C2RCC for Landsat with landmark?

snap
#1

Hello! I want to use the C2RCC tools to produce a TSM measurement for landsat imagery, but I also want to mask the land pixels. Is there a workflow for this with the GUI? I have successfully generated the C2RCC product from a landsat 8 L1TP image, but am struggling with getting a landmark.

Any suggestions?

Thanks!
Dulci

#2

C2RCC automatically masks land pixels.

#3

The C2RCC output for landsat 8 does not mask land pixels when the input is a L1TP Landsat 8 image.

Are there options that I need to choose to tell it to do so? I see the valid pixel values toggle in the processing parameters, but there is no flag with the L8 L1TP image data for land or water.

Am I starting with the wrong level of Landsat 8 processing?

Here is a screenshot of the RGB display of the RRs bands output by C2RCC:

image
image2672×1384 1.55 MB

Here is a screenshot of the TSM band with a random color bar assigned so all the pixels can be seen:

image
image2706×1402 1.44 MB

I have looked through the guides that I can find for using the C2RCC & have not found anything to help with Landsat 8 processing. A point in the right direction would be useful.

Thanks so much!

#4

you; re right, sorry. It was masked for Sentinel-2 data, so I assumed it would be the same for every input. But I tested it and got the same result at first:

grafik
grafik1901×946 835 KB

You can select “Add Elevation Band”, to get SRTM elevations into your product

grafik

Then you open the band properties of your cocentration raster and add
AND elevation > 0 (0 if you’re working in coastal areas, the constant value of your sea surface, when you are working with inland water bodies, I had to select 1117 meters for Lake Victoria)
to the valid pixel expression (there is already some term defined, so let’s leave it in there)

grafik
grafik586×572 15 KB

This is the result:

grafik
grafik1431×934 644 KB

Is this a work-around you could use?

By the way: The reason why the Land/Sea mask operator failed (it did for me) is probably, because SRTM is in lat/lon coordinates and the Landsat product is in UTM coordinates. So another option would be to reproject your data to WGS84 and run the Land/Sea mask operator again.

Problem with C2RCC Algorithm on Landsat 8 Data
#5

Ok. This is really helpful! Thank you. I am used to having to use some work-around with Landsat. I will try both these options and see how it goes.

Cheers!

#6

alright, great!
Just one more thing: When you reproject your data, make sure to un-check “reproject tie-point grids” unless you really need them, because these are just rasters with coordinates and most of the cases not required, but the whole processing takes considerably longer.