In GlobCover 2005 and 2009 project, the sensor used for spectral data collection was MEdium Resolution Imaging Spectrometer (MERIS). It observes the Earth in 15 spectral bands in visible and near infra-red region. But I am not getting which of those spectral bands is used for vegetation height detection. Land cover classes in GlobCover consist of vegetation distinguished on the basis of height. But without LIDAR, I do not think that data such as tree or shrub height can be determined. Then how is the vegetation height measured in GlobCover classification?
-
2I doubt any of the spectral bands would produce the vegetation height. I think your best bet would be to review the original project and sensor specifications to find out how this information has been derived. Have a look at earth.esa.int/documents/700255/707222/Vol11_Meris_6a.pdf and due.esrin.esa.int/page_globcover.php– DanCommented Feb 28, 2018 at 4:48
-
1Are you certain that vegetation classes in that product are distinguished on the basis of height...I see no reference to that in the documentation? Section 2.2 in this document (due.esrin.esa.int/files/…), describes the land classification techniques that were used.– DanCommented Feb 28, 2018 at 5:00
-
Yes. See page 18-19. The GlobCover legend consist of LCC with height (>5m or <5m) as a parameter.– Dark_KnightCommented Feb 28, 2018 at 5:07
-
1Have another read of section 2.2 Step IV...the classes are being labelled based on a global reference land cover database. I don't believe that the MERIS imagery is being processed in any way to derive the vegetation height. You probably need to go back a step and have a look into how the GLC2000 database was created and how the land cover classes were defined.– DanCommented Feb 28, 2018 at 8:44
-
2Perhaps this is a question in its own right...would it even be feasible to determine vegetation height from satellite imagery with a ground resolution of 300m at nadir, to the tolerances of greater than or less than 5m in height?– DanCommented Feb 28, 2018 at 8:50
|
Show 4 more comments
1 Answer
I'm responding to the comments here which will hopefully sort of answer the question.
I went on an LCCS training course in Rome a few years ago. LCCS is a methodology for creating a legend for your land classification. You build a rule set that says land cover X must have vegetation Y, in semi-arid areas and a periodic brackish water supply, etc etc. You might also state that it's vegetation must be between height X and Y metres.
So the report that @Dan links to is them saying they used the LCCS method but how you actually determined the height of the vegetation can be any method. Probably as @Michael_Stimson says:
LiDAR isn't the only way to determine height, LiDAR is relatively new and vegetation heights have been derived for decades from autocorrelation like the techniques used by drone software, also from stereo pairs where more accuracy is required.
So, I would not expect a spectral band in MERIS to be providing height data.
