A useful measure must be testable
A new measure is not credible merely because its map looks plausible. Readers need to know how the values are calculated, how the boundary is produced, whether the result can be reproduced, how sensitive it is to analytical choices, and whether the method still discriminates when moved to another city.
The Structural Intensity technical supplement addresses those questions directly. Its purpose is not to repeat the Washington monograph, but to expose the method to tests that could reveal instability, overfitting, or hidden discretion.
Begin with rules that do not move
Structural Intensity is estimated building-footprint coverage multiplied by normalized building height. A native raster cell qualifies when its value is at least 0.007987. Qualifying cells connect through strict eight-neighbor adjacency, and the largest connected component becomes the primary Structural Core.
The workflow does not smooth boundaries, bridge interruptions, merge components manually, or fill gaps. Calculations use full floating-point precision before thresholding. Those fixed rules make it possible for another analyst to determine whether the same inputs produce the same topology.
Reproduce the baseline exactly
The first test is reconstruction. For Washington, the documented workflow reproduces 9,014 qualifying cells and 992 strict eight-neighbor components. The largest component contains 3,957 cells and covers 13.978408 square kilometers, or 5.397094 square miles.
Exact reproduction matters because a visually similar map is not enough. A change in precision, connectivity, clipping, or component handling can alter boundaries and relationships even when the overall image appears familiar.
Test whether the threshold sits on a cliff
A threshold would be fragile if a small numerical change abruptly reorganized the principal geography. A fine sweep identified a local Washington stability regime from approximately 0.00690 to 0.00911. Its midpoint is about 0.008005—within roughly 0.23 percent of the fixed 0.007987 threshold.
Across the broader ±20 percent sensitivity test, the Structural Core remains the largest component and remains separate from the principal Southwest and Navy Yard–Capitol Riverfront areas. Connecting the Structural Core to Southwest requires a reduction of about 33.7 percent, outside that systematic test band. Some smaller component relationships are more sensitive, and the supplement reports those rather than hiding them.
Move the same method to a different city
A Washington-specific rule might simply reproduce an intuitive picture of Washington. The stronger test is to apply the same absolute threshold and connectivity rule elsewhere without city-specific adjustment.
In Philadelphia, the unchanged method identifies a 7.866-square-mile Structural Core, a separate 5.696-square-mile Center City–South Philadelphia Structural Area, and a separate 5.089-square-mile West Philadelphia Structural Area. It also preserves a localized interruption near the Vine Street Expressway rather than manually bridging it. The result looks distinctly Philadelphian, which is precisely what a transferable measure should allow.
State the limits with the result
The fixed value is empirically supported, but it is not presented as a uniquely optimal universal constant. Its defense is cumulative: documented provenance, a local stability regime, exact reconstruction, sensitivity testing, comparison with external building-footprint data, and cross-city differentiation all point in the same direction.
Structural Intensity is also an all-building measure. It does not classify land use, population, neighborhood identity, transportation access, or economic integration. It provides a reproducible structural reference against which those other systems can be compared.