In the documentation it looks like there should be “summary-classes” e.g. 60 for “Tree cover, broadleaved, deciduous closed to open (>15%)” and further detailed subclasses for closed cover (>40%) and open cover (15-40%) (classes 61 and 62 for broadleaved, deciduous). However the areas of these subclasses are not included in the summary class, as can be seen in the attached plot for 2022. Are we misunderstanding the documentation (p.34)?
Basically, we need to know which cover % classes the different bands refer to, because according to the legend it should only be 2 (open: >15 and <40 and closed: >40). But what is contained in the closed to open class?
Apologies for the delay in responding to your inquiry, and for the fact that your message was unfortunately overlooked. The Land Cover team will provide a detailed answer to your question by the end of July.
In the meantime, could you please let us know how you spatially aggregated the original Land Cover maps? Specifically, did you use the CCI LC user tool (available at ESA/CCI viewer)?
Thank you very much for your patience and understanding.
Thank you for your question and apologies for the delay in getting back to you.
I think the key point is how the hierarchy of the Land Cover legend should be interpreted in relation to the class codes stored in the map.
Taking broadleaved deciduous tree cover as an example, class 60 is the Level-1 class “Tree cover, broadleaved, deciduous, closed to open (>15%)”, while classes 61 and 62 provide a more detailed distinction according to canopy cover:
60 – broadleaved deciduous, closed to open (>15%)
61 – broadleaved deciduous, closed (>40%)
62 – broadleaved deciduous, open (15–40%)
However, class 60 is not a summary raster class containing the pixels assigned to classes 61 and 62. These are mutually exclusive class codes in the Land Cover maps. Therefore a pixel coded 61 or 62 cannot be also coded 60. Class 60 is assigned to pixels identified as broadleaved deciduous tree cover (>15%) for which the additional closed/open distinction is not provided.
This explains why the spatial extent of classes 61 and 62 is not included in the pixels coded as class 60 in your plot. The relationship is hierarchical at the legend level, rather than an aggregation of the corresponding class codes in the raster.
The same structure applies to needleleaved evergreen (70/71/72) and needleleaved deciduous (80/81/82) tree cover. Broadleaved evergreen tree cover is an exception in the legend. Class 50 has no corresponding closed/open subclasses. It means that all broadleaved evergreen tree cover pixels are represented without this additional canopy cover subdivision.
Therefore, if your intention is to derive the total area of broadleaved deciduous tree cover (>15%) irrespective of the closed/open distinction, you should aggregate classes 60, 61 and 62.
I hope this clarifies the interpretation of the legend and what you are seeing in the maps. Do not hesitate if some doubts remain.
May I suggest a different name for the 60/70 class, since it is not close to open, but “unclear” or “unclassified”?
However it is a bit unsatisfactory for practical use. If I want to include all closed broadleaved deciduous forests (> 40%), should I now include class 60 or not?
