Frouin, Robert: Scripps Institution of Oceanography, UCSD (Project Lead)
Project Funding:
2007 - 2010
NRA: 2006 NASA: EOS
Funded by NASA
Abstract:
n our preliminary work, we have introduced and developed an inversion methodology, based on nonlinear regression, to retrieve oceanic parameters like the chlorophyll-a concentration and the marine reflectance. The inversion model that we introduced is in the form of a function field. This is a model in which the angular information is accounted for in a specific manner that matches well the physical characteristics of the remote sensing problem. The obtained results suggest that our methodology, compared with other techniques, has the potential for improving the accuracy of the retrievals. In those experiments, we noticed, however, that large noise levels on the top-of-atmosphere reflectance might occur. The noise and uncertainties on the measurements constitute a severe issue for any inversion methodology. When not accounted for appropriately, the presence of noise may result in biases, trends and large errors on the retrieved values which, statistically, reflects the non-consistency of the inversion method.
In this proposal, our main goal is to cope with noise issues and turn our preliminary work into an operational procedure, suitable for remote sensing of ocean color, in either simple or complex water bodies (Case I and Case II). To this end, we propose first a procedure to assess the noise distribution. Next, we plan to use this estimated noise distribution in the design of a function field to obtain consistent retrieved values. Because of the special treatment of the angular information, the function field methodology is particularly well suited for merging concomitant data originating from various sensors, which is our final goal. Data from multiple sensors, when considered simultaneously, clearly brings additional information about the oceanic medium, especially when one is concerned with bidirectional aspects of the marine reflectance.
The investigation will provide an analytical inversion procedure, based on function field methodology, for the retrieval of marine reflectance. The regression models, owing to their mathematical structure, should enable more accurate retrievals of marine reflectance in Case 2 waters and in the presence of absorbing aerosols. They will be fast in execution, therefore suitable for use on an operational basis. The work on the fundamental issue of noise will result in models intended to be robust, at least statistically consistent, which will ensure continuity, across sensors, in the quality of the marine reflectance. Complete, operational processing lines will be constructed for atmospheric correction of current and planned ocean-color sensors, with the possibility of merging data from different sensors. The models will be applied to actual satellite imagery, compared with other algorithms, and evaluated using in-situ measurements. The regions where improvements are expected to occur include dust-contaminated and polluted areas and the coastal zone. They are large spatially and tend to contain very productive waters. The gain in accuracy, therefore, by allowing a better knowledge of rates of primary production, would be very significant.
The proposal is organized as follows. Our preliminary work on function fields is reported in Section 2. Our objectives are presented in Section 3. The technical plan is detailed in Section 4. We start by outlining the proposed approach, and describing it first in a general setting. Next, we describe how the function field methodology can be easily extended to merge data originating from various sensors and improve the retrieval of the bidirectional marine reflectance. Practical aspects regarding simulation, application, and validation are described in a dedicated subsection, prior to our work schedule.
2011 NASA Carbon Cycle & Ecosystems Joint Science Workshop Poster(s)
- Statistical Inference in Ocean-Color Remote Sensing
-- (Robert Frouin, Bruno Pelletier)
[abstract]
2008 NASA Carbon Cycle & Ecosystems Joint Science Workshop Posters
- Satellite Ocean-Color Remote Sensing in the Presence of Clouds
-- (Robert J. Frouin, Pierre-Yves Deschamps, Lydwine Gross, Hiroshi Murakami, Lucile Duforęt)
[abstract]
- A Stochastic Technique for Satellite Ocean-color Remote Sensing
-- (Robert J. Frouin, Bruno Pelletier)
[abstract]
More details may be found in the following project profile(s):