Atmospheric correction considerations of remotely sensed coastal waters

It is possible that the remote sensing reflectance values derived from digital images obtained in the Sea-viewing Wide Field-of-view Sensor (SeaWiFS) wavebands over shallow ocean could have a lower accuracy than those obtained over deep ocean. This is because the atmospheric corrections have been derived for deep ocean, but not completely validated for littoral waters. There is therefore a need to determine whether, and to what extent algorithms developed for the atmospheric correction of image data obtained over deep ocean can be extended to coastal environments. In order to determine in advance of the anticipated SeaWiFS launch whether the digital image data from this satellite can be accurately atmospherically corrected over shallow coastal waters using a modified version of the algorithm developed for deep ocean, a simulation study has been performed. The validated LOWTRAN radiative transfer model was used to predict the in-band SeaWiFS radiances at orbital altitudes to be expected from remote sensing water reflectance values which had been measured experimentally at the ocean surface. Gordon's modified algorithm was then used to calculate the remote sensing reflectance of the water. These values were compared to the initial surface measured reflectance values. In another experiment, hyperspectral image data was used to simulate a SeaWiFS multiband digital image, and image radiance values were projected to the ocean surface using Gordon's atmospheric correction algorithm. Problems with current data sets and with current measurement procedures are discussed and suggestions are made for improving the methodology and accuracy of the data collection and analysis process, especially for validation data sets.