Tropospheric Relative Humidity Profile Statistical Retrievals and their Confidence Interval from Megha-Tropiques Measurements

Ramses Sivira (1) , Hélène Brogniez (1) , Cécile Mallet (1) , Yacine Oussar (2)

1 SPACE - LATMOS
2 SIGMA - Laboratoire Signaux, Modèles et Apprentissage Statistique

The combination of the two microwave radiometers, SAPHIR and MADRAS, on board the Megha-Tropiques platform is explored to define a retrieval method that estimates not only the relative humidity profile but also the associated confidence intervals. A comparison of three retrievals models was performed, in equal conditions of input and output data sets, through their statistical values (error variance, correlation coefficient and error mean) obtaining a profile of seven layers of relative humidity. The three models show the same behavior with respect to layers, mid-tropospheric layers reaching the best statistical values suggesting a modelindependent problem. Finally, the study of the probability density function of the relative humidity at a given atmospheric pressure further gives insight of the confidence intervals.

Mots clés: Megha-Tropiques; Water Vapor profile restitution; Least Squares Support Vector Machines; Generalized Additive Models; Neural Networks

Domaines: Physique Atmosphérique et Océanique [physics.ao-ph]

Ramses Sivira, Hélène Brogniez, Cécile Mallet, Yacine Oussar. Tropospheric Relative Humidity Profile Statistical Retrievals and their Confidence Interval from Megha-Tropiques Measurements. 9th International Symposium on Tropospheric Profiling (ISTP), Sep 2012, L'Aquila, Italy. 4 pp. ⟨hal-00736777⟩