Page 1 of 1

VNP09GA VIIRS products

Posted: Tue Mar 09, 2021 10:38 am America/New_York
by pilar
Currently, I am building Maximum Value Composites (NDVI MVC) of 10 days time series, using VNP09GA VIIRS products. But I have doubts about what sensor zenith angle apply. With MODIS, I was applying sensor zenith angle < 38° to get the best quality pixels. But in the VNP13_User_Guide_ATBD_V2.1.2, I have seen the VIIRS Per Pixel Reliability assignment scheme, where all the pixels with Low Aerosol and sensor zenith angle > 40° are assigned Acceptable (Rank =2). Why these high values of sensor zenith angle are good quality pixels. In addition, what is the reason to consider all pixels with climatology aerosols as poor quality?

Re: VNP09GA VIIRS products

Posted: Thu Mar 11, 2021 1:25 pm America/New_York
by LP DAAC - jwilson
We have forwarded your inquiry to our Subject Matter Expert for assistance.

Re: VNP09GA VIIRS products

Posted: Thu Mar 11, 2021 3:55 pm America/New_York
by lien
We have asked the developer of the vegetation indices for particulars on why the higher angles are acceptable in VIIRS. VIIRS swaths are over 700 kilometers wider then MODIS so the daily equatorial gaps that exist in MODIS do not exist in VIIRS, so wider angles are acceptable in VIIRS, but we are trying to get particulars why. Also, climatology (historical) data is not used in the full resolution MODIS data, only in the CMG products because the climate modelers desired gap and cloud free data. As soon as I hear more I will pass it along.

Re: VNP09GA VIIRS products

Posted: Thu Mar 11, 2021 5:26 pm America/New_York
by lien
The product developer replied back and gave us a lot of insight, please see the following.
Please let us know if you have further questions.

Here is a quick and quite rough reply,

Let me try to clarify, as some of these topics tend to be a bit "subjective", so you have plenty of room to exercise your ideas/methodologies and you'll likely get reasonable results, provided you do not let BAD data in?

First, let's look at what matters from an observation (target status) perspective when compositing:
- Let us agree that the primary purpose of compositing is to remove clouds (& other contaminants)
- Removing clouds is based on MVC since NDVI is low in the presence of clouds
- We've added over the years what we call QA-MVC to avoid the pitfalls of bad VI-data and guarantee good data to the MVC algorithm
- Finally, we added the view angle constraining method (ex: < 20, < 30, or <40, etc.) to minimize BRDF impacts and it became QA-CV-MVC.
This has built-in accounting for pixel size and atmosphere correction performance (read on).

Second, we will try to favor the NADIR view (within reason) so we work with BINS (View Zenith between certain ranges are considered similar)
- In this case, we consider any View Zenith < 20 to be almost NADIR
- Between 20 ad 40 reasonable Angle, but not great
- > 40 to be a bad view angle but still acceptable if the Aerosol is low and other condition are fine

What we are after here is to see the target under the best conditions possible first (no cloud, no aerosols, etc...).

So if Aerosols are very low we consider all observations to be reasonable candidates for selection and degrade their RANK based on the View Angle.
The reasoning here is that Medium Aerosols (or Climatology) will not be corrected properly no matter what and the data is somewhat compromised, so we want to keep it separate from the data with a good atmosphere.

Remember also that a High View Zenith angle does two things:
- Path radiance becomes long (slanted view) making the atmosphere even thicker and harder to correct for
- Pixel size changes, but for VIIRS since it is a push broom this is not as bad as MODIS (with bow tie at extreme angles, pixels grow so much in MODIS to 8 times their nadir sizes, yet still considered and stored as the nominal size at nadir ??????). But VIIRS pixels remain reasonable at a high view angle

Our logic then, is that observations with low aerosols will always be better than the ones with medium or high or climatology aerosols no matter what the other conditions are.

Finally, Climatology means the aerosol algorithms did not converge, and could not estimate the aerosol thickness for that pixels and had to use climatology (sort of default values from external sources).
Usually, climatology corresponds to clouds (or bright targets like snow or bright deserts) so the data is not good in the first place.
For that reason, climatology is almost always associated with poor quality and useless data.

Hope that helps-