Page 1 of 1

Removing high chlor_a before binning

Posted: Fri Oct 28, 2022 2:29 pm America/New_York
by schckngs
Hey forum,
Apologies if there is an obvious answer. Is there a simple way to remove high chlorophyll-a values before they are included in binned images and composites?
I am making daily, 8day and monthly chlor_a composites at 1km resolution using L2Bin > L3Bin > L3Map on downloaded and reprocessed L2 MODISA images. Ideally I would like to remove all chlor_a > 40 so that it is not averaged in during this process.

Re: Removing high chlor_a before binning

Posted: Sat Oct 29, 2022 11:09 am America/New_York
by gnwiii
Using a fixed upper limit doesn't handle all the implausible pixels (e.g., a cloud edge pixel value outside the plausible range for the time and location). You can apply filters on individual level-2 images to identify pixels with value well outside the range of values for neighboring pixels. You may be able to modify the list of flags used by l2bin to remove some pixels with implausible values. In the Northeast Atlantic, implausible values are from shallow water can be masked using depth criteria when processing level-3 data, but that doesn't help with cloud edge pixels.

Re: Removing high chlor_a before binning

Posted: Mon Oct 31, 2022 2:28 pm America/New_York
by schckngs
Hey gnwiii,

As I understand you're saying that a threshold is maybe a bit too simplistic a way to remove values.

In this case I'm using regular flagging but even so in my region of interest (BC coast), at 1 km resolution many pixels on the continental shelf in the chlor_a product are very high (> 60 mg m^-3). The areas I'm more concerned about filtering are those not directly adjacent to shore, so in the composite binning process it would be nice to remove them (the value of an 8-day composite pixel is very different including vs. excluding high chl if it is more reasonable the rest of the week).

This doesn't get at all the complexities of the chl product quality, I'm generally confident using it in this area overall due to previous work showing a decent relationship to in situ chl, but using thresholds is a common filtering method so thought I would see if it is integrated in seadas binning somehow (looks like that answer is "no" :) )


Re: Removing high chlor_a before binning

Posted: Mon Oct 31, 2022 3:49 pm America/New_York
by OB.DAAC - SeanBailey

The answer is more a 'not yet'. The binner does do something akin to what you seek, but it's very inflexible.
There is a way to set a minimum at run time. The default minimum is zero, but some products are valid with negative values, so we have a way to allow the binner to accept values less than zero. The max value for chlorophyll is defined by CHLWARN flag, which is set for values above 100. The binner defaults to masking values with CHLWARN set. A long-in-the-tooth desire of mine is to have the valid range defined by the product definition of the valid min/max range with an option for overriding by the give users the ability to do exactly what you want to do :D

I won't promise a date, but will promise that it will happen (and since someone other than me thinks it'd be nice, I've got some additional motivation to see it through!)

Until then, if you modify the l2_flag array to set CHLWARN when chl > 40, the binner would happily exclude them (if you include CHLWARN in the flaguse parameter)


Re: Removing high chlor_a before binning

Posted: Mon Oct 31, 2022 5:05 pm America/New_York
by schckngs
Hey Sean,

Thanks for your answer! I'll adjust the CHLWARN flag for future. I agree this would be a super useful feature esp at the L2/L3 binning stage for a variety of products.

Cheers, happy halloween 8-) ,