Prior to reprocessing 4, the project looked at the flagging and masking that is currently being done. A primary question is whether any change can be made in the masks used in making the level-3 binned product. The initial test is to assess how much impact is made by the various flags used as level-3 masks. The results of that test show that the stray light flag is having the greatest impact in removing samples and coverage from the level-3 datasets. Based on these results, modifications to the stray light flagging were investigated.
The changes to the flags and masks results in a net gain of samples in the level-3 bin files of 24% compared to the currently used operational algorithms (reprocessing 3). The increase in the number of bins (ie., global coverage) in a level-3 product is 24% for the daily file and decreases for longer binning periods to 14% for the 8-day period and 6.5% for the monthly.
The existing turbid water flag is set when the measured reflectance in band 5 (555nm) exceeds the estimate of reflectance at 555nm derived from the Morel 1988 clear water model. Since the Morel model requires pigment as input, an invalid chlorophyl retrieval may cause the turbid water flag to be erroneously set, or prevent the flag from being set under highly turbid conditions. To remedy this the following approach is being applied:
|Turbid Water Flag|
Currently the high-Lt flag is set for a pixel if any band's DN value is above the knee in that band, and no further calculations are performed for that pixel (i.e., this flag is use as a mask at Level 2). A change is under consideration to disable this test on the visible bands (1 - 6) and only check for high Lt in the NIR bands (7 and 8).
|High Lt masking|
The current sun glint mask is set when the estimated glint reflectance exceeds 0.005. Wang and Bailey 2001 showed that glint radiance could be effectively removed from the data. The current processing code applies a sun glint correction to data where glint reflectance is present up to the current mask level of 0.005. Wang and Bailey showed that this correction could be applied up to the point where the detectors saturate, allowing the mask to be reduced or even eliminated. The threshold for the L2 flag has been raised to 0.01, and the L3 masking based on this flag has been eliminated.
The current stray light algorithm was developed using laboratory measurements of a bright target. Pre-launch TM 31 discusses the derivation of the corrections and masking and TM 41 describes the application of the algorithm. TM 31 states that it may be possible to relax the GAC masking so that two more samples along a scan with a bright target (the samples that are 2 pixels down- and up-scan of the bright target) may be corrected instead of masked. Stray light is a major cause of masked pixels in a time-binned file, causing a 32% loss in samples, so reducing the stray light masking by 1/3 could result in 10% more samples recovered.
Currently, stray light contaminated, but not corrected, pixels are flagged at Level 2 and masked in the Level 3 binning. For reprocessing #4, it is proposed that the stray light flagged pixel be masked at Level 2 since the data contained in those pixels has uncorrected (and uncharacterized) stray light and are concequently erroneous.
The satellite zenith threshold has been extended from 56 degrees to 60 degrees. This will allow pixels on the edge of the GAC swath that were being excluded due to high satellite zenith angle to be processed. The satellite zenith angle is affected by the spacecraft roll, which occasionally caused pixels on the edge of the GAC scene to have a satellite zenith angle slight larger than the 56 degree cut off.
In September of 2000, the SeaWiFS imaging duty cycle was increased to include additional data at higher latitudes. GAC data is now routinely collected at solar zenith angles as high as 83-deg. It has been found that the standard SeaWiFS cloud-detection algorithm is too restrictive at these extreme solar zenith angles. In fact, it was pointed-out by H. Fukushima, that the predicted Rayleigh path radiance alone is sufficient to trip the cloud flag when solar zenith angles exceed 75-deg. A proposed solution to this problem is to base the cloud test on a Rayleigh-subtracted quantity, such as surface + aerosol reflectance at 865 nm. A threshold on this quasi-surface reflectance field, Rhos(865), of 0.025 (2.5%) has been found to yield very similar results to the reprocessing #3 algorithm at moderate solar and viewing geometries, while still allowing high solar and view angle observations of apparently clear ocean pixels to pass through unmasked.
With changes in the visible band gain factors, thresholds used in the current absorbing aerosol flag algorithm are incorrect. These thresholds have been updated.
The current shallow water mask is based on ETOPO5. Since a finer resolution product exists (e.g. ETOPO2) a new shallow water mask based on this higher resolution data has been generated.