Dear All, Please can you advise as to the validation process that Dragonfly has been through to be considered robust enough to test the Urban Heat Island? Also, is it suitable to assess the value of green and blue infrastructure at a neighbourhood scale and how much of the surroundings need to be built in?
All done. Thanks for pointing it out. Any ideas on the validation question for Dragonfly? Only reason I ask is that I have realised Envi-Met has recently become a paid option and that makes me believe they have a USP they want to protect.
Like all of the insects under Ladybug Tools, Dragonfly itself is not an engine. It is merely a connection from the CAD interface to a validated, open source engine that is actually doing all of the calculation and heavy lifting.
For the case of Dragonfly, the engine that does all of the urban heat island calculation under the hood is the Urban Weather Generator (UWG) and it has a history going back ~8 years. During this time, the engine has been validated against measured urban and rural weather data four times:
- Downtown Singapore: Bruno Bueno originally built the UWG for his PhD thesis at MIT and, at the end of his thesis, he validated the engine against measurements of downtown Singapore.
- Toulouse France - Also a validation done by Bueno for his thesis.
Abhu Dhabi - A validation that was also done by Bueno for his thesis.
- Boston / Cambridge - A validation that was done by Mike Street for this thesis, in which he found what size of a representation of an urban area is needed to yield accurate results.
Many of these publications for these validation studies can be found on the Urban Microclimate Website
It is also worth noting that the engine has been enhanced by Aiko Nakano in 2015 who did an extensive sensitivity study of it. In 2016, Joseph Yang also improved the engine and added a range of building templates for his thesis. He did a validation against the previous studies to show that his contribution was an improvement.
Lastly, @SaeranVasanthakumar translated the UWG to python over the last year and matched the results against the older Matlab version that all previous contributors were using. This is how you get the python github link I cited at the very top of this response. And it is this python code that is copied to your machine when you install Dragonfly.