Separating measurement from definition: an object-based workflow for national forest mapping
Abstract
Forest is defined rather than observed, and definitions are written to serve different management objectives. In Norway, four authoritative datasets place forest between 35 and 43% of the mainland because each encodes a different rule. Categorical forest maps inherit this problem, since each serves only the definition it was built on. We present a documented object-based workflow that separates measurement from definition. Very high-resolution imagery, Sentinel-2, and a laser-derived canopy height model are combined with the existing national masks in a spatial database, and the result is segmented in two steps into tree crowns, tree points derived from them, and forest-stand segments carrying height, crown cover, species, and site index. A rule-based expert system then classifies the tree points. Applied to mainland Norway, the workflow produced 2.72 billion tree crowns and 57.8 million forest-stand segments covering 144,755 km2. Because structure is stored as attributed objects rather than a single class, thresholding one dataset reproduces forest areas from 15.8 to 44.7% and brackets all four national estimates. Independent interpretation of 702 sampled points gave a commission error of 0.6% for the accepted tree points. The resulting dataset is called Skogmaske.
Related
- National-scale mapping locates the mismatch between perceived and assessed flood risk in India
- Spatial Colorimetric Mapping of pH and Dissolved Carbon during Convective Dissolution of CO{\_2}: Constraints from Domain, Geometry, and Indicator Mix on Measurement Reliability
- National-scale mapping of ecosystem services in Canada to support watershed management
- AI Agent for Hydrologic Modeling: Definition, Development, and Application
- Separating Weight and Likelihood Tail Effects in Bayesian Neural Groundwater Inversion
BibTeX
@misc{gjertsen2026separating,
title = {{Separating measurement from definition: an object-based workflow for national forest mapping}},
author = {Arnt Kristian Gjertsen and Shivesh Kishore Karan and Knut Bjørkelo},
howpublished = {EarthArXiv},
year = {2026},
month = oct,
doi = {10.31223/x5vj7b},
url = {https://eartharxiv.org/repository/view/15370/},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Separating measurement from definition: an object-based workflow for national forest mapping
AU - Arnt Kristian Gjertsen
AU - Shivesh Kishore Karan
AU - Knut Bjørkelo
PY - 2026
DA - 2026/10/05
PB - EarthArXiv
DO - 10.31223/x5vj7b
UR - https://eartharxiv.org/repository/view/15370/
AB - Forest is defined rather than observed, and definitions are written to serve different management objectives. In Norway, four authoritative datasets place forest between 35 and 43% of the mainland because each encodes a different rule. Categorical forest maps inherit this problem, since each serves only the definition it was built on. We present a documented object-based workflow that separates measurement from definition. Very high-resolution imagery, Sentinel-2, and a laser-derived canopy height model are combined with the existing national masks in a spatial database, and the result is segmented in two steps into tree crowns, tree points derived from them, and forest-stand segments carrying height, crown cover, species, and site index. A rule-based expert system then classifies the tree points. Applied to mainland Norway, the workflow produced 2.72 billion tree crowns and 57.8 million forest-stand segments covering 144,755 km2. Because structure is stored as attributed objects rather than a single class, thresholding one dataset reproduces forest areas from 15.8 to 44.7% and brackets all four national estimates. Independent interpretation of 702 sampled points gave a commission error of 0.6% for the accepted tree points. The resulting dataset is called Skogmaske.
ER -