Ecotope features must be mapped before they are classified.
Ecotope features are first mapped and
then classified, because both feature recognition
and mapping are conducted by applying standardized, scale-explicit rules, providing a basis for standardized
feature classification. Features are mapped and classified by image interpretation
first, and then by groundtruthing, in feature interpretation & groundtruthing
cycles. Each cycle is just one iteration of the , which must be repeated at least
two times and then validated to complete an ecotope map of a sample cell.
|
|
1. Image Interpretation
Viewing imagery in a GIS, a trained mapper identifies and classifies features across
each sample cell.
|
1A. Feature mapping
|
Features are identified in imagery and drawn as polygons into a GIS data layer by
a trained mapper using a scale-explicit, rule-based, feature identification strategy.
|
1B. Feature Classification
|
Once mapped, each feature is classified by interpretation of imagery and context.
|
2. Groundtruthing
By visiting all features and confusing areas in the field, a trained mapper checks,
corrects, and improves feature mapping and classification across each sample cell
aided by imagery and interpreted ecotope feature maps.
|
2A. Feature mapping
|
Feature identification and edge mapping is checked and corrected in concordance
with scale rules.
|
2B. Feature classification
|
Feature classification in the field is generally more reliable than that from interpretation,
and many classes can only be identified by groundtruthing.
|
3. Repeat & Validate
This cycle of image interpretation and groundtruthing is repeated at least two times,
with at least one validation check by a different trained mapper to check for agreement
with standard mapping and classification rules.
|
|
The Feature Interpretation & Groundtruthing CYCLE
|
|
Citation for AEM: Ellis E. C.,
H. Wang, H. Xiao, K. Peng, X. P. Liu, S. C. Li, H. Ouyang, X. Cheng, and L. Z. Yang.
2006. Measuring long-term ecological changes in densely populated landscapes using
current and historical high resolution imagery.
Remote Sensing of Environment 100(4):457-473.