# El Segundo civic atlas — methods and limits

Pilot 01. Reviewed October 30, 2026, America/Los_Angeles. Independent project; not an official City of El Segundo publication.

## The directory

The research seed contained 21 records. This public edition retains 18 entries with retrievable government/operator identity evidence; three pending entries were excluded. Ten records have verified public coordinates. This is a hand-selected pilot, not an inventory of every service within 25 miles.

Every retained entry carries a stable local ID, public source, exact short supporting excerpt, retrieval/check date, and explicit unknowns. Review time does not mean current availability. Source publication dates are unknown unless stated. Unknown is neither zero nor proof of absence. Program restrictions, eligibility, fees and accessibility must be verified with the operator; neither proximity nor a map point establishes access.

The center is the [LA County Services Locator City Hall point](https://locator.lacounty.gov/lac/Location/3177269/el-segundo-city-hall): 33.92044971 N, −118.4153182 W, for 350 Main Street. The source HTML embeds this facility geometry. It is not a surveyed entrance. The ten pins use three published county facility points and seven Census address-range interpolations. Coordinate sources and methods are recorded in the JSON. A North/South street mismatch for the Inglewood center was rejected. Online services and transport service areas do not receive pretend facility pins.

The directory plot uses spherical great-circle distance and initial bearing, with mean Earth radius 3,958.7613 miles, rounded for display. It is a geographic point plot, not a street map or a routing service. Directory points are all well within the boundary; this calculation does not replace the ellipsoidal census boundary calculation.

## Resident portrait

The portrait uses **ACS 2020–2024 five-year period estimates**, not a 2026 headcount, and **whole census tracts chosen by their official internal points**, not an exact clipped circle.

1. Use complete [2024 California TIGER/Line tracts](https://www2.census.gov/geo/tiger/TIGER2024/TRACT/tl_2024_06_tract.zip).
2. Use the fixed county facility center above and a 40,233.6-meter WGS84 geodesic radius (25 miles).
3. Select each whole tract whose TIGER internal point lies within that distance. An internal point is not necessarily the polygon centroid or population-weighted center.
4. Join unique 11-digit tract GEOIDs to ACS summary-file `1400000US` + GEOID rows. Sum only additive cells. Compute percentages from the aggregate numerator and matching universe denominator. Do not average tract medians or tract percentages.
5. The selected geography has 2,087 tracts: 2,034 in Los Angeles County and 53 in Orange County. Selected resident sum is 8,032,239; headlines deliberately say **about 8.0 million**. These counties are not wholly covered.

Population B01001 matches B01003; household B11001 matches income-household B19001 and occupied tenure B25003. The analysis uses B01001, B11001, C16001, B15003, B19001, B25003, B08301 and B23025. Download exact counts, table cells, universes and matched-period LA County, California and US comparisons in [census.json](https://elsegundo.xyz/atlas/census.json).

Sources: [ACS summary-file documentation](https://www.census.gov/programs-surveys/acs/data/summary-file.2024.html); [2024 release schedule](https://www.census.gov/programs-surveys/acs/news/data-releases/2024/release-schedule.html). Bulk tables use `https://www2.census.gov/programs-surveys/acs/summary_file/2024/table-based-SF/data/5YRData/acsdt5y2024-b01001.dat` with the corresponding lowercase table code. Definitions and exact source excerpts are linked in the JSON and page.

### Boundary sensitivity and uncertainty

Using a 2,880-segment circle in a WGS84 azimuthal-equidistant projection, alternative whole-tract selections produce:

| Selection | Tracts | ACS resident sum |
|---|---:|---:|
| Fully contained | 2,024 | 7,750,621 |
| Internal point (published portrait) | 2,087 | 8,032,239 |
| Any intersection | 2,134 | 8,224,125 |

The 473,504-person spread is **selection sensitivity**, not a confidence interval, exact minimum or exact maximum. The all-intersecting selection includes two Ventura coastal tracts. Coastal tracts include water and can extend outside the circle; the selected union extends roughly 31 miles west of the center. No population is allocated by ocean area, and no residential density is calculated from π × 25². A nearby alternative address-interpolated center changes six boundary tracts, which is why the center remains fixed.

Approximate 90% sampling margins use root-sum-square aggregation. Subset proportions use the ACS subtraction formula, with the conservative addition fallback when the variance radicand is negative. The population sampling margin (about ±28,787) is for the selected tract aggregate only. It excludes boundary approximation, center uncertainty, model error and other nonsampling errors. Covariance is not fully captured. Radius and county comparisons overlap; no independent-sample significance claim is made. See [Census guidance](https://www.census.gov/content/dam/Census/library/publications/2020/acs/acs_researchers_handbook_2020_ch03.pdf).

Income is in 2024 inflation-adjusted dollars. Commute measures describe resident workers age 16+ during a period including pandemic years; they are not today's trip or travel-access measures. Language at home does not measure individual proficiency, citizenship or preferences. Degree attainment does not measure all skills. Tenure alone does not establish housing cost burden.

## Separate workplace portrait

LODES 2023 covered primary workplace jobs total **3,691,926** in a distinct block-point approximation. Covered primary jobs by workers' residence total **3,276,519**. Do not add either to ACS population or call the result daytime population.

The analysis selects 70,149 public crosswalk block points by WGS84 geodesic distance ≤40,233.6 meters from the same center. WAC has 50,724 nonzero selected rows; RAC has 57,807. Boundary blocks are wholly included or excluded. Statewide rows matched the crosswalk. LODES uses a different administrative job frame from ACS, including UI/OPM coverage, location imputation and disclosure protection. Primary jobs select one highest-paying covered job per worker; the frame concerns beginning-of-second-quarter employment. These are not all jobs, vacancies, current attendance or a commuting flow. No OD analysis is published. Survey sampling MOEs do not apply; nonsampling and modeling limitations remain.

Public inputs:

- [California LODES8 block crosswalk](https://lehd.ces.census.gov/data/lodes/LODES8/ca/ca_xwalk.csv.gz)
- [2023 primary workplace jobs, WAC](https://lehd.ces.census.gov/data/lodes/LODES8/ca/wac/ca_wac_S000_JT01_2023.csv.gz)
- [2023 primary residence jobs, RAC](https://lehd.ces.census.gov/data/lodes/LODES8/ca/rac/ca_rac_S000_JT01_2023.csv.gz)
- [LODES technical documentation](https://lehd.ces.census.gov/doc/help/onthemap/LODESTechDoc.pdf)

Industry concentrations compare the same 2023 primary-job definition against California. Location quotient is the selected area's job share divided by California's job share. It is descriptive, not a statistical test, growth measure or promise of business advantage. Broad information jobs are not all film or technology; broad manufacturing cannot isolate aerospace.

## Improvement benchmarks and future work

Source and coordinate counts describe this pilot only. Full regional coverage needs an official deduplicated denominator. Applicable verified hours, costs, eligibility and access fields should be measured separately; explicit null is not a completed field. Service capacity, wait times, outcomes, walking/transit accessibility and causal benefits have not been measured.

Future work includes block-weighted clipping with vintage crosswalks, category-specific denominators and licensed public tree inventories. Block population weights cannot automatically model age, language, income or household characteristics. Historical comparisons need matched geography and dollar normalization. No merchant or vehicle enrollment, physical tagging, private asset collection, agency outreach or service commitments occurred.

## Reproduction and review

Original research computations were performed from the cited public bulk files. OpenAI GPT-6 Astra independently downloaded the public inputs and reproduced all 76 indicator/geography results, the tract selections and boundary sensitivity, and the LODES totals and published industry comparisons. [Read the review](https://elsegundo.xyz/atlas/reproduce/review.json) and [download reproduction instructions](https://elsegundo.xyz/atlas/reproduce/README.md). The bundle includes Python scripts, selected tract GEOIDs and compact results with local source hashes. The review records the scope, corrections and limits; it does not certify exact circle coverage. Original research used WGS84; the independent GRS80/NAD83 check produces the same selected totals. Raw bulk source files are linked rather than republished.