Climate Data Licence Clearance — Investigation Report¶

Sub-chat deliverable — Session 17 prerequisite Scope: Licence clearance for every candidate climate / cooling-load data source the energy and lca.b6 modules could draw on, for the specific use case of committing values as static data in an AGPL-3.0-or-later open-source repository, distributed publicly via GitHub and served at runtime. Companion to EPD source licence clearance, which covers the same question for EPD programme operators. Date: 2026-08-29 · Revised: 2026-09-27 — the account of the CLTD/CLF/SCL method was corrected after Chiwara challenged it; see §9. Not legal advice. This is a technical and licensing assessment prepared for a decision session. The items in §5 need a qualified lawyer.

TL;DR¶

The licence is the blocker: ASHRAE CLTD/CLF/SCL table values cannot go into the public AGPL repository. ASHRAE asserts full copyright over the Handbook and grants only non-transferable permissions. No ASHRAE licence exists for these tables — its paid Climatic Design Data License for Software Developers ($2 000 one-time) covers only Chapter 14 climatic design data, and even that grants "the purchaser a non-exclusive license". Any permission ASHRAE did give would carry restrictions, and AGPL-3.0 §10 forbids imposing further restrictions on recipients, so a permission would not make the combined work distributable. Government reproductions of the tables do not launder the copyright, and no open equivalent of the tables exists.
The tables answer a different question — the lead technical reason. CLTD sizes cooling equipment for a design day. ASHRAE's own 1997 text reserves annual energy estimation for hour-by-hour methods. EN 15978 defines B6 as the operational energy used across the study period, in kWh. The reference workbook reaches kWh only by multiplying one CLTD day per month by 19–22 days (§3a).
It is also the least accurate of the established methods. Mao, Baltazar & Haberl (2019) rank CLTD last of five methods; in hot climates it over-predicts by roughly 20 % (Hong Kong) to 38–79 % (Malaysia). See Finding 7.
Superseded, not retired. RTS superseded CLTD/CLF/SCL in the 2001 Handbook–Fundamentals. The method was never formally withdrawn and is still in wide practical use.
The adopted method is an independently implemented EN ISO 52016-1 hourly kernel (AR-64), with coefficients derived in project code from layer properties. Methods are not copyrightable.
NASA POWER is the vendorable climate source; ERA5/Copernicus and PVGIS are the attribution- bearing reserves. All three are continuous global gridded datasets. That replaces the latitude-band lookup in the reference workbook's own LM Lat- month corr sheet — which pasted the 12°N block for a 14.7°N site instead of interpolating (AR-62) — with interpolation between grid points. Do not vendor climate.onebuilding.org ("All Rights Reserved") or Meteonorm (redistribution contractually prohibited, CHF 3 000-per-instance penalty clause).

§1 Question under investigation¶

Can values derived from ASHRAE Handbook CLTD/CLF/SCL cooling-load tables (which the reference workbook indexes by latitude band) be committed as source-controlled data files in this repository (AGPL-3.0-or-later), distributed publicly via GitHub, and served at runtime to drive the energy module? For each candidate source, what is the operative licence, what does it permit, what does it prohibit?

The operative distinction is the same one the EPD clearance draws:

Vendoring — committing the values as static data in the repository, redistributable to anyone who clones it.
Runtime / cited-only — loading values at runtime from a licensed local copy or the source's own service, with a citation but no static copy committed.

AGPL adds a third and decisive test the EPD clearance did not have to apply, because ÖKOBAUDAT's DL-DE-BY-2.0 passes it trivially:

Downstream parity — every recipient of the conveyed program must receive the same freedom to run, study, redistribute, and modify. A licence that grants rights to this project only fails, however generous those rights are to this project.

§2 Scope and method¶

Each source's terms of use, data licence, and/or software-developer licence were reviewed as published at the time of this investigation (2026-08-29). Where terms were absent or ambiguous, the conservative interpretation (not cleared) is applied — the same rule the EPD clearance uses.

Copyright doctrine was reviewed to assess residual risk in reproducing table values versus re-expressing them algorithmically: US law via Feist; EU law via Directive 96/9/EC and the BHB / Football Dataco line; and the ASTM v. Public.Resource.Org precedent, in which ASHRAE was itself a plaintiff.

§3 Key findings¶

ASHRAE's copyright assertion is explicit and broad. The Handbook–Fundamentals states that "no part of this book may be reproduced without permission in writing from ASHRAE… nor may any part of this book be reproduced, stored in a retrieval system, or transmitted in any form or by any means… without permission in writing from ASHRAE." Reuse runs through permissions@ashrae.org, which issues per-request reprint permissions and assumes the requester already owns the publication. A February 2025 AI policy additionally prohibits creating derivative works of ASHRAE IP using AI tools without express written permission.
The relevant paid product exists but does not solve the AGPL problem. The ASHRAE Climatic Design Data License for Software Developers ($2 000 one-time) grants, at §1.1, "the purchaser a non-exclusive license to use the text, tables and data from ASHRAE Research Project RP-1923… in derivative software, with no end date for the use of this information", with attribution "Climatic Design Data © 2025 ASHRAE, www.ashrae.org. Used with permission." Two things defeat it here: (a) it covers Chapter 14 climatic design conditions — dry-bulb, dew-point, wet-bulb, wind — not the CLTD/CLF/SCL cooling-load tables; and (b) it is non-exclusive to the purchaser, so downstream users who clone an AGPL repo receive nothing.
The legal background helps in the US and is uncertain in the EU. Under Feist Publications, Inc. v. Rural Telephone Service Co., 499 U.S. 340 (1991), facts are not copyrightable — "originality, not 'sweat of the brow,' is the touchstone of copyright protection" — and only original selection and arrangement of a compilation attracts thin copyright. Extracted numeric values re-expressed as regression or interpolation coefficients are therefore materially safer than reproducing literal tables. But the project's EU footprint brings the sui generis database right (Directive 96/9/EC) into play. British Horseracing Board v William Hill and Football Dataco limit that right to investment in obtaining or verifying data, not creating it — which helps, because CLTD/CLF/SCL values are computed outputs of ASHRAE's RP-472 methodology, whereas Chapter 14 design conditions are obtained and verified from weather-station records and are the stronger candidate for protection. Favourable, not free of doubt. Note too that contract can bind where copyright does not: ProCD, Inc. v. Zeidenberg, 86 F.3d 1447 (7th Cir. 1996).
The ASTM v. Public.Resource.Org line does not help. In 82 F.4th 1262 (D.C. Cir. 2023) the court held that "the non-commercial dissemination of such standards, as incorporated by reference into law, constitutes fair use", finding the use transformative — providing "the public with a free and comprehensive repository of the law". The Handbook is not a standard incorporated into law, and AGPL distribution inside a tool is neither non-commercial nor a repository-of-the-law use. Do not rely on it.
CLTD/CLF/SCL was superseded in 2001 by RTS; it was never formally withdrawn and is still in wide practical use. Per Spitler, Fisher & Pedersen, "The Radiant Time Series Cooling Load Calculation Procedure" (ASHRAE Transactions): the RTS method "effectively replaces all other simplified (non-heat-balance) methods such as the cooling load temperature difference/solar cooling load/cooling load factor method (CLTD/SCL/CLF)", and from the 2001 Handbook–Fundamentals RTS and Heat Balance are the methods ASHRAE presents. "Superseded" is the accurate word, not "deprecated" or "retired": practitioners still use CLTD, which is why this report does not rest on its status. Status alone would not justify leaving it; the licence (Findings 1–2), the quantity it computes (§3a) and its accuracy (Finding 7) do. Coefficient-based replacements — conduction transfer functions, response factors, and the RC node network of EN ISO 52016-1 — are derivable from first principles and documented in open sources such as FastCTF and the EnergyPlus Engineering Reference. Nothing in that chain requires an ASHRAE table.
Genuinely open replacement data exists, and it fits Sub-Saharan Africa better. NASA POWER is free, global, continuous, and carries no copyright restriction under NASA's open-data policy. ERA5 via the Copernicus Climate Data Store permits redistribution and commercial use with attribution to C3S/ECMWF. Both are meteorological and solar datasets rather than pre-computed cooling loads, so they entail the methodology change of Finding 5 — and both cover African locations where ASHRAE station density is thin.
CLTD is the least accurate of the established cooling-load methods, and it over-predicts in hot climates. Mao, Baltazar & Haberl (2019), "Comparison of ASHRAE peak cooling load calculation methods", Science and Technology for the Built Environment 25(2):189–208, rank CLTD last of the five methods they compare, behind Heat Balance, RTS, the Transfer Function Method (TFM) and TETD/TA. Hot-climate studies report over-prediction of about 20 % in Hong Kong (Mui & Wong 2007, Building and Environment 42:2498–2504) and 38–79 % in a Malaysian study. The venue of the Malaysian study has not been traced; cite that range as unverified until it is. This is what the earlier claim that "the accuracy improves" rested on. It was asserted without a citation until the 2026-09-27 revision.

§3a The tables answer a different question¶

Added 2026-09-27. This is now the lead technical reason; the licence (Findings 1–2) remains the blocker.

What CLTD computes. CLTD/CLF/SCL is a design-day method. It sizes cooling equipment by estimating the peak load on a representative day, using a precomputed temperature-difference profile for that day.
What ASHRAE says it is for. ASHRAE's 1997 Handbook–Fundamentals separates load calculation for equipment sizing from energy estimation, and reserves annual energy estimation for hour-by-hour methods.
What B6 requires. EN 15978 defines module B6 as the operational energy use of the building over the study period, reported in kWh. That is an annual energy quantity, summed across the study period, not a peak load.
What the reference workbook does. 1c_energetic_analysis.xlsx computes one CLTD day per month (sheet Calcul par mois) and multiplies it by 19–22 days (Données!K3:K14) to reach 14 396 kWh a year, or 49.8 kWh/m² (Données!K15:K16). That turns a design-day profile into an annual total, which ASHRAE's own guidance does not support. Its per-month daily totals also show the site's cooling peak in April–May, not July — see §9.

So even a lawful, perfectly licensed set of CLTD tables would compute the wrong quantity for B6. This is why the method chosen is an hourly EN ISO 52016-1 kernel run across 8 760 hours (AR-64), and why RTS / Heat Balance — equally clear on licence, and more accurate than CLTD — was also rejected: RTS still yields a repeating design day.

§4 Verdict table¶

Statuses follow the EPD clearance convention: CLEARED — values may be committed to the public repository; RESTRICTED — usable at runtime with citation but not committed; BLOCKED — do not vendor, clearance not obtainable or not obtained.

Source	Status for AGPL vendoring	Operative licence	Cost	Drop-in vs methodology change	AGPL downstream parity?	Notes
ASHRAE CLTD/CLF/SCL tables (Handbook–Fundamentals)	BLOCKED	All-rights-reserved; per-request written permission only	Permission fee varies; requires owning the publication	Drop-in against the current data model	No — permission is non-transferable, and any permission would carry restrictions AGPL §10 forbids passing on	Superseded in 2001 by RTS, never formally withdrawn, still in wide practical use (Finding 5). Computes a design-day peak, not the B6 quantity (§3a), and ranks last on accuracy (Finding 7). Reproducing literal tables also risks US compilation copyright; re-expressed coefficients are safer, but the licence and contract barrier stands regardless.
ASHRAE Climatic Design Data License for Software Developers (RP-1923, 2025)	RESTRICTED	Non-exclusive perpetual embed licence; attribution "Climatic Design Data © 2025 ASHRAE… Used with permission."	$2 000 one-time	Not the cooling-load tables — Chapter 14 design conditions; methodology change still required	No — non-exclusive to the purchaser	Lets this project embed the data in its software; cannot license it onward under AGPL. Viable only as a closed optional component or a runtime layer the end user provisions. Covers 12 424 stations.
ASHRAE Weather Data Viewer 2025	BLOCKED	Annual single-user subscription, terms-of-use gated	Annual per-seat subscription	Design conditions only	No	Subscription access, not a redistribution licence.
NASA POWER	CLEARED	US Government open-data policy — no copyright; commercial use and redistribution permitted; attribution requested as a courtesy	Free	Methodology change — raw meteorological and solar data, not cooling-load values	Yes	Solar irradiance components, dry-bulb, dew point, humidity, wind. ½° lat × ⅝° lon (MERRA-2) for meteorology, 1° × 1° (CERES) for solar. The best vendorable seed, and the adopted primary source (AR-60). A continuous grid replaces the workbook's latitude-band lookup with interpolation between grid points.
ERA5 via Copernicus CDS	CLEARED (attribution)	Copernicus licence — reuse, redistribution and commercial use permitted with attribution to C3S/ECMWF	Free	Methodology change — reanalysis climate data	Yes, with attribution retained	Reserve source: higher-resolution alternative or complement to POWER; larger data-handling burden (CDS queueing).
PVGIS (JRC, European Commission)	CLEARED (attribution), verify	EC reuse policy; built on SARAH / ERA5 (ERA5 = attribution)	Free	Methodology change — solar irradiance and climate	Yes, with attribution	Reserve source. Confirm the exact JRC reuse notice for the specific product used before relying on it.
climate.onebuilding.org (TMYx / EPW / DDY)	BLOCKED (as-is)	"Copyright © 2014-2026 Climate.OneBuilding. All Rights Reserved." — no redistribution grant	Free download	Would be near drop-in for design conditions (DDY)	No — no licence granted	Free to download is not licensed to redistribute; only a disclaimer and a citation request are published. The underlying NOAA ISD (US public domain) and ERA5 are redistributable at source — go to the primaries instead. Non-TMYx files carry separate national-service terms.
EnergyPlus EPW weather files (DOE / NREL)	MIXED	EnergyPlus Open Source License v1.0 covers software and docs; each weather file carries its own source terms	Free	Design conditions / hourly	Per-file	Many files are redistributable (US public-domain sources); others (e.g. ISHRAE © 2005, national met services) are "for use at no cost" only. Must be cleared file by file.
Meteonorm	BLOCKED	Proprietary EULA; redistribution of generated data to third parties, including via software or web apps, prohibited without a separate written agreement	~CHF 675 first licence (excl. VAT); ~CHF 350 each additional	Design conditions / hourly	No — explicitly forbidden	EULA §II.c.8.2 bars "indirect provision of access… by making available to third parties results based on the data (e.g. through web applications, simulation software)", with a CHF 3 000-per-instance penalty at §VI. Only single-project client documentation is permitted (§II.c.8.3). Fundamentally incompatible with a public repository.
ISO 15927 (parts) / EN ISO 52016-1 (method)	METHOD USABLE; text BLOCKED	ISO standards are copyrighted and sold as DRM PDFs	Standard purchase, per document	EN ISO 52016-1 is a calculation method that replaces CLTD/CLF/SCL tables outright	Yes for an independently implemented method; No for the standard's text and tables	EN ISO 52016-1 hourly is the adopted B6 kernel (AR-64). Methods are not copyrightable: the algorithm may be implemented, the text and tables may not be redistributed. Needs open weather data as input either way. ISO 15927-4 defines the TRY construction method used by TMYx.
Calculation coefficients computed from first principles	CLEARED	No third-party licence — physics plus this project's own AGPL code	Free (development effort)	Methodology change — EN ISO 52016-1 hourly (adopted); RTS / Heat Balance (rejected)	Yes	Removes the table-licensing question entirely. The adopted method is EN ISO 52016-1 hourly (AR-64). RTS / Heat Balance is equally clear on licence and more accurate than CLTD (Finding 7), but was rejected because it yields a design day, not annual energy (§3a). This row originally recommended RTS / Heat Balance; corrected 2026-09-27 to match AR-64.

§5 Legal-review items (pre-public-release blockers)¶

These gate Phase 3 (repository public, docs site published), not Phase 1 calculation. Tracked as AR-61 on the Action register.

EU database-right determination on any ASHRAE-derived values. Confirm with EU IP counsel that re-expressed CLTD/CLF/SCL coefficients — created outputs — fall outside Directive 96/9/EC per BHB / Football Dataco, and that Chapter 14 design conditions (obtained and verified data) are treated separately. This only becomes live if the project ever revisits the ASHRAE path; until confirmed, treat all ASHRAE-derived values as not cleared for vendoring.
Contract versus copyright. Even where Feist denies copyright in the numbers, the terms accepted when obtaining ASHRAE data may bind by contract (ProCD v. Zeidenberg). Counsel should confirm that no click-through or licence term is breached by residual ASHRAE-informed methodology knowledge.
NASA POWER / ERA5 / PVGIS attribution wording must be embedded in the docs site, API responses, and exported reports, mirroring the ÖKOBAUDAT attribution discipline already adopted for EPDs (AR-28). One discipline, two data families.
climate.onebuilding.org must not be vendored absent written permission from its maintainers; prefer the primary sources (NOAA ISD, ERA5).
EnergyPlus EPW per-file clearance if any specific EPW or DDY file is ever vendored.

§6 Practical implications¶

6.1 The only path that keeps the public repo buildable and lawful¶

Derive design inputs from open data — NASA POWER as primary, ERA5 or PVGIS where higher resolution or an EU-specific product helps — and compute annual energy from an independently implemented EN ISO 52016-1 hourly kernel in project code (AR-64). That one choice answers the licence question, the quantity question (§3a) and the accuracy question (Finding 7).

6.2 This is a from-scratch build, not a migration¶

No cooling-load code exists to migrate. domains/energy, domains/thermal, lca.b4 and lca.b6 are five-line scaffolds with no logic, and services/interpolation.py is a NotImplementedError stub with zero call sites. Nothing is being thrown away; the decision changes what gets built, before anything is built. See AR-60.

6.3 Continuous gridded data dissolves the interpolation problem it was going to solve¶

The services/interpolation.py policy in METHODOLOGY.md §5a was written against table bands — "treating a 14.7°N project as if it were at 16°N". The bands in question are the reference workbook's own: sheet LM Lat- month corr in 1c_energetic_analysis.xlsx tabulates a latitude–month correction to CLTD at 0°, 8°, 16°, 24° and on to 64°N in 8° steps, across seven month-pairs, plus 12°N and 20°N blocks interpolated by formula. The defect is not the bands; it is that the workbook's cooling chain reads a pasted copy of the 12°N block for a site at 14.7°N instead of interpolating. Measured cell by cell against true interpolation, that is a typical 0.75 K and a maximum 1.687 K error on an input that drives q = U·A·CLTD (AR-62; METHODOLOGY.md §5a rule 4).

What is and is not verified about the ASHRAE source of that table. Verified: the printed CLTD tables are for July at 40°N (Handbook–Fundamentals) or July at 24°, 36° and 48°N (the 1992 Load Calculation Manual); GRP-158 supplies a latitude–month correction to extend them to other latitudes and months; ASHRAE permits interpolation, but the method's authors bound the printed tables to 20–50°N, and the 1997 Handbook warns of lost accuracy below 24°N. At 14.7°N the site lies outside ASHRAE's own stated envelope, so even correct interpolation would be extrapolating the method. Not verified: the exact latitude bands of the published LM correction table. The workbook's table (0–64°N in 8° steps, seven month-pairs) is consistent with that mechanism but has not been checked against a primary ASHRAE copy. ASHRAE's own caveat on the LM correction is on record: it was "reasonably consistent during summer months but became much less realistic for early and late hours during traditionally noncooling load periods" (1997 Handbook–Fundamentals, ch. 28, p. 28.4).

NASA POWER is a continuous global grid, so the question changes shape entirely: from which tabulated latitude band, to bilinear interpolation between four surrounding grid points. The five rules of §5a survive unchanged in substance and now govern spatial grid-point interpolation; §5a has been revised to say so.

The hemisphere-transform concern raised in the interpolation-status audit — that the LM table is tabulated for northern latitudes and would need a six-month month-shift for southern-hemisphere projects — is moot. NASA POWER has no hemisphere-specific data gap; it returns the same parameters for a site at 14.7°N and one at 14.7°S.

6.4 The $2 000 licence remains a fallback, outside the AGPL distribution¶

The ASHRAE Software Developer License is a legitimate option only if the ASHRAE-derived data stays out of the AGPL distribution — an optional separately-licensed closed data module, or a runtime layer the end user provisions themselves. It cannot be committed to the public repository.

6.5 Internal Handbook use is unaffected¶

Chiwara's internal use of the Handbook, under membership or purchase, remains lawful for internal engineering calculations. It simply does not extend to public AGPL redistribution.

6.6 Representativeness caveat carries over from the EPD clearance¶

As with ÖKOBAUDAT factors (EPD clearance §4.4), open global datasets have resolution and local-microclimate limits. NASA POWER's ½° × ⅝° meteorological grid is roughly 55 km × 60 km at Senegalese latitudes — a real limit for a site with local topographic or coastal effects. This is a pedigree/DQI matter, documented per input under the Weidema axes and propagated by the Monte Carlo engine, not a reason to reject the data.

§7 Recommended path¶

Immediate (Session 17). Adopt the determination: do not vendor ASHRAE CLTD/CLF/SCL tables or Chapter 14 climatic design data in the public AGPL repository. Record the §4 verdict table in THIRD_PARTY.md. Resolve AR-31 on this basis. Open AR-60 (the methodology pivot) and AR-61 (the §5 legal-review items). Revise METHODOLOGY.md §5a to the continuous-grid framing.
Before the energy module is built (Phase 2). Build the climate-data service against NASA POWER, and the B6 kernel as an independently implemented EN ISO 52016-1 hourly calculation (adopted as AR-64, superseding this report's original RTS / Heat-Balance recommendation), with coefficients computed in project code from layer properties. Add ERA5 or PVGIS as reserve sources where resolution warrants, retaining their attributions. Note AR-62: the reference workbook's latitude selection is a CORRECTED formula under decision #23, not a parity target.
Before public release (Phase 3). Obtain EU-counsel sign-off on §5 items 1–2. Finalise the attribution wording of §5 item 3 across all three surfaces. Confirm that no EPW, climate.onebuilding.org, or Meteonorm-derived value has entered the repository.

What would change these recommendations¶

If ASHRAE offers an explicitly sublicensable and redistributable grant — of the sort it made for Guideline 36-2018's control sequences — for the cooling-load or climatic data, vendoring under that grant becomes re-evaluable.
If the EN ISO 52016-1 build proves infeasible on schedule, keep the ASHRAE Software Developer License data as a closed, optional module outside the AGPL repository as an interim, and continue the migration.
If EU counsel confirms that re-expressed coefficients fall outside both copyright and the database right and that no contract term binds, a narrowly-scoped vendoring of re-derived coefficients could be reconsidered — but the method change would still stand, because CLTD computes a design-day peak rather than the B6 quantity (§3a) and ranks last on accuracy (Finding 7), so this would not reverse the decision.

§8 Caveats¶

Not legal advice. The §5 items require a qualified lawyer: EU IP counsel given the project's EU footprint, plus US-law input given ASHRAE's US origin.
US and EU diverge, and the divergence is real. A "facts are not copyrightable" answer under Feist does not transfer to the EU, where the sui generis database right and contract terms can restrict reuse independently of copyright. The BHB / Football Dataco created-data limitation helps but does not guarantee.
Prices and terms change. The $2 000 Software Developer License fee, the Weather Data Viewer subscription price, and the Meteonorm list price were current at the time of investigation and should be re-checked at purchase time. The Meteonorm EULA notes that only the German-language version is legally binding.
Verdicts are as-published, not as-negotiated. Every BLOCKED and RESTRICTED verdict above reflects published terms. A written grant obtained directly from a programme owner supersedes the published position — as it would for any EPD source.

§9 Revision record — Chiwara's challenge (2026-09-27)¶

Chiwara raised four objections to the account of the ASHRAE CLTD/CLF/SCL method that this project had given — partly on this site and partly in correspondence with Chiwara. Checking them changed the account. The corrected version below was not the project's original position. It is recorded here so a reviewer can see what was claimed, what was wrong, and what survived.

#	Earlier claim	Where it was made	Verdict	Corrected account
1	The tables cover only seven wall groups and cannot represent an earth wall.	Correspondence with Chiwara (not on any site page)	Withdrawn — wrong.	ASHRAE research project RP-472 documents a procedure for an arbitrary wall. Describe each layer by thickness, conductivity, density, specific heat and resistance; identify the heaviest layer; read a wall group from 41 representative assemblies; take the conduction transfer function coefficients from there. A 40 cm earth wall at 1 800 kg/m³ is about 720 kg/m², inside the range those assemblies span. The reference workbook itself assigns wall groups by mass (`Données!E4:E11`). The real mismatch is that ASHRAE's heavy walls are insulated and an earth wall is not. That is a caveat on representativeness, not an obstacle.
2	Latitude: the site lies below the lowest table; tables exist at 0°/8°/16°/24°N.	The bands were attributed to ASHRAE's published tables on this site (for stakeholders §(c)); the "below the lowest table" framing does not appear on any site page	Corrected.	The Fundamentals Handbook publishes CLTDs at 40°N and the 1992 Load Calculation Manual at 24°, 36° and 48°N, and ASHRAE explicitly permits interpolation. But the method's authors bound the printed tables to 20–50°N, and the 1997 Handbook warns of lost accuracy below 24°N, so at 14.7°N the site is outside ASHRAE's own stated envelope. The 0–64°N bands in 8° steps are real, but they are in the reference workbook's `LM Lat- month corr` sheet; that they match ASHRAE's published LM table is not verified (§6.3). The 0.75 K typical / 1.687 K maximum error stands. It is measured from the workbook's own table, and the defect is that the workbook pasted the 12°N block instead of interpolating (AR-62).
3	The tables apply only to May–September.	Correspondence with Chiwara (not on any site page)	Corrected.	"May–September" describes an older factor for solar gain through glass, not the wall and roof CLTD tables. The printed CLTD tables are for July; GRP-158 supplies a latitude–month correction to extend them, which ASHRAE itself described as less realistic outside the summer cooling months (§6.3). The 1992 Manual also shipped CLTDTAB, a program for other latitudes and months, which is no longer available. At 14.7°N, July is not the hottest month: the workbook's own daily cooling totals peak in April (93.0 kWh/day) and May (85.5), against 50.8 in July (`Données!J6`, `J7`, `J9`).
4	ASHRAE deprecated the method.	This report (TL;DR and Finding 5)	Corrected.	RTS superseded CLTD/CLF/SCL in 2001. The method was never formally withdrawn and is still in wide practical use. Do not write "retired" or "deprecated".

What stands, and in what order.

The licence is the blocker (Findings 1–2, §4). No ASHRAE licence exists for these tables. Any permission would carry restrictions that AGPL-3.0 §10 forbids passing on, government reproductions do not launder the copyright, and there is no open equivalent.
The tables answer a different question (§3a). CLTD sizes equipment for a design day; B6 is annual energy in kWh. This is now the lead technical reason.
Accuracy (Finding 7) is new with this revision and stands on published comparisons: CLTD ranks last of five methods and over-predicts in hot climates.

The decision does not change. No ASHRAE table enters the repository, and B6 is computed by the EN ISO 52016-1 hourly kernel driven by NASA POWER (AR-60, AR-64). What changed is the reasoning given for it: two technical objections were wrong or overstated, and the lead technical reason is now the functional unit.