Résumé
Effusion rate is a key parameter to model lava flow advance and associated risks. Estimation of effusion rate from thermal remote-sensing using satellite data has matured to the point where it can be an operational monitoring tool, notably for volcanoes without a ground observatory. However, robust physical models, as required for quantitative interpretations, have not yet been adequately developed. The current and widely used method relates the satellite-measured radiated power to the flow effusion rate through the lava area, with an empirical fit that assumes a low surface cooling efficiency. Here we use novel fluid dynamic laboratory experiments and viscous flow theory to show that assuming low convective cooling at the surface of the flow leads to a systematic underestimation of the effusion rate. This result, obtained for the case of a hot isoviscous gravity current which cools as it flows, relies only on the respective efficiency of convection and radiation at the flow surface, and is independent of the details of the internal flow model. Applying this model to lava flows cooling under classical wind conditions, we find that the model compares well to data acquired on basaltic eruptions within the error bars corresponding to the uncertainties on natural wind conditions. Hence the thermal proxy deduced from the isoviscous model does not seem to require an additional fitting parameter accounting for internal flow processes such as crystallization. The predictions of the model are not correct however for thick lava flows such as highly viscous domes, because a thermal steady state is probably not reached for these flows. Furthermore, in the case of very large basaltic flows, extra cooling is expected due to self-induced convection currents. The increased efficiency of surface cooling for these large eruptions must be taken into account to avoid a gross and dangerously misleading - underestimate of the effusion rate. (C) 2013 Elsevier B.V. All rights reserved.