ISSS608 2018-19 T1 Assign Wong Yam Yip Task 2 Insights
What's Suffocating Sofia?
|
|
|
|
|
|
`
The above animation demonstrates the air pollution concentration measurements in different areas of the city over time
Contents
Civilian Sensor Coverage in Sofia, Bulgaria
Overall the sensors are distributed over the counties and there appears to be at least one sensor in each state of Bulgaria. Nonetheless, most of them congregate in Sofia, while a significant amount can also be found in Plovdiv (green), which is the second largest city in Bulgaria.
For this analysis, we would only like to define the boundaries of Sofia city by Longitude 23.191 – 23.457 and Latitude 42.603 – 42.788. This will leave us 1,949,114 readings from 647 sensors (above right). The sensors are generally well distributed across the city. There are more sensors towards the center of the city, where we can see more overlapping sensor marks.
Performance and Operation of Sensors
As seen above, some of the readings for temperature, humidity and pressure are skewed in an irrational way. For example, there are readings of 64°C and -5573°C, as well as pressure as high as 176,159 hPs, which are not humanly livable conditions. Similarly, negative humidity % not possible! Therefore, the accuracy and operability of these sensor readings are questionable.
Exploring PM10 and PM2.5 readings, the distributions are also highly skewed, reaching as high as 2000 and 1000 for PM10 and PM2.5 respectively. IF we assuming that the readings are accurate, then there may be a limit to the capabilities of these sensors.
Air Pollution Readings and Distribution
On first look, it seems that the highest readings are found in the areas highlighted by annotations, where the PM10 and PM2.5 concentration is the highest, 2 of which are in the outskirts of the city with PM10 readings 641 µg/m³ and 574.3 µg/m³ respectively. Seen in the below image, a series of maximum PM10 and PM2.5 readings were recorded for these locations, thus pushing up the average. It is possible that the readings are detecting very polluted air, reaching the sensor limits of PM10 and PM2.5. However, if we look at the vicinity of these sensors, they appeared to be the only one getting such high readings. It is also questionable that air pollution is contained within a specific geographical location. Thus, it may also be possible that these sensors are behaving abnormally and giving inaccurate readings.
To further our analysis on other areas, we exclude those 3 earlier sensors and now we can see other areas with relative higher readings and mostly in the city center (green). The highest average reading detected by the sensors here is PM10: 265.7 µg/m³ and PM 2.5: 166 µg/m³.
Air Pollution over Time
Looking at the readings for Sofia city over time, there are trends of higher concentration readings during end/start of the year. Highest average PM10 and PM2.5 readings are 456.6 µg/m³ and 255.2 µg/m³ respectively on the same day of 8 Jan 2018. The average reading for Jan 2018 was also the highest at 101.2 µg/m³ and 51.6 µg/m³. As there is insufficient data, it is not possible ascertain that this is the trend for other years, but these trends are in line with those of the official data.
Relationship between Air Pollution and other Civilian readings
Finally, we attempt to explore the correlation of PM10 and PM2.5 readings with the collected environment variables like Pressure, Humidity, Temperature and Latitude/Longitude. As we can see, it appears that there is little/no correlation between air pollution and the environmental measurements of the civilian sensors. However, we need to keep in mind that some of these measurements may not be reliable. This correlation analysis will be explored again with the Meteorology data in the next section.
Reference
The Tableau Workbook to the above images can be found here
Banner image credit to: MarcusObal
