ISSS608 2017-18 T3 Assign Jiang Yilin Answers

Rose-crested Blue Pipit: Where are you?

Overview

Methodology

Answers

Conclusions

Q1. Is there any trend or anomaly?

Starting from the first question raised before (whether the population of rose pipits is declining), we can derive the conclusion from Viz.1 that the Blue pipit's population is declining in fact.



From the graph,it's clear that the bird reaches its population peak in 2015, and the population drops since then.

Now, what about other species' population?



As shown by Viz.2, most of other species also have population decline since 2015 and 2016, some since 2014. For specific info please check the published story.

Geographic Visualization

Geographical Visualization for all birds, Rose-crested blue pipits, Green-tipped Scarlet Pipit, and Eastern Corn Skeet. The last two species are two most active species besides Rose pipits.

From All birds distribution, there used to be birds active around the dumping site in 2012-2014, the peak is in 2013. However, since 2015, birds become very scarce around the site. Another point is that the density dropped from e-4 to e-5 in 2017, which resonates with Viz.2.



The movement of rose pipit is even more obvious than the all birds distribution. In 2012-2014, there were many rose pipits active around the site. There are two clusters of them, one around the dumping site, and another cluster below that. But from 2015, there is only one cluster, the cluster around dumping site disappeared.

....... .......

Green pipits(left) and Eastern Corn Skeet(right) also shows some signs of getting away from the site in 2016 and 2017, but since the population is small and changes are not very obvious, here we cannot make the conclusion that the dumping site is affecting all birds. When exploring other reasons that may make rose pipit move away from the dumping site, I found an interesting point.
Click on the image and refer to the following dashboard and find it.



If you select all years, select "in" value for Pollution Circle filter, and change radius from 5 to 40, you'll find that when the circle's radius is smaller than 30, most records are in winter. While when the circle grows larger, the highest proportion is spring, which is same for the records outside the circle.

One inference for this finding is that in winter, the contaminant spreads much lower than in summer because of the temperature. So birds tend to be around the dumping site in winter more than summer(the lowest proportion when radius <30).

Apart from last finding, inside each season, 2013 and 2014 have the highest proportion, and 2017 has the lowest. That proves that the population decline around the dumping site is not affected by season. Rather, it is a trend that has to be explained by other reason such as pollution.

Q2. Are the sounds that Kasio company provided really Rose-crested Blue Pipits?

Using librosa package from python, I generated visualized audio waves for the 15 audios given by Kasios company(test file) and the audio package provided by the university(sample file).
The university bird audio package has 2081 files. Since birds have different sounds for call and song, randomly select one call audio and one song audio for each species. All sample file selected are of A or at least B quality, to avoid noise interference and make sure of the waveplot accuracy.

Since all the test audios has range inside [-0.5,0.5], and so is blue pipits sample audio(both call and song), I set the sample file to the same range. In this way, it automatically excludes many audios when matching them against the test audios.

Test Audio	Sample Audio

Of all the test audios, 13 is most likely to be Rose pipit singing, and 6 is likely to be Rose pipit calling. Other audios show no similarity with either rose pipit singing or calling.

Above is the plot for location of 15 birds collected by Kasios company. From the audio visualization. Only 6 and 13 are audios of Rose Pipit. And the singing one(13) is further away from the dumping site than the calling one(6).