A Spatial and Temporal Analysis of Atmospherics Parameters Retrieved by a Neuro-varationnal Method off the West African Coast
Daouda Diouf *
Laboratoire de Traitement de l'Information - ESP, Université Cheikh Anta Diop de Dakar, Sénégal.
Sylvie Thiria
Laboratoire d'Océanographie et du Climat: Expérimentation et Approches Numériques, Université Pierre & Marie Curie, France.
Awa Niang
Laboratoire de Traitement de l'Information - ESP, Université Cheikh Anta Diop de Dakar, Sénégal.
Julien Brajard
Laboratoire d'Océanographie et du Climat: Expérimentation et Approches Numériques, Université Pierre & Marie Curie, France.
Michel Crepin
Laboratoire d'Océanographie et du Climat: Expérimentation et Approches Numériques, Université Pierre & Marie Curie, France.
*Author to whom correspondence should be addressed.
Abstract
In this work, we study spatial and temporal atmospherics parameters evolution retrieved by neuro-variationnal method from SeaWiFS observations measured off the west African coast.
The SeaWiFS sensor measures the radiance above the top of atmosphere (TOA) solar irradiance.
SeaWiFS use standard algorithm to invert the signal in order to retrieve weakly absorbing aerosol optical thickness (AOT) less than 0.3 whereas the Senegalese coasts are frequently crossed by desert dust plumes from large optical thickness.
A neural algorithm, so-called SOM-NV, was developed to deal with absorbing aerosols and to retrieve their optical parameters, off the Senegalese coast, from SeaWiFS observations.
The impact of meteorological variables on these restitutions was studied over the entire period of the observations that we analyzed and over the whole studied area, on the one hand, but also in a more thorough way on three "sub-area" located in north, south and center. The results obtained showed that the composition of aerosols in the atmosphere is a function of the seasons. High altitude zonal U winds are correlated with non-desert aerosols of -62.16% in winter and autumn. The correlation is -60.32% between dust aerosols and the zonal wind.
Keywords: Kohonen map, neural network, SeaWiFS sensor, atmospherics parameters.