0

I want to do a time-series analysis on Landsat surface reflectance data for an extended period of time (2005-2018), which requires integration of Landsat 5, 7, and 8. Is there any procedure to make the data from ETM+ and OLI sensors compatible so they can be used in a single time series together? I looked at the scholarly publications but only found one paper dated 2014, suggesting a linear regression technique for converting OLI to ETM+.

https://www.researchgate.net/publication/277674991_Continuity_of_Reflectance_Data_between_Landsat-7_ETM_and_Landsat-8_OLI_for_Both_Top-of-Atmosphere_and_Surface_Reflectance_A_Study_in_the_Australian_Landscape

  • Your question is unclear. What you want as the output? Is there any procedure to bring the data from different sensors to the same ground? (the ground is the same). – If you do not know- just GIS Feb 25 at 18:45
  • By "ground" I didn't mean the Earth. Sorry for confusion, I edited my post to make my question more clear. – Shahriar49 Feb 25 at 22:21
0

You do not really give enough information. The systems are generally highly comparable and designed this way. If you are using unsupervised or supervised approaches you may need to do no changes at all. The same for other landscape methods. If you are trying to get identical TOA or SR values from the same ground pixel in the same bands then you many need some adjustment on some bands, although it is very nominal on the most often used bands.

The regressions equations provided in your link are for one country and a type of landscape. They cannot be applied elsewhere. It is also an MDPI journal.

Differences between the two systems appear entirely nominal in regards Top-of-Atmosphere and Surface Reflectance for the visual bands For almost all cases the following bands can be used as is:

OLI - ETM

2 - 1

3 - 2

4 - 3

6 - 5

Adjustment For the band below a linear regression adjustment is suggested in the MDPI paper but it is time, locations, and landscape specific; but as I said it may not be required depending on your approach and it may not work on your landscape.

5 - 4 (regression equation from your link)

Flood, Neil. (2014). Continuity of Reflectance Data between Landsat-7 ETM+ and Landsat-8 OLI, for Both Top-of-Atmosphere and Surface Reflectance: A Study in the Australian Landscape. Remote Sensing. 6. 7952-7970. 10.3390/rs6097952.

  • Many thanks for your reply and important point about the paper (being location specific). Is there any other problem with MDPI journal? – Shahriar49 Feb 26 at 2:13
  • And I knew the figure you gave above. As you see, only B, G, and R bands are good match between ETM and OLI and the bandwidth of other bands (especially for NIR and SWIR-1 bands) are very different. How the measurements can be compatible then? – Shahriar49 Feb 26 at 2:15
  • @Shahriar49 MDPI are considered a predatory publisher by many. They used to be on the predatory list. en.wikipedia.org/wiki/MDPI – If you do not know- just GIS Feb 26 at 4:00
  • @Shahriar49. As they do not need to be. As long as they are internally consistent to themselves then you can use them in most LULC change approaches. You are not usually interested in the values but what the values represent. – If you do not know- just GIS Feb 26 at 4:03
  • But I think if I want to make a whole time series by joining Landsat 7 & 8 and run a change detection algorithm on that, I exactly need that their values be consistent, isn't it? – Shahriar49 Mar 5 at 1:00

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.