WWF Space+Science has submitted 3 concepts to ESA’s summer of code in Space.
Through this program ESA will support student projects coding AI with Earth Observation data for the following:
Monitoring Waterholes for Wildlife with AI: wildlife, particularly migrating species depend on water for survival. In seasonal landscapes like East Africa, Southeast Asia, water accumulates and attracts wildlife – these areas vary in time, sometimes they dry up, others stay wet. We are interested in developing AI tools to identify training datasets for waterholes from high resolution satellite imagery. For example, gleaning waterhole features from Digital Globe’s GDBx platform or Google Earth to provide an annotated training dataset to develop a model to predict waterhole wetness, using information from Sentinels 1,2,3 and climate data to determine the probability that a waterhole in time is wet or dry. additional reading
Whales from Space! : Whales can be seen by the human eye in very costly, very high resolution satellite images, but not in the more publicly available high resolution imagery that conservationists can work with. We seek a coder to help us use very high resolution images (for example from GDBx or Google Earth) from the Mediterranean Sea to create an annotated data set to develop algorithms to detect whales in free images (Sentinel-1 and 2). We are trying to use very high resolution satellite images (e.g. World View 3) and other types of “conformational” data (https://obis.org) to identify whales in freely available image data, but lower resolution image data (e.g. the kinds you see on google earth).
The idea, is that we need to develop code that cannot only teach the computer how to identify whales in these lower resolution images, but can then analyze many thousands of images and do so on a regular (automatic) basis, like this. We are in the process of hiring a post-doctoral researcher to help lead this effort this summer and the successful intern would work with that researcher at the University of Brest and the European Institute for Marine Studies (both collocated in western France near Brest). additional reading
Orinoco Flooding Frequency Assessment: The Sustainable Land Use Project in Colombia aims to improve existing methods to reach a climate smart land use planning in the Colombian Orinoco grasslands, considering not only biodiversity and carbon content, but also hydrological dynamics and potential GHG emission savings. We wish to explore the hydrological dynamics in the Orinoco region, which are an important driver for the ecological connectivity of the area, characterized by extensive flooding during the rainy seasons. We need to better understand these dynamics by using both optical and/or SAR EO imagery, in order to develop monthly flooding frequency analyses through classification methods aided by AI. A coder will help us develop a model to assess flood timing and frequency over the Orinoco. additional reading
To apply, please send some information about yourself stating which project you are interested in, a c.v. and an outline of your proposed strategy for the project, which tools or approaches or ideas from any information you have researched.