Color difference

Hi All,

Good day. I am still waiting for your help and need suggestions regarding sharp boundaries on the RapidEye Ortho reflectance image (Geometric corrections and atmospheric corrections were made by the developer, Planet.com). 

The question is: I have a Rapid Eye mosaicked image from 2011. Due to some minor gaps in it, I took a few tiles from 2012 to make a complete mosaic. After mosaicking, sharp boundaries on the reflectance map are visible. Now the question is:

• How can realistic estimates of biomass be derived from maps that have different scores for reflectance depending on what year they have come from? 
• Why do the sharp boundaries on the reflectance map not result in sharp (and hence incorrect) boundaries in biomass?

My response to this questions are: when we apply remote seeing techniques to map biomass in a specific area, often there needs to be more images for a specific area. So remote sensing experts collect images from the following year or the previous of the same season, given biomass doesn't vary significantly with 1-year difference (indicating less variation in spectral reflectance). Those strips may also arise for multiple reasons, including edge effect, colour balancing, tile overlap or because of having different mosaic technique. Those strips can be removed after applying a resampling technique of the cell size. 

Please share your expert opinion on how to solve this issue. Please share your opinion on whether my response is correct or not. Any response in tis matter is highly appreciated.

Kind regards,

Sohel

Hi Alden (Planet Support),

Thanks for contacting me. Let me summarise my questions:

1. How can realistic estimates of biomass be derived from maps that have different scores for reflectance depending on what year they have come from? [Solved]

   Response from NASA Scientist: It is a common practice to fill gaps from previous/ following years  images. One key assumption in this aggregation process is that the relative biomass does not change significantly for most plant functional types within the grid.  For example: I reviewed the GEDI LiDAR biomass product developed for 2023 (available here: https://daac.ornl.gov/GEDI/guides/GEDI_L4B_Gridded_Biomass_V2_1.html) and noticed that the temporal coverage spans from April 18, 2019, to March 16, 2023. This means that the 2023 products are based on a combination of data from four years. This solved the problem of my first question. However, you can add further explanation on this. 

2. Why do the sharp boundaries on the reflectance map not result in sharp (and hence incorrect) boundaries in biomass?

   Response from me:  When I mosaiced 2014 tiles [no tiles from previous/following years], I see strips/sharp boundaries. I tried to apply various mosaicking technique such as smoothing, pixel maximum and median in R. Also did in QGIS. Still can see the strips. So, the question is  why do the sharp boundaries on the reflectance map not result in sharp boundaries in biomass? Please see the raw image below. Raw images can be found here 2014_SR_Tiles together with the mosaic file that has sharp strip. I didn't proceed to biomass calculation because its better to solve the strip issue before calculation of biomass. 

Thank you.

Kind regards,

Sohel