Merging SeaWiFS and MODIS/Aqua Ocean Color Data in North and Equatorial Atlantic Using Weighted Averaging and Objective Analysis
| Type | Article | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| Date | 2006-11 | ||||||||
| Language | English | ||||||||
| Author(s) | Pottier C1, Garcon V2, Larnicol G1, Sudre J2, Schaeffer P1, Le Traon Pierre-Yves3 | ||||||||
| Affiliation(s) | 1 : CLS, Space Oceanog Div, F-31520 Ramonville St Agne, France. 2 : CNRS, LEGOS, F-31401 Toulouse 9, France. 3 : IFREMER, F-29280 Plouzane, France. |
||||||||
| Source | IEEE (0196-2892) (IEEE), 2006-11 , Vol. 44 , N. 11 , P. 3436-3451 | ||||||||
| DOI | 10.1109/TGRS.2006.878441 | ||||||||
| WOS© Times Cited | 51 | ||||||||
| Keyword(s) | signal analysis, sea surface, marine vegetation, Biology | ||||||||
| Abstract | Two approaches of ocean color data merging were tested and compared in the North and Equatorial Atlantic Basin: the weighted averaging and the objective analysis. The datasets used were the daily level-3 binned data of chlorophyll-a from the Sea-viewing Wide Field-of-view Sensor (SeaWiFS) and Moderate Resolution Imaging Spectroradiometer on the Aqua satellite over the year 2003, which is the first common full year of operation. Since they represent input for both approaches, matchups between the satellite and the in situ data from the SeaWiFS Bio-optical Archive and Storage System and the Atlantic Meridional Transect were first studied to compute a spatial map of the root mean-square error and of the bias. Because of the log distribution of the chlorophyll fields, each approach was applied to untransformed and log-transformed values. The application of the weighted averaging to log-transformed values does not show significant differences in comparison to its application to untransformed values. This is not the case, however, for the objective analysis that gives better results when applied to logtransformed values. Both approaches give combined chlorophyll data of equivalent quality, although the objective analysis could be improved with a better statistical characterization of noise and signal covariance. The main advantage of the objective analysis is its ability to interpolate in space (and time) by taking into account the characteristic scales of chlorophyll-a. As a result, the spatial coverage of the combined data is at least twice as large in the case of objective analysis than weighted averaging. OCR NOT CONTROLLED | ||||||||
| Full Text |
|