Welcome to the Earthdata Forum! Here, the scientific user community and subject matter experts from NASA Distributed Active Archive Centers (DAACs), and other contributors, discuss research needs, data, and data applications.
by dsgrimmer » Thu Mar 30, 2023 2:11 pm America/New_York
Hi, I'm researching the effectiveness of the Hawkeye satellite compared to existing satellites. While I was going through images, I noticed that the chlorophyll concentrations changed significantly in comparison to the MODIS images over time. I'm attaching images in a pdf for context. Both are spatial subsets of the San Francisco area. The first (29 September 2021) is what most the images look like - it has high concentrations of chlorophyll. The second (4 September 2022) displays very low concentration, not matching the data shown in MODIS - which shows high concentrations throughout, or the expectated values. The change from high to low concentration was gradual, but the MODIS images did not show this same change. I generated these images using SeaDAS, and used the same concentration scale for all images. I'm wondering why the chlorophyll concentration disparity would be so great and why it would happen? There isn't a significant difference in weather conditions, and this pattern happens in other locations too. If someone has a theory about why the data might appear that way, please let me know, thanks!
by dsgrimmer » Sat Apr 01, 2023 7:18 pm America/New_York
Hi! I'm comparing the images to MODIS Aqua, but as I was going through images, I noticed that the Hawkeye images started to show lower concentrations relative to the MODIS images. The MODIS images showed high concentrations in all the images I'm using, and the Hawkeye images were the only ones that deviated. I used the two images below to show how much they changed.
I can't attach the images as they are .nc files, but I can paste links to the Ocean Color website where I got the images. Does this work?
There are a number of possible reasons for the differences you see. There certainly is some level of degradation in the Hawkeye instrument over time (see: https://oceancolor.gsfc.nasa.gov/data/hawkeye/on-orbit-changes/), however there is currently no way to account for any temporal changes in the radiometric calibration in the processing of the data. Also, as the NIR bands are suffering from uncorrectable ghosting, the data are processed using a modified "CZS-like" approach where a fixed aerosol model is used. This model is very likely not appropriate for all scenes for all time and how much of an impact that will have will vary by scene.
by dsgrimmer » Thu Apr 13, 2023 9:38 pm America/New_York
Hi! I'm reaching out to see if you might be able to provide more feedback on this issue? I'm doing outside research as well, but do you think the degradation could cause such extreme differences sometimes and not have an effect at other times? I'm attaching a pdf that shows multiple image comparisons - some with similar and some with differing concentrations. I had to take out a lot of the slides and put multiple on one page to stay within the file size requirements. If you would like to get the full version, I would need to provide it outside of the forum. I've seen this pattern occur in other locations I've focused on, and they are very extreme differences for a period of time, then it goes back to normal. In fact, most the images show similar concentrations. I would understand how the degradation could cause differences, but wouldn't that affect the sensors all the time? Such that all of the images have significant disparity? Please let me know if you have any thoughts or if you would like more information! Thank you so much!
I did not mean to suggest that the sensor degradation was the sole reason for differences you see between Hawkeye and MODIS-Aqua. In fact, most of the degradation occurred during the extended commissioning phase. While the instrument likely continues to degrade, the rate of change over time is much lower than earlier in the mission.
Without a thorough analysis, my suspicion is that the biggest issue is with the CZCS-like approach to removing the aerosol component of the atmospheric signal. Simply changing the aerosol selection process to a more traditional "GW94" (Gordon and Wang, 1994) approach can dramatically change the results (right panel was processed with GW94):
Screen Shot 2023-04-14 at 3.11.59 PM.png (317.89 KiB) Not viewed yet
One caveat to keep in mind here...the vicarious calibration wasn't tuned to GW94. If it were the results might differ.
Since you report that most of the images show similar concentrations, but also that there are periods when they differ (and can be extremely different), my guess is that when MODIS and Hawkeye agree, the aerosol contribution is similar to the fixed model chosen for Hawkeye processing and when they differ, the model is not capturing the real aerosol load. There is also the complicating factor that Hawkeye data collection is near coasts - and coastal waters can have non-negligible contributions in the red (and NIR). Since the red band is used by Hawkeye, and the "correction" applied for non-negligible reflectance is not terribly robust (but better than not doing it) for the CZCS-like processing, when the water signal in the red is high, the Hawkeye retrievals will suffer.
by dsgrimmer » Fri Apr 14, 2023 8:51 pm America/New_York
Thank you so much for your feedback. I really appreciate it. Do you think that if the instrument was larger, a different atmospheric correction approach could be used to more accurately reflect the chlorophyll patterns? My understanding is that the small size is responsible for the modified atmospheric correction approach, and that part of the mission of Hawkeye was "proof of concept"- to demonstrate the capabilities of the instrument for future implementation in larger scale missions. In other words, is this problem specific to the satellite or the instrument? I appreciate all your help - it has provided valuable information for my lab's work!