Answers are grounded in this paper and cite their sources.
Coarse raster maps can preserve woody area while altering the lines and field margins represented by that area. We separate the contributions of landscape geometry, threshold ties and aggregation history using 800 synthetic networks, 390 sensitivity realizations, a deterministic strip experiment and a published 2020 shelterbelt classification for Northeast China. The networks represent regular, rotated, irregular and offset layouts. Direct aggregation of 1 m footprints is compared with aggregation through a binary 10 m map, under rules that include or exclude exactly half-occupied cells. In regular networks, the latter pathway gives mean area ratios of 1.325 and 0.661 at 20 m under the two rules; both give 0.980 at 30 m because a nine-cell binary block cannot tie. Axis-aligned 20 and 25 m strips at 30 m support have exact phase-detection fractions of 2/3 and 5/6. Across 18 positive windows retained by the study-boundary screen, pooled 30 m area and skeleton-length ratios are 0.979 and 0.727 relative to the reprojected 10 m classification. Gross mapped losses of 28.10 ha and gains of 25.53 ha explain the small net area change. Dependence between length and field-edge adjacency varies with layout, resolution and rule. These results identify distinct sources of geometric change during map aggregation; they assess transformation of the input classification rather than accuracy against field observations.
No takes yet. Share an insight, caveat, or question.
Zhou et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: