17.9.25

Assessing Road Segment Data Quality

This week, I assessed the spatial completeness of two road datasets across a county: the federally maintained TIGER Roads and the local Street Centerlines. Completeness was measured by calculating the total length of roads within a grid overlay and comparing the results for each dataset.

At the county-wide scale, TIGER Roads had a greater total length, measuring 11,382.7 km, compared to 10,805.8 km for the Street Centerlines, a difference of 576.9 km. Using the assumption that more road length equals more completeness, TIGER appears more complete overall. However, more road data does not always mean better quality, as local datasets are often more current and detailed.

To evaluate spatial variation in completeness, both road datasets were clipped to the grid extent, then intersected with the grid to break road segments at cell boundaries. After recalculating geometry, road lengths were summarized per grid cell. These summaries were then joined to the grid, and a percentage difference in completeness was calculated using the local Street Centerlines as the reference, following a method similar to Haklay (2010).

Across the 297 grid cells:
  • 161 were more complete in the TIGER dataset
  • 134 were more complete in the Street Centerlines
  • 2 had equal road lengths
  • 2 had no road data in either dataset

The choropleth map below illustrates the spatial pattern of these differences. Red areas indicate where TIGER data had more road length, blue areas where county Streets data was more complete, and light gray tones show minimal difference. The 2 cells not show, or are the same dark gray as the background, are where there was no road data for comparison.



The analysis may be showing a pattern that while TIGER may be more complete overall, local data can provides more detail in specific areas.

No comments:

Post a Comment