MODIS SST product

Please enter here to ask a question about any NASA Science related topics!
GES DISC - jimacker
User Services
User Services
Posts: 110
Joined: Mon Sep 30, 2019 1:57 pm America/New_York

Re: MODIS SST product

by GES DISC - jimacker » Fri Apr 02, 2021 3:45 pm America/New_York

The previous post does a good job of covering the answers to both questions posed. The monthly SST data product (as for all the other monthly ocean color data products) is a product of binning and averaging -- all of the valid measurements for a specific location are "binned", and then at the end of the time period, all of the binned measurement values are averaged, providing the monthly data value. In cloudy areas, there may only be a few valid measurements, i.e. data acquired from the sea surface, in a month.

As the other response also noted correctly, SST data is derived from radiances in the IR/"near" microwave range, and this radiation can be detected through light cloud cover when visible radiation is blocked. So for a cloudy region, there may be more SST observations than visible wavelength observations.

Tags:

febryanto25
Posts: 8
Joined: Tue Mar 30, 2021 7:39 am America/New_York

Re: MODIS SST product

by febryanto25 » Sat Apr 03, 2021 1:07 am America/New_York

Thank you for the explanations. Could you please describe more detail about the terms of 'binned' based on your answer?

GES DISC - jimacker
User Services
User Services
Posts: 110
Joined: Mon Sep 30, 2019 1:57 pm America/New_York

Re: MODIS SST product

by GES DISC - jimacker » Mon Apr 05, 2021 2:32 pm America/New_York

"Binned" simply refers to the collection of valid observations. With 1 km (at nadir) resolution, a "bin" will be each 4x4 km region on the Level 3 data grid. Any valid 1 km pixel observation during the time period (8-day, monthly) will be added to the appropriate bin. At the end of the period, a mean value of all the observations in the bin is calculated, and that is reported as the data value for the Level 3 data product.

febryanto25
Posts: 8
Joined: Tue Mar 30, 2021 7:39 am America/New_York

Re: MODIS SST product

by febryanto25 » Wed Apr 07, 2021 3:35 am America/New_York

Thank you for the answer. Regarding the MODIS resolution at nadir is 1 km, why the resolution of MODIS products like chlorophyll-a, sst, and others are different from the sensor resolution of MODIS itself? For instance, the resolution of monthly SST Aqua MODIS level 3 is 4 km while the sensor resolution band 31 and 32 of Aqua MODIS are 1 km? Is it like a downscale or something? How can they (sensor resolution of MODIS = 1 km and the resolution of monthly ST Aqua MODIS = 4 km) different anyway?

gnwiii
Posts: 566
Joined: Fri Jan 29, 2021 5:51 pm America/New_York

Re: MODIS SST product

by gnwiii » Wed Apr 07, 2021 7:41 am America/New_York

Why bin to lower resolution?

For water, binning over time is affected by water motion, which has a smoothing effect. Using a high resolution for monthly data costs more in
storage, processing, and download times without adding information.

With a simple binning algorithm, binning at high resolutions introduces Moire patterns.

OB.DAAC - SeanBailey
User Services
User Services
Posts: 1115
Joined: Wed Sep 18, 2019 6:15 pm America/New_York

Re: MODIS SST product

by OB.DAAC - SeanBailey » Wed Apr 07, 2021 7:49 am America/New_York

...and the sensor native resolution is nominally 1km only at NADIR. Off NADIR, the pixel size grows, at the swath edge it's more like 3x5km.

Sean

febryanto25
Posts: 8
Joined: Tue Mar 30, 2021 7:39 am America/New_York

Re: MODIS SST product

by febryanto25 » Wed Apr 07, 2021 10:36 pm America/New_York

Thank you for the answers. Regarding binning over time is affected by water motion, which has a smoothing effect. What do you mean by that? And, could you plese describe more about Moire patterns?

OB.DAAC - SeanBailey
User Services
User Services
Posts: 1115
Joined: Wed Sep 18, 2019 6:15 pm America/New_York

Re: MODIS SST product

by OB.DAAC - SeanBailey » Thu Apr 08, 2021 8:17 am America/New_York

Moire patterns can appear when binning to higher spatial resolution than the inherent data. The binner (currently) assigns Level 2 pixels to the bin with a center location (in lat/lon) closest to the center of the L2 pixel location. While the L2 pixel viewed an area that may cover more than a single L3 bin, it contributes to only one bin. This results unfilled bins, but in a pretty pattern :D

We have updated the binner to allow an area weighting of the L2 pixel over the L3 bins it covers, so in the not-too-distant future, this ungainly side-effect of the current binning behavior will be eliminated.

Regards,
Sean

gnwiii
Posts: 566
Joined: Fri Jan 29, 2021 5:51 pm America/New_York

Re: MODIS SST product

by gnwiii » Thu Apr 08, 2021 8:59 am America/New_York

I've wanted to get time to make an example of the smoothing effect, but making your own examples or your area of interest will be a useful learning experience. Most of my work has been either global or North Atlantic open ocean, so includes significant current systems. Some students have focused on river outflows. Using the NASA Level-2 browser and your local knowledge, pick an area with dynamic features (here, that is the Gulf Stream) and a month that has few cloudy days. You can use both SST and chlor_a. Compare the level of detail in the level-2 images with the monthly level 3 mapped image for the same area.

If you have the OCSSW Processing System working, I recommend installing and running the SeaDAS benchmark (bottom of this page: https://seadas.gsfc.nasa.gov/build_ocssw/). This will give an end-to-end example from Level-0 to Level-3 mapped chlor_a data. It serves as a check on a recent installation or new version of the processing software and provides snippets of code you can adapt for your own batch processing. This script uses the new "area_weighting" to reduce Moire effects. To understand why they occur, the wikipedia article may be useful together with the fact that the ground footprint of level-2 pixels becomes much larger at pass edges. If you search for "supersampling" in the SeaDAS GUI you should get an explanation (wth pictures):
Supersampling

As long as the area of an input pixel is small compared to the area of a bin, a simple binning is sufficient. In this case, the geodetic center coordinate of the Level 2 pixel is used to find the bin in the Level 3 grid whose area is intersected by this point. If the area of the contributing pixel is equal or even larger than the bin area, this simple binning will produce composites with insufficient accuracy and visual artifacts such as Moiré effects will dominate the resulting datasets.
You can modify "seadas_benchmark.bash" to increase the resolution and disable "area_weighting" starting with the l2bin step and see Moire patterns.

Post Reply