I am trying to use the urban cooling model in a part of my dissertation and I am currently refining the urban cooling parameters. I’m a little bit skeptical about the “Air Temperature Maximum Blending Distance” parameter that the model requieres. I’ve look out for some questions about air mixing here in the forum and I found a topic that recommended reading Shatz & Kucharik 2015 and Shatz & Kucharik 2014 but I’m still having difficulties. I was wondering, as it is a very uncertain parameter, if the developers or someone who has used the model could make any recommendations on how to extrapolate this air mixing parameter from some data that is easier to find in the literature, such as the wind speed for example. Any idea will also helpful!
Thanks for the question! This is indeed a point of uncertainty in the modeling. Typically we default to 600m from the Schatz and Kucharik papers, and I would also point you to a recent publication that uses a different model but provides a similar distance parameter linking land use to air temperature change: Lonsdorf et al. 2021 https://doi.org/10.1016/j.landurbplan.2020.104022.
I like the idea of tying the parameter to a prevailing wind speed, which to my knowledge has not been done before! As it stands, the parameter is a simplification of air blending dynamics since it does not take into account wind speed, direction, or any actual wind modeling–this allows the model to function on relatively simple data inputs. If you have access to ground-truthed air temperature data, you can train some of these inputs (including the air mixing parameter) using a script built by Marti Bosch, a PhD student at EPFL in Switzerland, which is currently under review for publication. Here’s a link to the github repository:
We have been using this script to refine the model in Minneapolis, Minnesota and Paris, France, with air blending distance being a primary driver of model accuracy. I will try to look into any correlations with typical wind speeds in those cities, as that relationship is intriguing, but for now the above publications and model training code are all I can offer on this specific parameter.
Hi guys! Sorry for the delay in replying, but life is happening and it hasn’t been easy, but anyway… thanks for the suggestion! It was very very useful.
Hi just picking up this thread again as now the paper on calibrating parameters by Marti Bosch has been published. In this research the values that fit the temperature monitoring stations the most were cooling distance of 89.21m and a temperature mixing radius of 239m. They do mention that this is probably influenced by the relief of the city. As I’m looking into a mountainous city would you recommend deviating from the InVEST User Guide recommendation and applying the ones found by Bosch et al. (2021)? it would be great to calibrate the model as has been done in other papers but I won’t have time for the purpose of my research timeline so I’d appreciate any insights / recommendations on this. Thank you!