefore needs a precise footprint shp. file) but I am not sure if the input should be energy use per square meter, or something different, as it is never really specified in the documentation.

Thank you for your help.

Hi @Michael, thanks for posting! I’ve reached out to our science staff for the model who may be able to provide some insight here.



Thank you, that would help a lot.

For now, I thought maybe a better way to get a single energy use number / building type is looking at the total energy need of a sector (f.ex. offices) for a region and then dividing that kWh number trough the number of buildings tagged as office in that region (f.ex. in Open Street map).
This would give then a kWh/building number…?

Hopefully somebody can elaborate on these thoughts.

Thank you.

Hi @Michael,

Thank you for posting these questions about the Energy Consumption Table that is needed as an input to run the valuation component of the Urban Cooling InVEST model.

You correctly point out that consumption values ought to be in units of kWh/degC (as kWh is a measure of energy and kW measures power). We have now made this change in the User Guide based on your observation and apologize for any confusion the incorrectly listed unit caused you. Users like yourself are essential to our software development and we always appreciate feedback to help us improve models and documentation.

This model is not sophisticated enough to distinguish energy consumption levels between different floors of a building, and instead applies a single value to all buildings of each specific type. The values in your table should not be per m^2 necessarily, but simply kWh/degC; the amount of energy needed to raise the temperate of X building type by one degC.

Please see the image below of an example table. Notice the order of magnitude is in the 100s of kWh/degC.

I hope this is helpful and that I’ve answered your questions. If not, please do reply with your unresolved concerns. Again, thanks for bringing the User Guide’s incorrectly listed unit to our attention and accept our apologies for the confusion.

GIS Analyst | Natural Capital Project

Hi Jesse,

thanks a lot for the answer, and no worries, the kWh error did not cause me any concerns.

Thank you for clearing my question up and providing an example table.
I guess it is now clear that kWh/m^2 is not the way to go and, if these numbers are used, they should be upscaled to the building level. By doing so with the numbers I gave in the above example, you do arrive in the 150’ish kWh/*C range, which seems fitting considering your table.


Hi there @jesseG. I am having a similar problem to @Michael. We are only able to obtain values per m^2. For our local region we found that a commercial building uses 0.07 kWh/m^2 for an increase in 1 degree Celcius, but how do we convert this to a value per building type? The statement “the amount of energy needed to raise the temperate of X building type by one degC,” implies that we somehow need to derive a mean value. Any recommendations on how to go about this would be helpful.

Furthermore a couple of other questions:

  1. For the example data, how were the values (in the magnitude of 100s) derived?
  2. On what time scale are the energy savings? Is it per annum, month, or day? There is no indication of this.

Our study area has relatively low use of A/C that is generally confined to 4 months of the year with very low residential use.

I assume the model works on the area of each building as the output values differ per polygon but this begs the question of how to derive the value for each building type - do we need to use the average area of buildings in our study area, for example? For reference, the shapefile I have has over 36 000 polygons of 6 building types.

Hi @JoshAnchor,

Thanks for your questions about the Urban Cooling model. Frankly, I have minimal experience running this model, so I have reached out to our Urban team to request a more detailed reply to you from @chris. Meanwhile, I will do my best to try to provide some basic clarity.

Since you have energy values in kWh/m^2 needed to raise each building type’s temperature by 1 degC, I would multiply those values by the average square meter floor area of all the buildings of that same type to derive the mean kWh/degC for each type. So, “Commercial” being one of your 6 types, for example,
0.07 kWh/m^2 * (mean Commercial building floor area) = mean kWh/degC for Commercial buildings.

I am not sure how the values in the example data were derived, but I can say that they were taken from an example in an online InVEST training course. I shared that table with Michael as an example of orders of magnitude for ‘Consumption’, but I would not take the exact values too literally.

The energy savings (in kWh, and $ or USD) are NOT measured on any particular timescale. Those values are the savings realized from the reduced energy (avoided consumption) that would otherwise be required to change a building’s temperature by 1 degC, or savings per degC. How long it would take for that 1 degC temperature change to take place depends on the duration over which that required energy (in kWh) was applied.

I hope these answers are at least somewhat useful for you. Hopefully you will hear back from Chris soon whom has much greater experience with this model, but I wanted to provide a reply sooner so that perfection would not become the enemy of the good here.

Be well,

1 Like

Hi @JoshAnchor,

As a small add-on to Jesse’s answer, the calculation uses the building classes in your land cover map, not the building footprint vector. The vector is only used in the ‘building intensity’ calculation. The table with energy consumption links to the land cover raster. If you would want to use the 6 building types in your vector you would need to rasterize them and merge them with your land cover raster (assuming that raster does not already have these building classes).

Best, Roy

Hi @royremme. What you are saying doesn’t seem to be implied by the model interface.
I am speaking about the Valuation Model under which the Building Footprints Vector (shapefile) looks like it is required.


Sorry @JoshAnchor, my bad, I was mixing the two components up. Yes, the vector is used for the valuation step, the raster for the building intensity calculation.


Wow ! that is super usefull ROY, I was in the process of creating Energy Star based building types for the energy use piece. Land use identifiers are not very detailed ESP for urban land cover !

We are in the process of using an Energy Star based calculation to create detailed building types in the hopes that we can consider changes in building type and associated impacts. This is only good for cities where detailed building information is publicly available.

from appendix B of “Grading buildings on energy performance using city benchmarking data” Sokratis Papadopoulos, Constantine E. Kontokosta⁎
“The EnergyStar grading method for multifamily buildings consists of a linear regression model, trained on 322 sample buildings across the U.S. The model specification is as follows:
EUI = 140.8 + 52.57∗cUnitDensity + 24.45∗cBedroomPerUnit−18.76∗LowRise + 0.009617∗cHDD + 0.01616∗cCDD”