Advanced geophysical reduced-order model construction from image observations – GERONIMO

The GERONIMO project aims at devising new efficient and effective techniques for the design of geophysical reduced-order models (ROMs) from image data. The project both arises from the crucial need of accurate low-order descriptions of highly-complex geophysical phenomena and the recent numerical revolution which has supplied the geophysical scientists with an unprecedented volume of image data.

The precise characterization of geophysical phenomena is a crucial need in many domains of everyday life and can have a dramatic impact in many environmental and economical fields. We think, among others, to applications related to climate studies, oceanographic analysis or meteorological forecasting which are of paramount importance for the study of global warming, the tracking of polluting sheets or the prediction of catastrophic events. Unfortunately, the laws ruling such geophysical processes depend on state variables evolving in huge dimensional spaces and are thus totally intractable. In this context, resorting to ROMs, gathering the main features of geophysical systems into a few degrees of freedom, is almost unavoidable.

The construction of relevant ROMs is a tricky problem since it has to trade carefully between two contradictory goals: decrease the complexity and preserve the model accuracy. On top of that, the construction of ROMs has often to cope with numerous uncertainties on the system parameters (for example unknown intricate boundary conditions or unknown model parameters). These uncertainties are usually ignored in the current literature, with possible dramatic consequences.

Our credo is that the huge amount of information contained in image data should be exploited to reduce the uncertainty on the unknown parameters of the models and improve the reduced-model accuracy. The GERONIMO project is thus placed in this context: we aim at processing the large amount of incomplete and noisy image data daily captured by satellites sensors to devise new advanced model reduction techniques. The construction of ROMs will be placed into a probabilistic Bayesian inference context, allowing for the handling of uncertainties associated to image measurements and the characterization of parameters of the reduced dynamical system.

The objective of the GERONIMO project is ambitious and will require the combination of many different fields of expertise shared by the project partners. We are however convinced that the new perspective proposed in the GERONIMO project will renew the ROM construction paradigm in many respects.