In the co-location map, however, the correlation is more visible. There is a total of 2302 observations that share the same quantile, which makes out roughly one third of the data. They are more or less evenly spread out between them, with number a little higher in the extremes. The locations in Geneva that appear in the lower quantiles, in blue, are grouped mainly in the eastern part and on the lake shore. The high values however, are shown to appear in the city center, which is situated at the tip of the lake and around the south-western part of the city, close to where the airport is located.
Discussion
The low correlation showed in the scatterplot is mainly due to the repartition of the observation. They are gathered around a middle ground and this clustering influences heavily the regression line. The slightly positive relationship however, is a good sign that we are heading in the same direction. It shows that we are in a situation where there is a trend suggesting that there is a relationship between our variables, but the scatter plot shows correlation too weak to assert it definitively.
The co-location map, however, shows a really clear separation of the quantiles between the different parts of the city. The eastern part, mainly composed of smaller, less dense villages is clearly a different entity than the densely populated and built center. The human activity, which is the underlying indicator that is measured through the noise follows this trend well. It is interesting to note the pattern of how the red squares are distributed. We can see that they often follow straight lines, indicating big roads with a lot of traffic. The roads tend to heat up and radiate a lot due to the flow of vehicles and this is shown in the data.
The data also shows that day noise is a good enough indicator of the human activity. As expected, areas with high density of population, businesses, traffic and buildings are mostly clustered at the center of the town, where the noise during the day is the highest. That leads to the conclusion that, while activity itself is hard to quantify, the noise it produces can be a good indicator of how it is laid out in an urban area.
Conclusion
Overall, the hypothesis that there is a positive correlation between noise and thermal radiation is hard to verify directly, especially when looking at all the observations at once. However, once it is analyzed spatially, it is possible to define areas in which the correlation is clear. The article shows that the thermal radiation as well as the noise rise when you are around a populated center, and that it is especially related to roads and traffic. The urban planning of the city must take this phenomenon into account, otherwise, the probability of a heat island occurring can become very realistic and the quality of life in the city will deplete accordingly.