A simple method for the assessment of the loss of spatial information as a function of ground pixel size: application to SeaWiFS

摘要

The effects of pixel size on the spatial variability that would be observed by a satellite ocean color sensor were investigated using ship-borne data collected during a field campaign in the lower St. Lawrence estuary. Pigment concentration images were also computed by applying simple two-band SeaWiFS Case I type radiance ratio algorithms to imagery acquired by an airborne imaging spectrometer (CASI for Compact Airborne Spectrographic Imager). Semivariograms were used to analyze the spatial behavior of in-situ chlorophyll-a data in order to determine the response of these algorithms to scale change effects. Chlorophyll gradient trends ～0.5 +- 1 mg@m~(-3) per km were used to show that the SeaWiFS pixel size, relative to a derived (algorithmic) noise equivalent distance, is appropriate. It was found that most of the fine-scale intra-pixel variation is eliminated and SeaWiFS does not detect fine-scale variation that is not useful. On the other hand, the SeaWiFS pixel size was typically found to be larger than the maximum pixel size associated with the maximum tolerable intra-pixel variability standard of 35% of the mean pixel chlorophyll concentration. This means that SeaWiFS will filter out an important amount of fine-scale systematic variability which can be 2 to 7 times the 35% standard for the range of turbidities found in the case II St. Lawrence estuary waters which we surveyed.