top of page

Voronoi or Thiessen?

Andres mentioning everyone's favourite topic in the urban climate lab
The audience for the TEC meeting raptly listening to Dr. Moral

Every Friday during the semester, the Tropical Environmental Change Research Group (TEC) holds a lunchtime meeting in which the wide diaspora of physical geographers in the department (faculty, postdocs and grad students) show up to (a.) listen to the latest in TEC-themed investigations, and/or (b.) see what everyone else packs (or tah-pows in the local lingo) for lunch. It's always a fun gathering in which serious research is discussed in an informal, supportive atmosphere.

Last Friday, our postdoc Andres Simón-Moral presented updates of his ongoing research; specifically, his presentation was on a GIS (Geographical Information Systems)-based urban classification of Singapore towards the eventual application in a coupled urban canopy-regional climate model. A major research thread in urban climate is how we can get a good representation of the city surface (that is, how we parameterise the physical structure of real-life roads, walls, roofs etc. as representative numbers such as the aspect ratio of urban canyons). These numbers can then be used by numerical models to calculate physical processes.

An obvious challenge is to determine if existing methods of compiling these spatial units of cities can be accurately - and widely - applied in conjunction with the finer spatial resolution of urban climate models; About 10 years ago, these models "saw" the urban surface at 2+ km scales. These days, the models are much more computationally powerful - Andres' model (and others) can represent complex surfaces at 300 m horizontal resolutions.

A common approach is to use a weighted average of urban parameters within the 300 m "square" grid to parameterise urban areas. That method may work in cities where regular landuse patterns dominate (e.g. large, homogeneous suburbs in parts of the US), but it's not really applicable in tropical cities like Singapore. Thus, it was fascinating to see how Andres uses a GIS-based approach to derive Voronoi diagrams of Singapore's land cover over various distance resolutions (e.g. does aspect ratio change if the polygon increases/decreases in size?) and over different wind directions (e.g. does the aspect ratio change if predominant wind direction changes from NE to SW?). Answers to these questions can be seen when Dr. Moral gets his research published!

I greatly enjoyed Andres' talk - others also did given the wide discussion among the TEC group that followed. One thing that stood out for me once the meeting concluded was how different disciplines have field-specific terms that are often confusing to other folks, and in this case it was how folks who know the term "Voronoi" were confused when I mentioned "Thiessen" (which was how I was taught to describe these polygons as they are applied in hydrology and in meteorology).

A challenge in inter-disciplinary research work is that this confusion in terminology often arises, and forums such as these lunch meetings are great places for Earth scientists to understand how the old wine of common concepts can often be poured into different bottles by others.

80 views0 comments
bottom of page