Citipax Landfill Capacity Density — CLCD v1
CLCD v1 is the empirical solar capacity density of closed-landfill caps — DC megawatts per acre — published on two bases: per acre of built array footprint and per acre of total capped area, derived from measured built New Jersey landfill-solar projects.
Coefficient v1 — frozen 2026-06-26 (engine model ground-mount-capacity-v1) · statewide envelope built on it: flagship v1.1, revised 2026-07-10 · symbol ρCLCD · maintained by Citipax Atlas
Why two bases, and why this name. Most public confusion about landfill solar comes from quoting an array-packing number as if it were a whole-cap number — so the standard is the two-basis structure, not a single figure: CLCD-cap for total capped acreage, CLCD-array for built array footprint. The two differ by roughly 2× by construction and are not interchangeable. Alternatives considered and rejected: “NJ-CLCD” (the coefficient derives from New Jersey built projects but is not definitionally NJ-bound — the envelope is NJ-specific, the coefficient is portable) and “Citipax Cap Coefficient” (loses the density framing that ties it to the Bolinger benchmark line).
1. The coefficient
Headline values, exactly as the receipts support:
| Basis | Value | Support |
|---|---|---|
| CLCD-array (built-array basis) | 0.25 MW_dc per built array-acre (median; ≈4.0 ac/MW_dc) | n = 11 unique built projects; range 0.169–0.40; cross-checked against an independent n = 75 median of 0.235 |
| CLCD-cap (whole-cap basis) | ≈0.20 MW_dc per total-capped acre = 0.50 cap buildable-fraction × 0.40 MW_dc/array-acre packing | Cap buildable fraction empirical band 0.45–0.55, with conservative whole-property floors of 0.32 (GEMS) and 0.37 (Kearny 1A) |
| AC equivalent | implied ILR ≈1.2 → ≈0.16–0.17 MW_ac per capped acre | derived from the above |
Receipts: capacity-coefficient research memo, 2026-06-26 (full per-project evidence table, 66 rows — ~30 unique projects after source deduplication: MW / acres / MW-per-acre / basis) · engine calibration register, “Ground-mount capacity — non-farmland.”
Engine constants
Model ground-mount-capacity-v1: mw_dc = area_acres × usable_fraction × 0.40 MW_dc/acre; mw_ac = mw_dc / 1.20; usable_fraction 0.50 default / 0.55 ceiling, constraint-derived clamp [0.05, 0.55].
Receipts: engine calibration register, “Ground-mount capacity — non-farmland.”
Confidence, stated plainly
This is a screening coefficient, not an engineered yield. n = 11 supports roughly one significant figure of confidence. The cap usable fraction (0.45–0.55 band) is the dominant swing variable, and only a site-specific cap and gas-infrastructure survey resolves it.
Receipts: capacity-coefficient research memo, 2026-06-26 (dossier-framing section) · engine calibration register.
Relation to the published open-field benchmark
The reference line for utility-scale PV density is Bolinger & Bolinger 2022 (“Land Requirements for Utility-Scale PV: An Empirical Update on Power and Energy Density,” IEEE Journal of Photovoltaics 12(2)): from 736 plants (35,482 MW-DC, 38 states, 2007–2019, ~92% of the U.S. utility-scale fleet), median power density 0.35 MW-DC/acre fixed-tilt (2.8 ac/MW) and 0.24 MW-DC/acre tracking (4.2 ac/MW); energy density 447 / 394 MWh/acre. CLCD v1 is a ~29% derate vs. Bolinger open-field on the array-footprint basis (0.25 vs 0.35) and ~43% on the whole-cap basis (0.20/0.35 = 0.57).
Receipts: academic literature map, 2026-06-25 (Bolinger 2022 replaces Ong 2013 as the density authority) · derate arithmetic in the calibration register and the 2026-06-26 coefficient memo.
2. Derivation
Each step below carries its receipt. The chain runs: registry → LiDAR cap-surface measurement → usable acres → density coefficient (benchmarked against Bolinger 2022) → two-sided capacity envelope → decay-audited net bracket.
Step 1 — Site universe. All 412 closed landfills in the NJDEP solid-waste registry (15,042 footprint acres; extract 2026-06-15).
Receipts: flagship study memo, 2026-07-02 — published as the statewide study.
Step 2 — LiDAR cap-surface measurement (model landfill-envelope-v0). Slope and aspect are computed for every site from the statewide LiDAR-derived 10-ft DEM (NJDEP/NJOGIS “statewide_2021,” EPSG:6527; source flights 2014–2021 by county) using Horn’s 1981 kernel. Cells classify prime (≤8%), conditional (8–15%), or excluded (>15% — the side-slope regime, typically 25–33% and never arrayed).
Receipts: flagship study memo, 2026-07-02 (method sections).
Step 3 — Usable acres. The buildable candidate is filtered to 8-connected contiguous components ≥2 acres, flagged (never dropped) against a 5-acre developable floor, and mild aspect factors above 5% slope (S 1.00 / E–W 0.95 / N 0.85) yield effective acres. Statewide: 11,439 slope-screened buildable acres; 294 of 412 sites (71%) meet the 5-acre floor. Distributional stats across all 412 sites: implied buildable fraction (envelope ÷ footprint, pre-aspect) median 0.815, IQR 0.566–0.928; 47 sites with zero buildable cells; 118 below the 5-acre floor. Sites without DEM coverage read “not measured — never zero.”
Receipts: flagship study memo, 2026-07-02 (§5.3, §3.7, §5.5) · flagship data pack summary.json.
Step 4 — Apply density: CLCD v1, benchmarked to Bolinger 2022. Because no landfill-specific density figure existed in the literature (§5), the coefficient was derived 2026-06-26 from measured built NJ projects (§1) and benchmarked against the Bolinger 2022 open-field medians. Capacity is always published as a two-sided bracket:
- anchor = total footprint acres × 0.20 MW_dc/capped-acre (CLCD-cap, which bundles infrastructure/setback/slope exclusions into its 0.50 buildable fraction), and
- bound = measured effective envelope acres × 0.40 MW_dc/array-acre empirical packing density.
The two are deliberately not like-for-like; reality lives between them.
Receipts: flagship study memo, 2026-07-02 · capacity-coefficient memo, 2026-06-26 · engine calibration register.
Step 5 — Statewide envelope (pre-decay). Gross 3,008–4,489 MW_dc; net of 17 known-built sites (EPA RE-Powering name-match, December 2024 matrix) 2,769–4,151 MW_dc. Built-to-bound ratios on the 17 built sites: 0.03–1.10, median 0.21. Per-county anchor/bound MW for all 21 counties published in the study’s Table 5.1 and the open CSV.
Receipts: flagship study memo, 2026-07-02 (§5.1, §6.1, §5.5, Table 5.1) · flagship data pack summary.json, county_table.csv, epa_match_review.csv.
Step 6 — Decay audit → current public bracket. A 2026-07-10 stratified imagery audit (71 sites: census 31 + random 40; per-site consumed fractions in 0.1 steps) measured ≈300 MW_dc consumed by post-LiDAR redevelopment (≈6–8% of the gross bound), triggering the ratified decision rule. Verbatim from the decay memo: “Revised net bracket = published × (1 − decay fraction): floor 2.8 × (1−0.067) ≈ 2.61; bound 4.2 × (1−0.067) ≈ 3.92 → publish 2.6–3.9 GW_dc.” The carve-out multiple becomes 9–13×.
Receipts: revision decision — decay-check memo, 2026-07-10 (“The decision rule, applied”; headline: “Revised bracket ≈2.6–3.9 GW_dc”) · public declaration — the study page (dek, “Revision, July 10, 2026” box, and conclusion).
Validation, three independent ways
(1) Built precedent: Mount Olive — 23.2 MW screen bound vs 25.6 MW built. (2) Resolution: 11×-finer direct point-cloud extraction — 408 of 412 sites carry a fine-grid envelope (4 sites fall outside all LiDAR tile coverage and are reported unmeasured); 361 site pairs, r = 0.998, fine grid mean 11% tighter, rank order preserved. (3) Time: two-epoch settlement grading — American Cyanamid’s 462-acre cap, +0.03 ft median over eight years; 17 Somerset sites graded in the pilot.
Receipts: flagship study memo, 2026-07-02 (v1.1 validation sections).
Reproducibility
Everything reproduces from versioned model outputs — constants and commit hash snapshotted per row.
Receipts: flagship study memo, 2026-07-02.
3. What CLCD is not — limits carried in full
CLCD v1 and the envelope built on it are screening instruments. Every published use must carry these limits:
- Not an engineered yield. The coefficient is a screening input: n = 11, ~1 significant figure of confidence; the cap usable fraction (0.45–0.55 band) is the dominant swing variable and only a site-specific cap/gas-infrastructure survey resolves it.Receipts: capacity-coefficient memo, 2026-06-26 · engine calibration register.
- Envelope MW is an upper bound by construction. The LiDAR screen measures slope and contiguity only — gas wells/flares, leachate infrastructure, cap-edge setbacks, access roads, and stormwater are invisible to it. The anchor-bound gap (1,481 MW gross) is the study’s own estimate of everything the screen cannot see.Receipt: flagship study memo, 2026-07-02 (§4.1.5).
- Proximity ≠ interconnectability. The grid-proximity layer is rank-only; no envelope or CLCD figure asserts interconnectable capacity. Interconnection/hosting-capacity reality is the field’s documented blind spot.Receipts: interconnection-layer design note (“Proximity != interconnectability”) · academic literature map, 2026-06-25 (S4 gap #5) · no-warranty framing in the 2026-07-01 market-research memo.
- DEM vintage varies by county (source flights 2014–2021; Bergen derives from 2014): envelopes reflect the flight date, not the present. The 2026-07-10 freshness gate found 11 of 31 priority sites materially changed or compromised by post-flight construction; a current-imagery check is standing pre-commitment procedure, and figures carry verification-tier labels (model-claimed / record-verified / ground-truthed). Sites without DEM coverage read “not measured — never zero.”Receipt: freshness-gate memo, 2026-07-10.
- Registry-polygon semantics. The 1,114-acre Leesburg polygon (443 MW_dc, 9.9% of the statewide bound) may map a property boundary rather than fill extent; excluding it entirely lowers the statewide bound to 4,046 MW_dc. Terrain confirmed physically flat by fine-grid extraction; boundary provenance remains unverified.Receipt: flagship study memo, 2026-07-02 (§7.6.1).
- Known-built netting is conservative. Name-matched against EPA’s December 2024 matrix; projects completed 2025–26, buffer-sited arrays, and unreported small systems are not subtracted — plausible “remaining” overstatement of tens (not hundreds) of MW.Receipts: flagship study memo, 2026-07-02 (§7.6.3) · flagship data pack epa_match_review.csv.
- The decay estimate is itself screening-grade. Stratified 71-site imagery audit, per-site consumed fractions in 0.1 steps, extrapolated remainder ≈94 MW with a crude 95% band of ≈40–150 MW; statewide honest band ≈250–360 MW consumed.Receipt: decay-check memo, 2026-07-10.
- Aspect weighting factors are judgment constants (S 1.00 / E–W 0.95 / N 0.85 above 5% slope) — the model’s weakest — with <2% leverage on statewide totals.Receipt: flagship study memo, 2026-07-02 (§7.6.5).
- Settlement grades are prioritization signals, never geotechnical determinations. NJDEP requires NJPE-sealed settlement and side-slope analyses at permitting regardless.Receipt: flagship study memo, 2026-07-02 (§4.4.3).
- Model lineage is disclosed. The original farmland-only suitability model read municipal landfills as 0 MW;
ground-mount-capacity-v1(the empirical landfill coefficient) closed that gap — v0 had used a generic 0.65 usable × 0.35 Bolinger open-field placeholder.Receipt: engine calibration register (non-farmland section). - Privacy wall. Publication is county-aggregate plus public NJDEP registry site identities with physical measurements only — no parcel identifiers, no block/lot, no owner names, and no distress signals attached to named sites; machine PII sweeps run on every output.Receipt: flagship study memo, 2026-07-02 (§4.7).
4. Versioning and corrections policy
Versioning. CLCD carries a major version (v1); the statewide envelope carries its own study version (flagship v1.1). Any change to the coefficient values, the buildable-fraction band, or the anchor/bound construction increments the CLCD version; envelope revisions (e.g., decay adjustments) revision-stamp the bracket without changing the coefficient. Model constants and commit hashes are snapshotted per output row.
Corrections are carried openly — the correction is the credibility story. The published bracket of 2.8–4.2 GW_dc (flagship v1.1, 2026-07-02) did not survive the 2026-07-10 redevelopment-decay audit (≈300 MW_dc consumed post-LiDAR, ≈6–8% of the gross bound) and was revised to 2.6–3.9 GW_dc. The public study page carries the revision box verbatim: “Revision, July 10, 2026 — bracket updated 2.8–4.2 → 2.6–3.9 GW_dc.” This history is a permanent part of the record, not an erratum to be buried; where the superseded figure survives in earlier internal drafts and marketing copy, it is corrected — never silently.
Open aggregate. The county-aggregate table (anchor/bound MW for all 21 counties) is published as open data under CC-BY 4.0: use it, cite it (§6), and send corrections — corrections are invited and credited. The published bracket is declared authoritative for the open county-aggregates CSV. Email ben@citipax.org with the figure and your source.
What is never published: parcel identifiers, block/lot, owner names, or distress signals attached to named sites (§3, limit 11); machine PII sweeps run on every output.
5. The literature gap
The authoritative empirical density reference for U.S. utility-scale PV is Bolinger & Bolinger 2022 (IEEE Journal of Photovoltaics 12(2)): 736 plants, 35,482 MW-DC, 38 states, 2007–2019, ~92% of the U.S. utility-scale fleet — median 0.35 MW-DC/acre fixed-tilt and 0.24 MW-DC/acre tracking. That study is fleet-wide and open-field; it publishes no landfill-specific capacity-density figure. Within the Citipax academic literature map’s sweep, no published landfill-specific capacity-density coefficient was identified — which is why the engine derived its own on 2026-06-26 from measured built New Jersey projects. The claim is scoped to that sweep: “no landfill-specific capacity-density coefficient has ever been published” as surveyed, not a claim of exhaustive global absence. The same map documents interconnection/hosting-capacity reality as the field’s blind spot (S4 gap #5), which is why CLCD is bounded away from any interconnection claim (§3, limit 3).
Receipts: academic literature map, 2026-06-25 (density authority; S4 gap #5) · capacity-coefficient memo, 2026-06-26 (derivation rationale, verbatim: “no landfill-specific capacity-density coefficient has ever been published”; derived 2026-06-26).
6. How to cite
The coefficient:
Citipax Atlas. Citipax Landfill Capacity Density (CLCD) v1: an empirical solar capacity-density coefficient for closed-landfill caps. 2026. CLCD-array = 0.25 MW_dc per built array-acre (median, n=11; range 0.169–0.40); CLCD-cap ≈ 0.20 MW_dc per total-capped acre (0.50 cap buildable-fraction × 0.40 MW_dc/array-acre packing). Derived 2026-06-26; screening-grade (~1 significant figure). https://citipaxatlas.com/research/landfill-solar
The statewide envelope built on it:
Citipax Atlas. Remaining Landfill Solar Potential in New Jersey (flagship study v1.1, 2026-07-02; bracket revised 2026-07-10). Net remaining statewide envelope: 2.6–3.9 GW_dc across 412 closed NJDEP-registry landfills (revised from the published 2.8–4.2 GW_dc after the July 2026 redevelopment-decay audit). County-aggregate data: CC-BY 4.0.
Shorthand: “CLCD v1 (Citipax, 2026)”. When quoting a single number, quote the basis: CLCD-cap for whole-cap acreage, CLCD-array for array footprint — the two differ by 2× by construction and are not interchangeable. Any citation of the statewide bracket must use the current 2.6–3.9 GW_dc figure and may note the open revision history (§4). Benchmark attribution: density benchmark per Bolinger & Bolinger 2022, IEEE Journal of Photovoltaics 12(2).
Receipts live in the engine’s versioned internal research memos (dated 2026-06-25 through 2026-07-10), the calibration register, and the flagship study data pack; the public receipts are the statewide study and the open county-aggregates CSV. This page is method and evidence; it is not investment, engineering, or legal advice.
Atlas services are provided by Citipax Atlas LLC. Screening signals are rank-grade prioritization for site selection — not a pro-forma, appraisal, valuation, or interconnection study.