GLC2000 GLOBAL LEGEND

1
GLC2000 GLOBAL LEGEND
GLC 2000 FIRST RESULTS WORKSHOP JRC Ispra,
18-22 March 2002
2
GLC land cover classification approach

• No fixed legend and no fixed class names

• Land cover described by classifiers and
  attributes (of vegetation / non-vegetation)
• at detail possible - using the FAO LCCS
  standards software

• Detail of class description at a given
  geographical level depends on the
• information obtainable from Remote Sensing
  (VEGETATION)
• capability to map a class consistently
• auxiliary map or information sources

• Different legends or classes can be generated by
  grouping of classifiers and attributes

• Compatibility at higher levels by grouping of
  classifiers and attributes

3
Arguments for this approach

• the user is not forced to squeeze a regional /
  local class in a predefined global class name or
  legend
• gtgte.g. forest gt 60 percent tree cover
• regional or sub-regional maps can be produced
  with legends and class names which meet the needs
  at this level
• gtgte.g. detailed steppe and desert classes,
  detailed boreal forest types
• direct links are possible to the national scale
  or field observations - if these are also coded
  with LCCS
• legends can be adapted to specific purposes at
  the same geographical level
• gtgte.g. desertification monitoring, forest cover
  assessment,
• carbon stock inventory,
• gtgt enlarge the potential applications and user
  group

4
Minimum set of required classifiers

• Objective of GLC2000
• Global land cover classification
• uniform
• consistent
• comparable across regions
• DEFINITION OF A MINIMUM SET OF REQUIRED
  CLASSIFIERS
• mandatory for describing vegetation cover classes

5
Example for a general class Cropland
(terrestrial)
Cultivated and Managed Terrestrial Areas
no further classifiers obligatory as Minimum
6
Example for more detailed requirements Natural
vegetation (terrestrial)
Note In case of missing ancillary information on
additional sparse layers the classifier
layering could be dropped.
7
Definition of a class by dominant life form with
classifiers and attributes
Definition of legend and classes depending on
geographical level and purpose
8
Global aggregation scheme
9
Significant LCCS codes and classifiers for
extraction
to be checked / complemented
10
Issues to be addressed during the working groups

• at global level
• changes at the global aggregation scheme only if
  absolutely required
• can the regional classes be converted into the
  global level?
• can the global aggregation level be consistently
  achieved?

• at regional level
• more detailed cover classes possible / desired
• can an agreement on a regional legend be
  achieved?
• proposed starting topic of first WG

11
Remarks to LCCS coding

• Threshold between closed and open forests appears
  to be more appropriate at
• 40 canopy cover
• Forest closed LCCS class closed 40 -100
  LCCS class open 40-65 ,
• Forest open LCCS class open 15-40
• Can the 40 threshold be achieved as it was not
  set in the initial requirements?

• Present version of LCCS does not allow thematic
  mix of 2 classifiers in one step
• a mixed unit needs to be defined with 2
  sub-components
• User comment // is required in order to
  differentiate a thematic from a spatial mix!

User labels (names) should be assigned with
regional / local class names
User comments in case a class does not meet
global legend Example Closed open gtgtassigned
to closed
12
the end