Measure and exploitation of multisensor and multiwavelength synergy for remote sensing: 2. Application to the retrieval of atmospheric temperature and water vapor from MetOp

In the companion paper, classical information content (IC) analysis was used to measure the potential synergy between the microwave (MW) and infrared (IR) observations from Atmospheric Microwave Sounding Unit-A, Microwave Humidity Sounder, and Improved Atmospheric Sounding in the Infrared instruments, used to retrieve the atmospheric profiles of temperature and water vapor over ocean, under clear-sky conditions. Some limitations of IC were pointed out that questioned the reliability of this technique for synergy characterization. The goal of this second paper is to develop a methodology to measure realistic potential synergies and to construct retrieval methods able to exploit them. Three retrieval methods are considered: the k nearest neighbors, the linear regression, and the neural networks (NN). These statistical retrieval schemes are tested on an application involving IR and MW synergy. Only clear-sky, near-nadir radiances over ocean are considered. The IR/MW synergy is expected to be stronger in cloudy cases, but it will be shown that it can also be observed in clear situations. The inversion algorithms are calibrated and tested with simulated observations, without any loss of generality, using similar theoretical assumption (same radiative transfer model, observational noise, and a priori information) in order to truly compare the IC and the direct statistical retrieval approaches. Multivariate and nonlinear methods such as the NN approach show that there is a strong potential for synergy. Synergy measurement tools such as the method proposed in this study should be considered in the future for the definition of new missions: The instrument characteristics should be determined not independently, sensor by sensor, but taking into account all the instruments together as a whole observing system.