SF HSH/DPH & 90-Day Drug Calls — Spatial Comparison

Earlier wide-view observation

August 2026 post comparing citywide HSH/DPH and 911sf drug-call maps

Why the apparent center changes with scale

The two highlighted rectangles are best understood as views of a broader contiguous central-city concentration, not necessarily as isolated clusters. At tight radii, the Tenderloin emerges as the densest core. As the measurement window expands, the optimal center shifts toward Civic Center/Mid-Market because the larger window increasingly captures Tenderloin, Mid-Market and SOMA together.

The radius analysis is orientation-neutral: circles do not depend on how a rectangle is rotated. That provides a useful check on the rectangle comparison.

Important: the map measures spatial association. It does not establish that HSH/DPH properties caused drug calls. Calls may come from providers, residents, staff, neighbors or others, and the dispatch data identify intersections rather than caller origins.

431

HSH/DPH records

2,710

drug calls

658

intersections

Map layers

BasemapProximity radius

Concentration rectangles

#1 same-shape HSH/DPH maximum
70 properties (16.2%)
361 drug calls (13.3%)
19 drug-call intersections

#2 Larkin–Jones / Golden Gate–Post
52 properties (12.1%)
567 drug calls (20.9%)
20 drug-call intersections

Call intensity vs nearest property

Calls/int.	n	Median	≤500 ft
1 call	349	706 ft	36.4%
2–3	163	427 ft	57.1%
4–9	92	332 ft	72.8%
10+	54	194 ft	90.7%

Interpretation: this shows spatial association only. Dispatch records identify an intersection, not the caller's exact location or the exact location of the underlying activity. Calls may originate from residents, staff, neighbors, or others. Proximity does not establish causation.

Method and recalculated rectangle results

The rectangles shown on the interactive map use the embedded coordinates in this file. Recalculation gives #1: 70 HSH/DPH records, 361 drug calls at 19 unique call intersections, and #2: 52 HSH/DPH records, 567 drug calls at 20 unique call intersections. The earlier illustrative graphic used preliminary rectangle figures; these are the figures tied to the actual polygons in this HTML.

The 431 count is a record/program count and can include multiple records at the same physical address. A future unique-address version should be shown alongside it. Drug calls are aggregated to 658 distinct geocoded intersection points in this HAR snapshot.

Social cards

Corrected cards use the verified 90-day HAR query window: July 3–Oct. 1, 2026. #2 uses the final polygon figures: 52 HSH/DPH records, 567 drug calls, 20 unique call intersections.

$Spatial analysis dashboard card 1$