A New Perspective

Due to the COVID-19 pandemic, protected areas and other forms of wilderness areas (e.g., arboretums, beaches, parks, marine monuments) have been closed worldwide. Your challenge is to lead the effort to examine any potential impacts of reduced human traffic in such local protected natural environments.

The HummingWaves Project

Summary

Investigate links between maritime traffic, noise pollution and aquatic life proliferation using aerial images, open data and AI.

How We Addressed This Challenge

The COVID-19 pandemic presented a unique opportunity to measure the aquatic ecosystem's response to diminished human activity. We embarked on a challenge to link the data that will allow us infer new knowledge about the impact of anthropogenic ocean noise pollution on the behavior of cetaceans.

How We Developed This Project

We were inspired by the observation that there's less noise around us, and a thought that it is probably not only true on land, but also under the sea. We began with searching for existing ways to determine the state of cetacean populations - as expected, successful attempts at AI recognition of whale appearances from aerial data are known in literature [2], [13]. One project has been even funded by ESA [1]. We attempted to replicate these results using Tensorflow detection model zoo [18] model of a Convolutional Neural Network and a Colab environment - project page containing more details and the repository are linked in the subsequent sections. One particular finding was that openly available satellite photos rarely reach the required resolution (~1m per pixel) - hence, to apply the given method to a broad range of territories, cooperation with commercial programmes (like WorldView3 by DigitalGlobe) or aerial photography providers would be required.

At the same time we looked at the maritime traffic. While AIS data is abundant online, many of the sources do not meet the criteria of openness while some others not yet provide the timeframes required for accurate determination of the changes as results of restrictions caused by the pandemic.

One of the sources that presents the ongoing changes in the maritime traffic can be found at [12]. Comparison of vessels and route density in March 2020 with the same period last year shows a significant reduction in human activity on the European waters. The same source reports a drop in fishing activities by nearly 40% on Western Mediterrean Sea, compared to the previous year.

We verified our findings with a marine biology SME, who confirmed our suspicions that many data gathering activities are now suspended because of the pandemic. We expect this to influence some of the popular statistics like the number recorded whale sightings, hence the increased need of an automated way of examining the status of marine populations.

Our conclusion is that machine learning algorithms have the potential to help us better understand the complex relation between human noise and changes in cetacean behavior. A multi-input solution, taking into account not only aerial imaging data, but also acoustic data and GPS tracking, may improve the accuracy even further.

Data & Resources

ESA resources:

[1] https://business.esa.int/projects/spacewhale

[2] A. Borowicz et al., Aerial-trained deep learning networks for surveying cetaceans from satellite imagery*: https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0212532

COVID-19 impact:

[3] https://www.aquablog.ca/2020/04/quiet-oceans-has-the-covid-19-crisis-reduced-noise-in-whale-habitats/

[4] https://iqoe.org/articles/ocean-sound-covid-19-era

Noise affecting marine mammals:

[5] https://dosits.org/tutorials/effects-introduction/behavioral-changes/

[6] R.M. Rolland et al., Evidence that ship noise increases stress in right whales: https://royalsocietypublishing.org/doi/full/10.1098/rspb.2011.2429

[7] https://noaa.maps.arcgis.com/apps/Cascade/index.html?appid=c653c78262a7487da42149ebc86f80c2

[8] S. Veirs et al., Ship noise extends to frequencies used for echolocation by endangered killer whales: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4800784/

[9] https://www.youtube.com/watch?v=GLlNXd1dJkA

Maritime traffic:

[10] Vessel ID System: https://www.nasa.gov/mission_pages/station/research/news/b4h-3rd/eo-tracking-global-marine-traffic

[11] https://oceanservice.noaa.gov/facts/ocean-noise.html

[12] https://www.emodnet-humanactivities.eu/blog/?p=1258 - COVID-19 impact on fishing

Marine mammal presence detection:

[13] E. Guirado et al., Whale counting in satellite and aerial images with deep learning: https://www.nature.com/articles/s41598-019-50795-9

[14] https://www.kaggle.com/c/noaa-right-whale-recognition/data

[15] https://github.com/Microsoft/belugasounds

[16] https://www.fisheries.noaa.gov/west-coast/marine-mammal-protection/whalewatch

[17] https://www.photolib.noaa.gov/Collections

[18] https://github.com/tensorflow/models/blob/master/research/object_detection/g3doc/detection_model_zoo.md

* the paper is an outcome of the SPACE WHALE project funded by ESA.

Tags
#noisepollution #covid19 #marinetraffic #silentseas #oshhean #cetaceans #whales #spacewhales #whalesinspace
Global Judging
This project was submitted for consideration during the Space Apps Global Judging process.