# Seawall Cost Per Linear Foot — Dataset Methodology

**Version:** 2026-07-27  
**Publisher:** Seawall Repair Cape Coral Research is the independent research and reference section of seawallrepaircapecoral.net.  
**Target query:** `Seawall Cost Per Linear Foot`  
**Last verified:** July 27, 2026

## What this release contains

This release keeps four evidence types separate:

1. **★ Official-record-derived permit observations:** 3,199 City of Cape Coral permit records with a positive value recorded in the annual export and an explicit, usable wall length in the official work description.
2. **● Publisher-stated ranges:** 15 reverified cost pages. Their complete range-producing samples and calculations were not available on the checked pages, so their ranges are never pooled with the permit observations.
3. **★ Model-derived unit costs:** state cost totals divided by state seawall mileage from *High Tide Tax*, Table 1. These are derivations from a planning model, not measured residential prices.
4. **★ Documented public projects:** official agency records in which a project value, wall length, or both are published. A per-foot figure is calculated only where the official records tie both quantities to the same project segment.

## Metric definition

The principal local metric is:

`permit-recorded value per stated foot = value recorded in City annual permit export ÷ selected wall length in official work description`

The row-level CSV retains the legacy fields `permit_declared_value_usd` and `permit_declared_value_per_ft_usd` for release compatibility. The exact legal and administrative sequence is more precise than that label: Florida Building Code section 109.3 says the applicant provides an estimated permit value including materials and labor, and the building official sets the final building permit valuation. Cape Coral says it rarely adjusts contractor-provided valuations and uses RSMeans or ICC tables when a significant discrepancy is evident.

The exported value is therefore an administrative permit valuation. It is **not an invoice, bid, executed contract, amount paid, or measured market average**.

Official valuation source: https://www.capecoral.gov/Documents/Document%20Hub/Open%20Government/Building%20Division%20JLAC%20Audit/City%20Response/Assertion%203.pdf?t=202512121723580

## Official source population

The City of Cape Coral lists annual **All Permits Issued** files here:

https://www.capecoral.gov/departments/development_services/building_permit_reports/index.php

| Year | File | Per-foot eligible? | Reason |
|---:|---|:---:|---|
| 2019 | All permits issued 2019.xls | Yes | Contains the permit class, work description, recorded value, and issue date used by the study. |
| 2020 | ALL PERMITS ISSUED 2020.xls | Yes | Contains the permit class, work description, recorded value, and issue date used by the study. |
| 2021 | All permits issued 2021.xlsx | Yes | Contains the permit class, work description, recorded value, and issue date used by the study. |
| 2022 | All Issued Permits 2022.xlsx | No | The annual export does not provide a work-description field from which wall footage can be recovered. A January 2022 year-to-date file is only a partial month and was not substituted for an annual population. |
| 2023 | Permits Issued 2023.xlsx | Yes | Contains the permit class, work description, recorded value, and issue date used by the study. |
| 2024 | 2024 ALL ISSUED PERMITS YTD.xlsx | Yes | Contains the permit class, work description, recorded value, and issue date used by the study. |
| 2025 | Public_All_Permits_Issued_2025_Report.xlsx | No | The annual export does not provide a usable work-description or wall-footage field. |

The five eligible source files were re-downloaded from the City on July 27, 2026. Their SHA-256 hashes match the hashes stored on every released observation. For the 2021, 2023, and 2024 `.xlsx` files, all 1,956 included records were also matched back to the official workbook on permit number, issue date, recorded value, and normalized work description. The 2019 and 2020 legacy `.xls` files were verified by exact file hash in this pass.

## Record flow

| Step | Rows |
|---|---:|
| Candidate rows in eligible permit classes before deduplication | 5,094 |
| Unique year + permit-number records | 4,626 |
| Excluded by one or more automated flags | 1,379 |
| Clean repair/cap-only records set aside | 47 |
| Manually excluded panel-dimension ambiguity | 1 |
| **Final included observations** | **3,199** |

Automated flag counts are non-exclusive because one record can fail more than one rule:

- `no_explicit_length`: 780
- `mixed_scope`: 538
- `alternative`: 478
- `returns`: 61
- `low_relevance`: 59
- `nonpositive_value`: 34
- `ambiguous_tie`: 1

## Candidate classes

The source rows were limited to permit classes labeled `SEAWALL`, `SEAWALL REPAIR`, or `Marine Improvement`. Records were collapsed to unique permit-year and permit-number combinations before summary statistics were calculated.

## Length extraction

A usable wall length had to be:

- explicit in the official work description;
- between 10 and 500 feet;
- written with `'`, `ft`, `feet`, `foot`, `LF`, `LFT`, `linear feet`, `lineal feet`, or a close abbreviation;
- most strongly associated with `seawall`, `sea wall`, or `bulkhead` when the description contained more than one measurement.

The parser rewarded linear-foot units and proximity to the wall term. It penalized area dimensions and measurements near words such as `dock`, `pad`, `patio`, `deck`, `boat`, `lift`, `piling`, `return`, and `cap`. Equal-scoring different measurements were rejected as ambiguous.

The manually excluded record was `WEB20-12915`, whose description, `Seawall replacement 10' Panels`, supplies a panel dimension rather than a total wall length.

## Scope rules

The observation table excludes:

- seawall alternatives and slope/sod alternatives;
- mixed scopes involving docks, boat lifts, patios, concrete pads, decks, pilings, dredging, pools, fences, stairs, pavers, driveways, landscaping, riprap, grading, or sod;
- wall-return descriptions, because one selected measurement may not equal the full wall run;
- repair-only, cap-only, recap, tieback-only, crack, patch, void, joint, injection, grouting, and similar limited-work records;
- records without a positive value;
- records without an explicit usable wall length.

The included scope classes are:

- `new` — description explicitly says new, construct, install, build, or proposed;
- `replacement` — description explicitly says replace, replacement, remove-and-replace, R/R, or rebuild;
- `unspecified_wall` — clean wall-only description with explicit length but no dependable new/replacement label.

## Summary statistics

Annual and scope tables publish sample size, P10, first quartile, median, third quartile, P90, median stated length, and median recorded value. Percentiles use the **R-7 / Excel `PERCENTILE.INC`** interpolation method. No outlier trimming or winsorization was applied.

The field `aggregate_declared_value_divided_by_aggregate_length_usd_per_ft` is retained for completeness but is not the headline statistic because large projects and extreme records can dominate it.

### Annual results

| Permit year | n | P10 | Q1 | Median | Q3 | P90 | Median length |
|---:|---:|---:|---:|---:|---:|---:|---:|
| 2019 | 508 | $110.00 | $115.00 | **$125.00** | $171.67 | $375.00 | 80 ft |
| 2020 | 735 | $115.00 | $115.00 | **$115.00** | $150.00 | $188.76 | 80 ft |
| 2021 | 1,141 | $120.00 | $120.00 | **$130.00** | $165.00 | $193.75 | 80 ft |
| 2023 | 482 | $215.00 | $225.00 | **$285.36** | $337.50 | $498.03 | 80 ft |
| 2024 | 333 | $190.00 | $200.00 | **$250.00** | $312.59 | $473.81 | 80 ft |

## Distribution checks

The following figures reproduce directly from the 3,199-row observation CSV:

- 2,363 records, or 73.9%, have a per-foot value divisible by $5.
- 1,960 records, or 61.3%, state exactly 80 feet.
- In 2020, 348 of 735 records, or 47.3%, equal exactly $115.00 per stated foot.
- Every annual and scope summary row in the summary CSV reproduces from the released observations using R-7 percentiles.

These clustering facts are measurements. Any explanation for the clustering is an editorial inference and must be labeled as such.

## National model derivation

The state unit-cost file uses *High Tide Tax*, Table 1: RCP4.5, year 2040, one-year storm surge, 50th percentile. For each state:

`implied unit cost = published state cost ÷ (published state seawall miles × 5,280)`

The report calls its results planning-level estimates and says they should not replace detailed engineering analysis. The report does not identify its values as “2019 dollars” or provide a price-base quarter. The release therefore labels them **report-published dollars** and applies no inflation escalation.

## Publisher-range ledger

Fifteen pages were opened and reverified on July 27, 2026. For each page the ledger records publisher, geography, page date or vintage, displayed range, scope, material bands, stated inclusions and exclusions, and whether the complete sample and calculations behind the displayed range were disclosed. None of the 15 checked pages disclosed a complete range-producing sample and calculation method sufficient to reproduce a national average. Selected examples on a page are not treated as the full sample behind its range.

The final ledger excludes two pages included in the working draft because their current stated figures could not be reverified during this pass. The final page makes no claim from those pages.

Repeated bands are counted as repeated page statements, not independent measurements. LandPerc's $100-$800 statement is explicitly linked to HomeAdvisor and is not treated as an independent estimate.

## Public-project ledger

Public project records are included only from official agencies. A unit value is calculated only when official records tie both an awarded value and a wall length to the same project segment. Bundled totals that include drainage, roads, landscaping, pump stations, sidewalks, lighting, or other work are not converted into a wall-only price.

## Reproducibility boundary

The released observation CSV is sufficient to reproduce every published included-row statistic, quartile, scope table, clustering percentage, and annual total. The row-level per-foot convenience field is rounded to cents with decimal half-up; summary percentiles are calculated from the unrounded quotient of the two source columns and then rounded the same way. The manifest verifies the exact released files.

Reconstructing the full path from the 5,094 candidate rows to the 3,199 included rows also requires the original annual workbooks, the parser implementation, and the excluded-row log. The original workbooks are linked and hashed; the parser implementation and excluded-row log are not part of release v2026-07-27. The record-flow counts are therefore disclosed as the study's processing audit, while the final included-row statistics are independently reproducible from the release.

## Verification tiers

- **★ official-source-derived:** read directly against an official code extract, fee schedule, government dataset, or public agency project record and reproduced from it.
- **● publisher-stated range:** transcribed from the named publisher's checked page; the page did not disclose the complete sample and calculations used to produce the displayed range.

## Limitations

1. Permit-recorded values are not final paid costs.
2. Only records with an explicit usable wall length can enter the per-foot study.
3. The 2022 and 2025 annual exports are absent from the per-foot series.
4. Administrative descriptions can contain typos, shorthand, scope omissions, or inaccurate values.
5. The dataset does not normalize for inflation, wall height, embedment, material, saltwater exposure, soil, access, demolition, engineering, or exact contract inclusions.
6. No outliers were removed.
7. The 2023 and 2024 exports do not support a dependable residential-versus-commercial split for this filtered sample.
8. Publisher ranges are not pooled or averaged.
9. The High Tide Tax values are model outputs, not measurements, and their price-base quarter is not identified in the report.
10. Among the sources reviewed, no broad public dataset of final paid residential seawall costs was found.

## Files in this release

- `seawall-cost-per-linear-foot-2026-permit-observations.csv`
- `seawall-cost-per-linear-foot-2026-summary.csv`
- `seawall-cost-per-linear-foot-2026-source-range-ledger.csv`
- `seawall-unit-cost-by-state-2026.csv`
- `documented-public-seawall-projects-2026.csv`
- `cape-coral-seawall-permit-fees-2026.csv`
- `seawall-cost-per-linear-foot-2026-data.json`
- `seawall-cost-per-linear-foot-2026-methodology.md`
- `seawall-cost-per-linear-foot-2026-manifest.sha256`