Discussion

The results obtained in the previous part are not very glaring. First, we can observe that 176 thermal values without green roof are higher than the maximum of a green roof case. However, we have to be careful, because more data without green roofs have been observed. If we consider that the global repartition of thermal value is a random variable, it can explain this higher range. To go deeper in detail, we need to take attention to the figure 2. We can observe that, for the four first figures, no real distinction can be made between green roofs and thermal response for small thermal values. It may significate that:
1) The response is not visible (zero hypothesis);
2) The thermal resolution is too rough;
3) The grid resolution (20m) is too rough;
4) The response is only visible for high values; otherwise they are too diluted with other parameters.
From the author’s point of view, the fourth option is the more relevant. As previously said, if we look at the last subgraph of figure 2, we can observe a certain tendency of values to be smaller with the presence of a green roof than without. A supposition that can be made is that green roofs have an impact of the thermal response in Geneva, and then potentially on a potential urban heat island but the impact is only measurable when the reflectance effect is noticeable enough.
In summary, the results seem valid, and express the idea that green roofs have a significant effect against UHI. To control or improve this supposition, it would be necessary to test a higher number of green roofs, in a more dense city, with higher buildings and higher temperature values (more UHI effect)[5]. Increasing the resolution of the thermal layer and the grid may also be a good thing, but it should not change the general conclusions (at least, for Geneva).