The European Starling is a bird native to most of temperate Europe and western Asia. I do not remember who was the first person that talked to me about it, Tim?, but since then I knew I wanted to develop something with it. The reason is funny/sad story behind it. The bird, that was only native in Europe and Asia, got introduced in lot of different places at the end of the 19th century. Since then it has spread to all continents and has been treated like a pest in lot of places, and for example, in Australia there is still efforts to prevent its introduction in the West. On the other hand in Europe it has been declining a lot, and it is now actually covered there by the Red List.
Specially appealing is the story on how it got introduced in North America. I will just quote Wikipedia for that:
Although there are approximately 200 million starlings in North America, they are all descendants of approximately 60 birds (or 100 [1]) released in 1890 in Central Park, New York, by Eugene Schieffelin, who was a member of the Acclimation Society of North America reputedly trying to introduce to North America every bird species mentioned in the works of William Shakespeare.
I knew this was a story that could really be catchy. Specially if we could use scientific primary data to show this story. While working with Tim Robertson and Andrew Hill we started thinking about using Clustr, from Flickr, to create polygons out of primary data and see if we could display this story. I demoed this in Geoweb and TDWG this year and the feedback was most of the time really good. You can watch the video at Vimeo.
The challenge for that was that there is more than 1 million observations of the starling now available on GBIF and the classical point in map did not work well, the visualizations were tedious... well, kind of complicate. But the second semester of this year we started to see interactive maps that seemed to be analyzing raster images on the fly in Flash. This is really really cool. And since then we were just thinking more and more in raster representation of data to further filter in the client and allowing much more rich story telling. And then, one day, I showed the work from Andrew Cottam from WCMC on sea level rise and Google Maps for Flash. That was awesome! And being such a nice guy he is, he publish his code and saved me the time of figuring out the bitwise operations needed for at least one band raster. I am not sure if he wants me to put a link to his ongoing work so I will wait for him to publish it first (maybe in this blog ;) ).
So I could not resist and with the help of Tim preparing the raster tiles for the starling, and Sergio doing some UI, we prepared the following demo application.
(Click the image to open)
Drag the slider from 1880 to 2010 to see the accumulative records (by date recorded) for the data available on the GBIF network. While you drag the slider you will be presented with tooltips mostly taken from Wikipedia.
Soon we will release all the source code, once a bit cleaned, and will share more technical details. And the best is yet to come... we only used one band on this demo, but we have 3 to play with!!
I hope you like it and want to share some comments.
Ah! Dont forget to turn on sound!
Tuesday, December 15, 2009
Subscribe to:
Post Comments (Atom)
6 comments:
Very cool viz guys!
Really awesome. There are a lot of people out there doing citizen science projects dealing with monitoring invasive species where such viz. techniques would be amazing and useful! Too cool.
Nice! Do you get any meaningful results if you use non-cumulative counts of records per year, for example a shift (to the West) or decline (UK?) in distribution? Or do we need more data for that?
Hey Peter. Well the data in GBIF is really basic primary data, not really inventories or things like that. So no, if we group the data by, let say year, you will not get much information about the distribution of the species.
So yeah, lack of data, and also lack of other kind of data, like populations, inventories, etc. With just primary data aggregated from multiple sources it is almost impossible to recreate any species story on a global base.
But as Rob is pointing I believe this could be a great technique for controlled experiments like bioblitz and things like that.
We are also trying the same technique now for species distribution models, fire risk and rasters like that.
Finally the latest experiments we are doing are making use of 2 and 3 bands in the raster. I will be able to publish something about this soon.
Post a Comment