The general scope of the AIRSEA project-team is to develop mathematical and computational methods for the modelling of oceanic and atmospheric flows. Available mathematical tools involve both deterministic and statistical approaches. Domains of applications range from climate modelling to the prediction of extreme events.
Because of the complexity to describe accurately the dynamic of the ocean or the atmosphere, mathematical models are in practice idealised and simplified representations of the reality. Observations are then necessary to monitor and forecast the evolution of these geophysical states. This is done through so-called data assimilation methods that combine numerical model, observations and a priori statistical informations. Although well known, the application of such methods can still be challenging, for example when observations are available as photographic images. This is in particular true for oil spill tracking, which is the topic of this internship.
During this post-doctoral work, a focus will be made on oil spill tracking using both image data and simplified fluid mechanics models. This is an important topic due to its severe ecological and economical impact. In this context, data assimilation and modelling could provide a better knowledge of the location of the source and volume of the spill, and of the future evolution of the pollutant.
One of the main ingredient of data assimilation is the ability to compare model output with observations. In general this is not trivial, since observations may be indirect measurements of the physics processes models represent. Moreover, the choice of the distance to be used may not be obvious and depends on the matter of interest. For instance sequence of images contains a lot of information about the dynamics of the system (there are things moving around) that may not be well captured by a pixel-to-pixel comparison (i.e. use of a L 2 distance), which measure amplitude rather than position errors.
Possible steps of the study will be :
This work program is ambitious. The priority topics will depend on the applicant profile and preferences.
Applicants must have a PhD in a relevant field. A solid background in applied mathematics, as well as programming skills are required. Expertise and experience of the applicant within the following areas are highly advantageous : development and application of numerical models, inverse problems and data assimilation, image analysis and computer vision.
 S. Belongie, J. Malik, and J. Puzicha. Shape matching and object recognition using shape contexts. IEEE Trans. on Pattern Analysis and Machine Intelligence, 24(24) :509–522, 2002.
 A.J. Chorin, M. Morzfeld, and X. Tu. Implicit sampling, with applications to filtering and data assimilation. Chinese Annals of Mathematics, 34B :89–98, 2013.
 D. Huttenlocher, G. Klanderman, and W. Rucklidge. Comparing images using the hausdorff distance. IEEE Trans. on Pattern Analysis and Machine Intelligence, 1993.
 H. Ling and D.W. Jacobs. Shape classification using the inner-distance. IEEE Trans. on Pattern Analysis and Machine Intelligence, 29(2) :296–299, 2007.
 P.J. van Leeuwen. Data Assimilation for the Geosciences, chapter Particle filters for the Geosciences. Oxford University Press Blue Book Series, 2014.
Team’s website : http://team.inria.fr/airsea
AIRSEA project team (INRIA, LJK lab.)
Bâtiment IMAG – 700 avenue centrale
Campus Universitaire de Saint Martin d’Hères, Grenoble (France)
(c) GdR 720 ISIS - CNRS - 2011-2015.