Version 5 (modified by mark1, 15 years ago) (diff)


Q: Why do my geocorrected Eagle/Hawk files have unexpectedly red/green/blue/yellow pixels?


The Eagle and Hawk instruments have a limited dynamic range and must be set to capture data over the appropriate range of signal strength. For example, if the area of interest is dark, then the instrument will be configured to capture as much low light detail as possible. This configuration is set based on operator experience, the principal investigator's indication of the areas of importance and the prevailing conditions. Inevitably, some pixels are unexpectedly bright – e.g. sunglint over water or part of a cloud. These pixels may exceed the maximum capture level and overflow. Overflowed pixels are marked with an invalid value (65535) in delivered level 1 data. Typically they are not in areas of interest, but should be accounted for.

Eagle uses a frame transfer CCD, where data are read out in rows. Incoming light continues to accumulate in unread rows during the transfer and is removed by “smear correction” software, which relies on data from one row to correct the next. If a pixel overflows, information is lost and all subsequent pixels in that column cannot be fully corrected. In Eagle, the net effect is that an overflow at 600nm will cause all bluer bands (600nm -> ~400nm) to be undercorrected for that spatial pixel. When Eagle data with overflows are delivered, we mark as bad all bands following an overflow as they will incorporate some unknown additional light.