Description

We propose a physical sound and time-consistent method for estimating and tracking horizontal dense motion fields at various atmospheric depths from satellite sequences of pressure images.

Based on a vertical decomposition of the atmosphere, we propose a dense motion estimator relying on a perfect dynamical model (a shallow-water multi-layer model adpated to observations of cloud top pressure). This estimator use a framework derived from data assimilation and are applied on noisy and incomplete pressure difference observations derived from satellite images.
More precisely, dense and time-continuous pressure difference maps are reconstructed according to shallow-water model on each cloud layer. While performing this reconstruction, the variational process estimates on the whole image sequence time-continuous average horizontal wind fields of the multi-layer model.

Results

Reconstruction of dense pressure difference maps:

True pressure map	Noised and occluded pressure map	Reconstructed pressure map

Synthetic experiment : simulated pressure image related to a given layer (image on the left); using pressure observations (middle image), dense and time-continuous pressure maps (image on the right) are reconstructed according to the a priori dynamic law for atmospheric layer evolution.

Estimation and tracking of layer motion :

t=0h00min
t=2h30min
	High layer	Middle layer	Low layer

While performing the pressure map reconstruction, the variational process estimates on the whole image sequence time-continuous average horizontal wind fields of the multi-layer model : first (above) and last (below) estimated horizontal wind fields superimposed on observed pressure difference maps (Real world experiment).

Reference

N. Papadakis, P. Héas and E. Mémin. Image assimilation for motion estimation of atmospheric layers with shallow-water model . Asian Conference on Computer Vision (ACCV), Tokyo, Japan, November 2007.

N. Papadakis, P. Héas and E. Mémin. Motion estimation of 2D atmospheric layers with variational assimilation techniques. 15th Satellite Meteorology and Oceanography Conference of the American Meteorological Society, Amsterdam, Netherlands, September 2007.