Figure 10a (animation)

1:250K source generalized by RDP (left) and QTM (right)

click on top figures to re-animate, bottom figures for metadata

Nantucket MA 250K data RDP v. QTM gen.

Nantucket MA 250K data @ 1:250K

View Companion Figure 10b

back to paper

METADATA:

DATA: 1:250,000 NOAA/NOS digital nautical charts, vector coordinates in latitude and longitude

SOURCE: Coastline Extractor -- http://crusty.er.usgs.gov/coast/getcoast.html

EXTENT: 70¤ 14' W - 69¤ 58' W (W-E); 41¤ 14' N - 41¤ 23' N (S-N)

WIDTH: 22.00 km; HEIGHT: 17.50 km; AREA: 131.7 km2

LENGTH: 102.5 km; POINTS: 525; AVG SEG LEN: 0.2 km

PROJECTION: None for QTM generalization; plat carré (cosine) for RDP generalization and displayed maps.

INSETS: The smaller maps are drawn to the scale at which the QTM generalization is intended to be displayed. The scale you see depends on monitor resolution (images were produced at 75 DPI).

REMARKS: Data vectorized by NOS from digital radar imagery, conditioning unknown. The 1:250K scale specified for it implies a resolution between 100 and 200 m (assuming a map resolution of 0.5 mm). As the average line segment length in the dataset is 200 m, this seems reasonable.

In generalizing the data, RDP (Ramer-Douglas-Peucker algorithm) tolerances ( 100, 205, 320 m) were chosen to yield equal numbers of points as did the corresponding three levels of default non-hierarchical QTM (quaternary triangular mesh) generalization (levels 16, 15, 14 -- representing 150, 300 and 600 m ground resolution respectively).

No special treatment of local sinuosity was applied in these experiments. The sinuosity statistic gives the path-to-trendline ratio, averaged across the set of selected points (on a scale of 1 - 7). Note that the sinuosity of RDP solutions is consistently higher than that of QTM solutions; this is indicative of the "critical point" bias of RDP.

back to paper