L2gen process the estuary and coast have some problem

[xiao_0.026]

When I process L1b image of the estuary use the L2gen, I find the result image L2 have the NaN data in the estuary, And I look for the L2gen algorithm for the solve the problem.
So I find the shallow water <30m of areas are not treated. the estuary and coast be contained to the shallow water.
I would like to ask the teacher how to solve this problem.

[gnwiii]

You haven't mentioned the problems you hope to address using ocean colour data.   You best hope is that someone has already worked out an approach to a similar problem.

There are a number of factors to consider for near-shore and shallow water optical remote sensing.  The response of the sensor has a 2-D Gaussian shape.  Land is generally much brighter than water, so the values of a near-shore "pixel" is often a mix of bright light from a small land area (the "tail" of a Gaussian curve) and lower light levels in a much larger water area.  The current l2gen algorithms assume that all the light is from water that is relatively clear if you filter out the phytoplankton.  If the water is clear, however, light will reach the bottom in shallow areas.   In this case, the colour of the light is affected by the nature of the ocean floor -- is it white sand, dark mud, or vegetation?  

In some cases, images from ocean colour sensors using "false colour" based on the reflectances can help track river plumes or characterize bottom types.  Since these ad-hoc images are outside the existing calibration and validation efforts, so require in situ sampling to validate the ocean colour results.   One example of this is planning airplane flights for lidar depth measurements.   In that case, the goal is to find periods of time when the water is clear enough for the lidar signal to reach the bottom.  Another example is mapping coral reefs.

If you have high concentrations of phytoplankton so there is little light at the bottom and dark land (dense vegetation), the optical data might be usable with relatively minor tweaks to the existing algorithms, but again you need  in situ sampling.

[xiao_0.026]

Thank gnwiii very much for your answer,

I understand what you mean. At present, this problem has not been solved, but by adjusting parameters, some pixel values in the estuary area are not empty values. I have a question, L2 image by atmospheric correction of L2gen obtain the reflectivity value of water pixel? and if use the l2gen process the Nile estuary image of MOD02HKM, Do I need to look up some atmospheric parameters of the Nile estuary, such as solar zenith Angle and satellite zenith Angle and so on.

wish you have a good time.
xiao_0.026

[gnwiii]

Ocean-color remote sensing of the Nile delta shelf and SE Levantine basin ... mentions the lack of public in situ data for this region.  Simply adjusting parameters in an attempt to reduce the number of pixels flagged as invalid you may end up with nonsensical values and without some  in situ data you have no way to determine whether your results are nonsense or contain useful information.   You might be able to test your adjustments with some other estuary where there are public in situ data.

[xiao_0.026]

Thank you for your answer
I have adjusted the parameters to make the reflectivity of the estuary appear, but I don't know if the values obtained by this method are correct, I am look for  other data to verify. And I am looking for MOD23 - Suspended solids, on oceancolor couldn't find it, Do you know where to get this data?

wish you a good mood
xiao_0.026

[gnwiii]

See Topic 1602.  The mentioned products are
listed in the $OCSSWROOT/run/data/common/product.xml file.  For "tsm":


    <product name="tsm">
        <standardName>mass_concentration_of_suspended_matter_in_sea_water</standardName>
        <units>mg l^-1</units>
        <category>Derived</category>
        <range>
            <validMin>-1</validMin>
            <validMax>5000</validMax>
            <displayMin>0</displayMin>
            <displayMax>100</displayMax>
        </range>
        <algorithm name="swim">
            <cat_ix>258</cat_ix>
            <prod_ix>1</prod_ix>
            <description>Total Suspended Matter, SWIM Algorithm</description>
            <reference>L. I.W. McKinna, P.R.C. Fearns, S.J. Weeks, P.J. Werdell, M. Reichstetter, B.A. Franz, D.M. Shea, and G.C. Feldman, "A semianalytical ocean color inversion algorithm with explicit water column depth and substrate reflectance parameterization", Journal of Geophysical Research 120, 1741-1770, doi:10.1002/2014JC010224 (2015).</reference>
        </algorithm>
    </product>


Note that this was developed for the Great Barrier Reef, so would need modifications for your region of interest.   The article notes the need for further validation using in situ data, which is why it is called an "evaluation" product.

[xiao_0.026]

Thank gnwiii very much for your answer

In the nowdays,I have solves the question.Thank you very mcuh.

Now, I have a Seadas gpt?Graph Processing Toolkit?question.
How can I use it to reprojection in the nc format remote image, the image is the L2gen result.

[gnwiii]

There are many ways to get images from level-2 files.   Do you want images from a single level-2 pass or are you interested in images that combine level-2 data from multiple passes (e.g., to cover a larger geographic area than a single pass or to get images with less missing data due to clouds, etc.)?

The OCSSW programs are generally faster and use fewer resources than GPT.   For a single pass there is l2mapgen.   If you want to combine passes you should investigate binning using l2bin to create (level-3) binned files from level-2, then l3bin to combine level-3 files (over space and time), and then l3mapgen.   The alternative to binning is Mosaic, one of the GPT operators, per  GPT Cookbook: Mosaic

IOCCG Report 4 provides background information on binning.  Many forum posts discuss binning.   You should be aware that you can't view level-3 binned files until they have been mapped, e.g., using l3mapgen.

[xiao_0.026]

Thank gnwiii very much for your answer

While I was working on this problem, I came across another one.Does L2gen's TSM_SWIM algorithm work close to shore?
Are there any parameters that can be adjusted in the TSM_SWIM algorithm in L2gen to better match the inversion of suspended sediment concentration in the coastal and estuarine regions?

[gnwiii]

Were you able to provide the NetCDF files with depth and bottom reflectances for the various bottom types in your area?   As I mentioned, optical remote sensing has limitations for near shore and shallow water.  SWIM tries to handle shallow water (if you provide good depth, bottom type, and reflectance data), but you need in situ data to determine if it is successful.   Since access to in situ data for your area is problematic, you might look for data from an area that is more similar to yours than the GBR.