Characteristics of speckle errors of Sea-viewing Wide Field-of-view Sensor (SeaWiFS) chlorophyll-a concentration
were analyzed, and its removal process was presented in the East Japan Sea from September 1997 to December 2007.
Level-3 data of SeaWiFS chlorophyll-a concentration provided by NASA showed significant speckle errors in the East
Japan Sea. The speckles with anomalously high concentrations were randomly distributed and showed remarkably high
bias, compared with their neighboring pixels. The speckles tended to appear frequently in winter, which might be related
to cloud distribution. Ten-year averaged cloudiness of winter was much higher over the southeastern part, with frequent
speckles, than the northwestern part of the East Japan Sea. Statistical analysis results showed that the number of the
speckles was increased as cloudiness increased.
Herein, we present a methodology of how to remove the speckles with highly anomalous chlorophyll concentration
data using Level-2 data and how to composite the chlorophyll-a data to generate Level-3 data. Considering seasonal
variations of the speckles and their statistical characteristics, dynamic threshold methods were given. Additional
methodology for high values during spring bloom was also developed by considering the chlorophyll-a concentration
frontal zone. After applying the methodology to ten-year Level-2 data, data composite of Level-2 was carried out to
produce Level-3 product and compared with the NASA product. The results showed that most speckles were disappeared
and more than 10% errors of 5°×5° mean values were reduced at speckle regions in the southeast East/Japan Sea.
This study raised the issue about speckle errors in chlorophyll-a concentration composite from SeaWiFS data in the
East Japan Sea for the first time and presented regionally-optimized composite method for more reliable chlorophyll-a
data in oceanic application researches.
|