Urban cooling model: building intensity values are too low?

Hi everybody,

I am using urban cooling model to estimate night temperature. My case study area is quite big and the results I got from running the building intensity option are not what I expected. The userguide suggests to calculate building intensity as normalized building area/land area. My LULC is very detailed so the building intensity basically equals to the number of floors in the building. That ranges from 0 to 30 so when normalized there are mostly low values as majority of the buildings don’t go higher than 6 floors, and the high rise buildings are rather an exception. For example, the LULC class of apartment buildings is having average intensity of 0.2, other average building intensity values for LULC classes don’t really go any higher. However, the city centre is known for night heat island and is built quite densely. The models’ output of Tair is not really showing that and I was wondering if there is some way to adjust this or if the problem is rather somewhere else?

Thank you for any suggestions.


Hi Helena, thanks for posting your question. In this case it will probably help to see the results that you are looking at. Can you include screenshots with legends of Tair and other rasters that you are looking at? Please also include the logfile from your run so we can see some of the input parameters you used.

Your building intensity values sound reasonable to me. Though maybe you would benefit from a new LULC class that represents the areas of high-rise buildings? If they are currently in the same class as the lower apartment buildings?

part (1)

Hi Dave,

Thank you for your answer and suggestion. And sorry for having my reply in separated parts (new users can post only one attachment per reply).

This is the log from the building intensity settings run

InVEST-Urban-Cooling-log-2022-09-06–14_35_29.txt (37.0 KB)

part (2)

and the T_air output.

The temperature there ranges from 15.8 to 16.05 C. The hotspots seem correct (more or less), however, the temperature should range somewhere between 15.8 (rural Tref) to 19.8 (based on meteorological station data in the city centre).

part (3)

I tried to run it with the weighted factor settings and this is the output.

InVEST-Urban-Cooling-log-2022-09-06–14_51_58.txt (38.3 KB)


The range of the temperature for the weighted factors run is closer to the expected results but since the input values are for the midnight temperature, it does not make sense to use weighted factors and building intensity would be more fit.

None of the average values for building intensity of LULC classes is above 0.15. I was thinking about your suggestion to have a separate LULC category for high rise buildings but that would still leave the residential buildings (in the centre, basically the red area on the map, usually 5-6 floor buildings) with the low value for building intensity.

I think I somehow need to increase the values for building intensity, I was wondering if I could normalize (0-1) the building intensity values in biophysical table again? I am not sure if that would be a legit solution though.


Hi @ee17hd , as it turns out, the user’s guide does say that the building intensity column should be normalized between 0 and 1, so yes! I would suggest that you try it out as it is the way the model was designed to be run. Here’s the quote I’m looking at, from the Data Needs section:

The ratio of building floor area to footprint area, normalized between 0 and 1. Required if the ‘intensity’ option is selected for the Cooling Capacity Calculation Method.

Let us know how this goes!

1 Like

Hi James,

thanks you. I guess I misunderstood it in the user guide because it does not specifically say to normalize the column. So what I did was to calculate the intensity and normalized the values for each building in the map, then averaged the normalized values for LULC categories. Normalizing the column after getting the average for categories makes more sense, it seems to work correctly now.

Thank you!


Oh that’s a good distinction to make. I’ll update the user’s guide to clarify this. Thanks for letting us know!

1 Like