“When you can’t change the direction of the wind - adjust your sails.” – H. Jackson Brown, Jr.

Project Info

Swainson’s Hawk has one of the longest migration routes in the world. My initial interest in the bird came from seeing the hawks migrating in and around my hometown of San Diego, CA. Yet in terms of migratory patterns, Swainson’s Hawk has one of the longest migrations in terms of distance covered, spanning from the North Western United States down to South Eastern Argentina1. Not only is the distance covered by their migration interesting, but also the indicator for when the birds decide to migrate. The hawks begin their migration in the Fall2. Specifically, their migration is theorized to begin when the surface wind blows south in their region3. I found the specificity of this interesting and wanted to combine migratory and surface wind vector data to see this in action.

Relevance

Aside from having an interesting migratory pattern, Swainson’s Hawk has been considered an endangered species4. The nature of their long distance migration makes their ecosystem much more fragile, as changes in any region along their migratory path carry implications to the survivability of the hawk. For example, pesticides used by farmers in Argentina to control local pests5 had resulted in a significant number of these hawks being poisoned. It is important to understand the fragility of the ecosystems that live around us. The Swainson’s Hawk is particularly a good example of how far-reaching ecosystems are, spanning beyond our own locale. If anything, it is important to have a heightened awareness of the implications of our actions. So, there is an importance to observing migratory patterns such as the Swainson’s Hawk.

Data Sources

Migration Data

Firstly, my migratory data came from MoveBank6, an opensource platform that allows researchers to archive and analyze animal movement data. From MoveBank, I was able to successfully identify the Swainson’s Hawk dataset. Originally, the dataset was composed for a study7 of “migration routes, length of migration, and duration of migration” for Swainson’s Hawks in 1998. Each bird was radio tracked using the ARGOS satellite system8, a satellite telemetry service for “scientific and environmental applications.” The dataset identifies each individual bird with a unique individual indentifier (i.e SW1, SW2, etc.), and provides time-stamped coordinates tracking the hawk’s movement.

With this datasource, initial exploration of the data allowed me to see the movement of a hawk overtime by seeing each given bird’s location over time. So, my goal was to aggregate my data by individual identifier (SW1, SW2, …) and map their migratory path over time using the Python9 programming language. I needed to create shapes beyond the POINT shapes I currently had. So, I found LineStrings. In a GeoJSON10 object, LineStrings are shape objects constructed by a series of coordinate pairs. The LineString will then create line shapes between the points. Initially, I had made LineStrings mapping out the paths of each individual bird using Shapely11. However, I realized that for animation’s sake, it would be more helpful for the user to be able to see the path of the bird develop overtime. So, I rewrote my script to create LineStrings that are timestamped.

To visualize this data, I took advantage of an open source tool called Kepler-gl12, a tool for geospatial data analysis that I think particularly excels at visualizing time series data. I was able to simply import my dataset in GeoJSON format to the WebGL component of Kepler-gl. There, I was able to use a timestamp filter to create the animation of the bird migration over time.

Migration Routes

Wind Vector Data

For my wind vector data, the biggest challenge was finding surface wind data that covered the time period of my migration dataset; 1995 - 1998. I found the wind vector data, titled NCEP Reanalysis provided by NOAA/OAR/ESRL PSL13, which had surface wind vector data from 1948 to the present. As such, I downloaded two datasets filtered from 1995-1998. The first is called Zonal Velocity (u) which is horizontal wind velocity at coordinate points. The second dataset is called Meridional Velocity (v), which is vertical wind14. With this, I was able to create vectors in Panoply15 by loading both datasets, then creating a merged plot which created a wind vector (as an arrow).

So, each day has its own array of Meridional and Zonal datasets, with vectors calculated between both arrays. Below, you are able to see the average direction of the wind for each day between 1996-07-01 to 1997-07-07.

Author: Andrew Caietti. Last edited: 2020-11-18.

Attributions

  1. Global Raptor Information Network. 2020. Species account: Swainson’s Hawk Buteo swainsoni. Downloaded from http://www.globalraptors.org on 18 Nov. 2020↩︎

  2. Fuller, M.R., Seegar, W.S., Schueck, L.S., 1998. Routes and Travel Rates of Migrating Peregrine Falcons Falco peregrinus and↩︎

  3. Hawk Migration Association of North America. Copyright(c) 2019 HMANA. Accessed 2020-10-30. Online: https://www.hmana.org/new-to-hawkwatching/↩︎

  4. Wildlife Heritage Foundation. Accessed 2020-10-28. Online: https://www.wildlifeheritage.org/gallery/swainsons-hawk/↩︎

  5. Goldstein, M.I., T.E. Lacher Jr, B. Woodbridge, M.J. Bechard, S.B. Canavelli, M.E. Zaccagnini, G.P. Cobb, E.J. Scollon, R. Tribolet, and M.J. Hooper. 1999. Monocrotophos-induced mass mortality of Swainson’s hawks in Argentina, 1995-96. Ecotoxicology 8: 201-214.↩︎

  6. Wikelski M, Davidson SC, Kays R 2020. Movebank: archive, analysis and sharing of animal movement data. Hosted by the Max Planck Institute of Animal Behavior. www.movebank.org, accessed on 2020-10-28. Online: https://www.movebank.org/cms/movebank-content/about-movebank↩︎

  7. Kochert, Michael N et al. “MIGRATION PATTERNS, USE OF STOPOVER AREAS, AND AUSTRAL SUMMER MOVEMENTS OF SWAINSON’S HAWKS.” The Condor vol. 113,1 (2011): 89-106. doi:10.1525/cond.2011.090243↩︎

  8. About ARGOS. Collecte Localisation Satellite (CLS). Accessed 2020-10-17. Online: https://www.argos-system.org/argos/who-we-are/international-cooperation/↩︎

  9. Van Rossum, G., & Drake, F. L. (2009). Python 3 Reference Manual. Scotts Valley, CA: CreateSpace Online: https://www.python.org/↩︎

  10. GeoJSON. Accessed 2020-11-06. Online: https://geojson.org/↩︎

  11. Shapely: manipulation and analysis of geometric objects. Writted by Sean Gillies and others (2007), toblerity.org. Online: https://github.com/Toblerity/Shapely↩︎

  12. Kepler-gl. Open-source geospatial software. Accessed 2020-10-30. Online: https://docs.kepler.gl/docs/api-reference↩︎

  13. Kalnay et al.,The NCEP/NCAR 40-year reanalysis project, Bull. Amer. Meteor. Soc., 77, 437-470, 1996. Online: https://psl.noaa.gov/data/gridded/data.ncep.reanalysis.surface.html↩︎

  14. Department of Eart & Climate Sciences, San FRancsico State Universty. Accessed 2020-10-30, Online: http://tornado.sfsu.edu/geosciences/classes/m430/Wind/WindDirection.html↩︎

  15. Panoply. Developed at NASA Goddard Institute for Space Studies. Accessed 2020-11-08. Online: https://www.giss.nasa.gov/tools/panoply/↩︎