Urban Heat Islands in the GTA

In 2016, Toronto recorded the hottest summer with a daily mean of 23.07 °C. If you've walked a little bit around town, you certainly noticed a difference in temperature depending where you are in the city. In fact, it is not uncommon to observe a variation of 5 degrees in areas just a few blocks apart! This thermal fluctuation where urbanized areas experience higher temperatures than their rural counterparts is called urban heat islands. 

Inspired by the excellent work of Dr Steif, I decided to do a similar analysis for my hometown, Toronto. I wanted to locate urban heat islands throughout the GTA and at the same time find places that can provide a respite from heat waves. To do this, I calculated the land surface temperature (LST) and the normalized difference vegetation index (NDVI) by using a satellite imagery (i'll talk about that later). Speaking of satellite images, available resources for Canada is not as abundant compared to our southern neighbours. Luckily, I discovered a great imagery browsing platform called Libra and was able to find the right Landsat data for this project. The satellite imagery used for this analysis was taken on August 24, 2016 at 4:03PM.



Normalized Difference Vegetation Index

The first step is to calculate the Normalized Difference Vegetation Index (NDVI). NDVI is an indicator displaying the density of vegetation based on a satellite image (LandSat 8 for this analysis). By comparing distinct colours (wavelengths) and near-infrared sunlight reflected by vegetation, we can calculate the NDVI with the following formula:

Where NIR = Near Infrared
          VIS = Visible light

NDVI values always range between -1 and 1, but typically falls between 0 and 0.5 in an urbanized area. Higher values indicate denser greenery while values around 0 and below indicate places with no vegetation such as sand, rocks, snow, or in our case paved surfaces and construction materials.

Here's what the NDVI look like for Toronto. Not surprisingly, NDVI index is higher in parks, wealthy neighbourhoods like Rosedale and lower in highly commercial and industrial areas. 


Land Surface Temperature

Now that we have our NDVI, we can proceed in calculating the land surface temperature (LST). This task basically involve two steps: 
1- Converting digital numbers to radiance
2- Converting radiance to kelvin
To learn more about this, you can go check out this video tutorial for more in-depth explanations.

Here is the LST in Toronto for August 24, 2016 at 4:03PM. As you can see, there are noticeable differences in the city. First, what surprised me is how low the LST in downtown Toronto is. This can be explained by the skyscrapers casting large shadows and thus cooling the environment. I suspect the results might be different if the calculations were made in the evening. Construction materials store heat throughout the day and continue releasing it even after sunset.  


A Comparison

A good example of how much NDVI and LST can varied based on the built environment is the two adjacent neighbourhoods Parkdale and Liberty Village. The differences from the west side and east side of Dufferin Street are pretty striking, but not surprising. With numerous mature trees, Parkdale achieve a high NDVI index of 0.25 and a LST of 27.5 °C on average. On the other hand, Liberty Village is more characterized by large paved surfaces (a lot of parking lots) and generally have less trees to provide a canopy cover and shade area. Approximately, the NDVI of Liberty Village barely reach 0.05 and has a LST of 31.5 °C. 


Crossing NDVI and UHI

Now that we have an idea on the NDVI and LST, we can cross them and see if there's a link between them. On the day and hour the satellite imagery was taken, the mean temperature in Toronto was 26.9 °C. To show more clearly the phenomenon of urban heat islands, I identified areas with temperatures of 31 °C and higher. Then I superposed them on the NDVI map and added Toronto's open spaces. Here's what it gives.

This visualization can help locate problematic areas and improve public policy where needed. For instance, city officials can incorporate mitigation strategies in their official plans or zoning by-laws to require new constructions to use reflective surfaces or determine where specifically to plant new trees.  

Finally and most importantly, this topic is also a question of public health and social justice. In The Star, an article talks about how less fortunates tend to live in low "leafy environments", have less financial means to cool themselves (e.g. buying an air conditioner), and consequently are more vulnerable to heat. 

In 2010, Clean Air Partnership with the collaboration of Toronto Public Health published a heat vulnerability map. This dynamic map locates which part of the city are most vulnerable to heat based on sociodemographic data such as senior and families with low income. It's a great visualization to identify which census tract we should direct our attention to. The NDVI and LST map can complement it by pinpointing with more precision which city block are considered problematic.