i The Direct Answer
EPA’s Risk Management Program — Clean Air Act Section 112(r), implemented at 40 CFR Part 68 — does not contain a standalone rule titled ‘weather monitoring.’ Any source telling you it does is wrong. What Part 68 contains instead are three distinct provisions that each require site meteorological data, and together they create a real and ongoing documentation obligation.
First, 40 CFR § 68.22 sets the meteorological parameters for offsite consequence analysis. Worst-case modeling must assume a wind speed of 1.5 meters per second and F atmospheric stability class, the highest daily maximum temperature in the previous three years, and average site humidity — with those temperature and humidity figures drawn from data gathered at the stationary source or a local meteorological station.
Second, § 68.22(b) contains an option most facilities never exercise: if the owner or operator can demonstrate that local meteorological data show a higher minimum wind speed or a less stable atmosphere at all times during the previous three years, those site-specific minimums may be substituted for the conservative defaults. Higher assumed wind speed disperses a release faster, which shortens the modeled distance to endpoint.
Third, 40 CFR § 68.42(b) requires that the five-year accident history report weather conditions at the time of each reportable release — a record that must be accurate years after the event and is submitted to EPA through RMP*eSubmit.
A fourth requirement — evaluation of natural hazards in the process hazard analysis, added by the 2024 Safer Communities by Chemical Accident Prevention rule — is currently subject to a proposed EPA rollback and should be treated as unsettled. See the regulatory status section below.
This article is written for EHS managers, process safety engineers, RMP coordinators, and compliance staff at facilities covered under 40 CFR Part 68 — refineries, chemical plants, ammonia refrigeration operations, water and wastewater treatment, agricultural retailers, and oil and gas processing. It covers each weather-related provision with the regulatory text, the site-data substitution option in § 68.22(b), the current status of the natural hazards requirement, and what a defensible meteorological documentation program looks like.
⚠ Regulatory Status Warning: Part 68 Is Actively in Flux
RMP has changed direction three times in three years, and any guidance written before 2026 may describe requirements that are being withdrawn.
March 11, 2024 — EPA published the Safer Communities by Chemical Accident Prevention (SCCAP) final rule, effective May 10, 2024. Among other changes it emphasized the requirement for regulated facilities to evaluate natural hazard risks including associated loss of power, and added a backup power provision for monitoring equipment. Most compliance dates fall on May 10, 2027, with RMP resubmittal a year later.
March 12, 2025 — EPA announced it was reconsidering the 2024 SCCAP rule.
February 2026 — EPA proposed the Common Sense Approach to Chemical Accident Prevention rule, which would rescind or weaken a substantial portion of SCCAP. Provisions explicitly proposed for rescission include the backup power requirement for monitoring equipment, and the requirements for documenting and reporting declined recommendations relating to natural hazards, power loss, and siting.
Before relying on any natural-hazard or backup-power provision described in this or any other article, confirm the current status of the Common Sense rulemaking at epa.gov/rmp. The § 68.22 and § 68.42 provisions discussed below are long-standing, date to the original 1996 rule, and are not part of either rulemaking.
Requirement 1: Offsite Consequence Analysis Parameters — 40 CFR § 68.22
Every facility with a covered process must perform an offsite consequence analysis: at minimum one worst-case release scenario, plus alternative release scenarios for Program 2 and Program 3 processes. The modeling is not free-form. Section 68.22 prescribes the meteorological assumptions, and those assumptions directly determine the calculated distance to endpoint — which in turn determines how many public receptors fall inside your vulnerability zone.
Wind speed and atmospheric stability — § 68.22(b)
◆ Verbatim — 40 CFR § 68.22(b)
“For the worst-case release analysis, the owner or operator shall use a wind speed of 1.5 meters per second and F atmospheric stability class. If the owner or operator can demonstrate that local meteorological data applicable to the stationary source show a higher minimum wind speed or less stable atmosphere at all times during the previous three years, these minimums may be used. For analysis of alternative scenarios, the owner or operator may use the typical meteorological conditions for the stationary source.”
The defaults are deliberately conservative. A wind speed of 1.5 m/s — roughly 3.4 mph — combined with F stability class describes a nearly calm, strongly stratified nighttime atmosphere in which a dense gas cloud travels far without dispersing. It is close to the worst dispersion condition that can physically occur.
For alternative scenarios, EPA’s RMP Offsite Consequence Analysis Guidance assumes 3 meters per second and D stability class, representing typical daytime conditions.
Ambient temperature and humidity — § 68.22(c)
◆ Verbatim — 40 CFR § 68.22(c)
“For worst-case release analysis of a regulated toxic substance, the owner or operator shall use the highest daily maximum temperature in the previous three years and average humidity for the site, based on temperature/humidity data gathered at the stationary source or at a local meteorological station; an owner or operator using the RMP Offsite Consequence Analysis Guidance may use 25 °C and 50 percent humidity as values for these variables. For analysis of alternative scenarios, the owner or operator may use typical temperature/humidity data gathered at the stationary source or at a local meteorological station.”
Note the phrase that appears in both subsections: data gathered at the stationary source or at a local meteorological station. The regulation contemplates on-site measurement as the first-named option. It also imposes a three-year lookback — meaning the underlying dataset has to exist, and has to be retrievable when the RMP is prepared or revised.
Temperature of the released substance — § 68.22(g)
A third temperature dependency is easy to miss. Under § 68.22(g), for worst-case analysis, liquids other than gases liquefied by refrigeration must be considered released at the highest daily maximum temperature based on data for the previous three years appropriate for the stationary source, or at process temperature, whichever is higher. Release temperature affects vaporization rate, which affects the source term feeding the dispersion model.
Surface roughness — § 68.22(e)
Section 68.22(e) addresses terrain. Urban or forested settings with buildings and obstructions generate mechanical turbulence that mixes a plume and shortens its reach; flat open rural terrain lets a dense cloud travel comparatively undisturbed. EPA’s guidance permits selecting urban or rural surface roughness accordingly, and the choice materially changes the modeled distance. Like the meteorological parameters, the selection and its justification are records under § 68.39 — a facility classifying itself as urban should be able to explain why.
The definition that ties it together
Part 68’s definitions section defines typical meteorological conditions as the temperature, wind speed, cloud cover, and atmospheric stability class prevailing at the site, based on data gathered at or near the site or from a local meteorological station. Four parameters, site-based, used throughout the alternative scenario analysis.
The § 68.22(b) Substitution Most Facilities Never Use
Read the second sentence of § 68.22(b) again. It is permissive, and it is consequential:
→ The Provision
“If the owner or operator can demonstrate that local meteorological data applicable to the stationary source show a higher minimum wind speed or less stable atmosphere at all times during the previous three years, these minimums may be used.”
In plain terms: if three years of site data prove that the wind at your facility never drops to 1.5 m/s, or that F-stability conditions never occur there, you may model your worst case using your actual site minimum instead of EPA’s default.
Why this matters operationally: wind speed is in the denominator of dispersion. Doubling the assumed wind speed roughly halves the modeled centerline concentration at a given distance, which pulls the distance-to-endpoint inward. A shorter endpoint distance means a smaller vulnerability zone, which can mean fewer public receptors — schools, hospitals, residences — inside the circle reported in your RMP and displayed in EPA’s public data tool.
That figure is not merely administrative. It shapes community perception, local emergency planning committee expectations, insurance and lender scrutiny, and in some cases whether a process qualifies for a lower program level.
! What the Demonstration Actually Requires
The bar is deliberately high, and the phrase ‘at all times during the previous three years’ is doing the work. A facility cannot use an average, a typical value, or a seasonal figure. It must show that the condition never occurred across the full three-year record.
That requires a continuous, gap-free dataset from a calibrated instrument at the site. A handful of spot readings will not support the demonstration, and neither will data from an airport twenty miles away — the regulation asks for data applicable to the stationary source.
Practically, this means the decision to pursue the substitution has to be made three years before the benefit is realized. A facility that starts logging site meteorological data today has the record available for its next RMP resubmittal cycle. One that does not will be using EPA’s defaults indefinitely.
Consult your process safety engineer and, where warranted, counsel before relying on a substitution. The demonstration must be defensible to the implementing agency.
Requirement 2: The Documentation Rule Itself — 40 CFR § 68.39
Section 68.22 tells you which meteorological parameters to use. Section 68.39 is the provision that makes your choice of those parameters a record you must keep. It is titled, simply, Documentation.
◆ Verbatim — 40 CFR § 68.39(a)
“The owner or operator shall maintain the following records on the offsite consequence analyses: (a) For worst-case scenarios, a description of the vessel or pipeline and substance selected as worst case, assumptions and parameters used, and the rationale for selection; assumptions shall include use of any administrative controls and any passive mitigation that were assumed to limit the quantity that could be released…”
Three words carry the weight: assumptions and parameters used, and the rationale for selection. The meteorological values fed into a dispersion model are parameters. Whether you took EPA’s defaults or substituted site data is an assumption. Why you chose one over the other is the rationale. All three must be on file.
This is why a facility cannot simply state a wind speed in a modeling output and consider the matter closed. If an inspector asks where the highest daily maximum temperature figure came from, or which station supplied the humidity average, or why the default 1.5 m/s was used rather than site data, § 68.39 is the provision that says an answer should already exist in writing.
The review cycle — § 68.36
◆ Verbatim — 40 CFR § 68.36
“(a) The owner or operator shall review and update the offsite consequence analyses at least once every five years.”
“(b) If changes in processes, quantities stored or handled, or any other aspect of the stationary source might reasonably be expected to increase or decrease the distance to the endpoint by a factor of two or more, the owner or operator shall complete a revised analysis within six months of the change and submit a revised risk management plan as provided in § 68.190.”
Two points matter for meteorological documentation. First, the five-year review means the underlying weather data has to be re-examined on a cycle, not established once and forgotten — and the § 68.22 three-year lookback for highest daily maximum temperature moves with it. Second, the factor-of-two trigger is set high: EPA’s own guidance notes that for most substances, the released quantity would have to change by more than a factor of five to double or halve the endpoint distance. Meteorological assumptions are among the few inputs that can move the result materially without any change to the process itself.
Demonstrating That Your Weather Data Is Representative
Whether a facility uses its own instrument or an off-site station, § 68.39 requires the rationale for that selection. EPA has not set a numerical distance limit for an acceptable off-site station, which means the burden falls on documented technical judgment. These are the factors a defensible representativeness narrative addresses:
- Distance from the facility. No regulatory cap exists, but closer is more defensible, and a station tens of miles away requires a correspondingly stronger argument.
- Terrain similarity. Flat versus complex, coastal versus inland, urban versus rural. A coastal station standing in for an inland site — or vice versa — is difficult to defend on dispersion grounds.
- Anemometer exposure and height. Wind measurement convention is 10 meters above ground with unobstructed fetch. EPA’s meteorological monitoring guidance has long advised that obstructions be no closer than ten times their height from the sensor in simple terrain. A sensor in a building wake understates wind and would actively undermine a § 68.22(b) substitution.
- Climatic similarity. Seasonal temperature, humidity, and wind patterns at the data source should reasonably approximate the facility’s.
- Period of record and completeness. Part 68 does not prescribe a data capture rate, but EPA air quality modeling practice generally treats 90% or better completeness as the threshold for a credible dataset. Document what percentage of expected observations were captured and how any gaps were handled.
i What a Complete Meteorological Package Contains
A useful organizing convention is a dedicated Meteorological Basis section in the RMP file, holding: station identification and full metadata (coordinates, elevation, sensor types and heights, exposure description); the period of record and a completeness statement; summary statistics including a wind rose and a stability class frequency distribution; the representativeness narrative; the dispersion modeling input files with date stamps, whether RMP*Comp, ALOHA, or another model was used; and the § 68.39 rationale for each parameter selection.
Where both default and site-specific runs were performed, archive both. The comparison is itself evidence of diligence, and it answers the obvious inspector question before it is asked.
Requirement 3: Weather Conditions in the Five-Year Accident History — 40 CFR § 68.42
Section 68.42(a) requires the owner or operator to include in the five-year accident history all accidental releases from covered processes that resulted in deaths, injuries, or significant property damage on site, or known offsite deaths, injuries, evacuations, sheltering in place, property damage, or environmental damage.
Section 68.42(b) then enumerates the data required for each reportable release. Among the required fields — alongside date, time, duration, chemical, estimated quantity, NAICS code, release event and source, on-site and known offsite impacts, initiating event and contributing factors — are the weather conditions at the time of the release.
⚠ This Is a Retrospective Requirement With No Do-Over
Weather conditions at the moment of a release cannot be reconstructed after the fact with any precision. Either the data was captured when it happened, or it was not.
Standard practice among RMP practitioners is to pull the figure from on-site weather instrumentation where it exists, and fall back to the nearest National Weather Service station for the time of the release where it does not. The fallback is workable but weaker: it reports conditions at the station’s location, which in a release investigation is precisely the thing at issue, because wind direction and speed at the source determine where the plume went.
The five-year accident history is among the most scrutinized sections of an RMP. It is public, it feeds EPA’s enforcement targeting, and under the 2024 rule an accident meeting § 68.42(a) criteria can trigger a third-party compliance audit. Reporting is due within six months of the release or at the next RMP update, whichever is earlier.
There is a second, less obvious consequence. Wind speed and direction at the time of release are the primary inputs to any after-the-fact plume reconstruction — the analysis that determines who was actually exposed, which drives the offsite impact figures also required by § 68.42(b), and which will be central to any litigation. A facility with a continuous on-site record can answer that question. One relying on a distant station is estimating.
Requirement 4: Natural Hazards in the Process Hazard Analysis — Status Unsettled
The 2024 SCCAP rule emphasized the requirement for regulated facilities to evaluate the risks of natural hazards, including any associated loss of power, within the process hazard analysis for Program 3 processes under § 68.67, with parallel hazard review obligations for Program 2. EPA paired this with a provision requiring backup power for monitoring equipment and with documentation requirements covering recommendations a facility declined to implement.
The reference case for this requirement is the Arkema facility in Crosby, Texas, where Hurricane Harvey flooding in August 2017 disabled primary and backup refrigeration for temperature-sensitive organic peroxides, leading to fires and a release. It is the standard illustration of how a natural hazard, a loss of power, and a temperature-dependent process combine into a chemical accident — and it is why the natural hazards and power loss provisions were written together.
The rationale was straightforward: a significant share of RMP-reportable accidents have a weather or natural-hazard component — hurricane flooding, freeze events disabling instrumentation, lightning strikes on tankage, high wind damaging process structures — and the pre-2024 rule did not require these to be systematically evaluated.
! Where This Stands as of September 2026
EPA announced reconsideration of the 2024 SCCAP rule on March 12, 2025, and in February 2026 proposed the Common Sense Approach to Chemical Accident Prevention rule.
Under that proposal, the SCCAP backup power requirement for monitoring equipment would be rescinded, along with the corresponding compliance date at § 68.10(g)(1). The requirements for documenting and reporting declined recommendations for natural hazards, power loss, and siting would also be rescinded.
The proposal had not been finalized at the time of writing. Facilities should verify current status before making compliance decisions, and should be aware that a rescission of documentation requirements does not necessarily eliminate the underlying obligation to identify and evaluate hazards in a PHA — the general PHA requirements at § 68.67 predate SCCAP.
The practical advice most process safety practitioners are giving: do not dismantle a natural hazard evaluation you have already built. The analysis has independent value for insurance, for business continuity, and for the general duty obligation under CAA Section 112(r)(1), regardless of what happens to the specific documentation mandate.
RMP Weather Data Obligations at a Glance
| Provision | What It Governs | Requirement | Status |
| § 68.22(b) | Wind speed and atmospheric stability for OCA modeling | 1.5 m/s and F stability for worst case; typical conditions for alternative scenarios. Site data may substitute if it shows higher minimum wind speed or less stable atmosphere at all times over three years | Long-standing — 1996 rule, not affected by current rulemaking |
| § 68.22(c) | Ambient temperature and humidity for OCA modeling | Highest daily maximum temperature in previous three years plus average site humidity, from site or local station data. EPA guidance default is 25 °C and 50% humidity | Long-standing — 1996 rule |
| § 68.22(g) | Release temperature for liquid source term | Highest daily maximum temperature over previous three years, or process temperature, whichever is higher | Long-standing — 1996 rule |
| § 68.42(b) | Weather conditions at time of reportable release | Must be reported for each release in the five-year accident history; submitted via RMP*eSubmit within six months or at next RMP update | Long-standing — 1996 rule |
| § 68.67 / § 68.50 | Natural hazards evaluation in PHA or hazard review | Emphasized by 2024 SCCAP rule, including associated power loss | Unsettled — documentation provisions proposed for rescission in February 2026 |
| § 68.10(g)(1) | Backup power for monitoring equipment | Added by 2024 SCCAP | Proposed for rescission in February 2026 |
Building a Defensible Meteorological Documentation Program
Step 1 — Determine which data you actually need
At minimum: temperature, relative humidity, wind speed, and wind direction, logged continuously. Barometric pressure supports dispersion modeling. Rainfall matters for facilities where flooding is a credible natural hazard scenario. Lightning proximity matters where tankage, flares, or loading operations are exposed.
Step 2 — Site the instrument defensibly
The regulation asks for data applicable to the stationary source. An instrument placed in a sheltered courtyard understates wind and would undermine rather than support a § 68.22(b) substitution. Standard practice follows meteorological siting conventions: unobstructed exposure, standard measurement height, away from building wake effects and heat sources. Document the siting rationale — it is the first thing a reviewer will question.
Step 3 — Log continuously and retain
The three-year lookback in § 68.22 and the five-year window in § 68.42 mean the retention horizon is at least five years, and practically longer given the RMP resubmittal cycle. Automated logging removes the failure mode where a record exists in principle but was never captured on the night it mattered.
Step 4 — Integrate with the incident response procedure
Write into the emergency response and incident investigation procedures that meteorological conditions at the time of the event are captured as part of the initial response. If the station logs automatically, this is a retrieval step rather than an observation step — which is the difference between having the data and hoping someone wrote it down.
Step 5 — Feed the natural hazard evaluation
Whatever the final disposition of the SCCAP documentation provisions, a PHA team evaluating flooding, freeze, high wind, or lightning scenarios is far better served by several years of site-specific data than by regional climate normals. Site history is also what distinguishes a credible scenario from a speculative one.
→ How cyclonePORT Supports RMP Meteorological Documentation
cyclonePORT measures temperature, relative humidity, wind speed and direction, barometric pressure, rainfall, and lightning proximity from a single industrial-grade station at the facility itself, logging continuously with automatic timestamped retention.
For § 68.22: the continuous multi-year record is the dataset a § 68.22(b) substitution demonstration requires, and the source of the highest-daily-maximum-temperature and average-humidity figures § 68.22(c) calls for.
For § 68.42: conditions at the time of any reportable release are already captured and retrievable, rather than reconstructed from a distant NWS station months later when the accident history is compiled.
For natural hazard evaluation and emergency response: configurable threshold alerts deliver simultaneously to the control room, the EHS manager, and shift supervision — covering high wind limits on crane and loading operations, freeze thresholds for instrumentation and piping, and lightning proximity for tankage and transfer operations.
RMP Weather Documentation Compliance Checklist
Items marked (F) relate to provisions currently in regulatory flux — verify status before relying on them.
| Program Element | What to Confirm | ✓ | ✗ |
| OCA meteorological basis documented | The RMP file records which § 68.22 parameters were used, whether defaults or site data, and the source of each figure | ☐ | ☐ |
| Three-year temperature record available | Highest daily maximum temperature for the previous three years is retrievable and sourced from site or an identified local station, per § 68.22(c) | ☐ | ☐ |
| Average site humidity documented | Average humidity figure used in worst-case modeling is documented with its source | ☐ | ☐ |
| Release temperature basis recorded | For liquid scenarios, the § 68.22(g) determination — highest daily maximum versus process temperature — is documented | ☐ | ☐ |
| Alternative scenario conditions justified | Typical meteorological conditions used for alternative scenarios are based on site or local station data, not assumed | ☐ | ☐ |
| § 68.22(b) substitution evaluated | The facility has assessed whether three years of site data would support substituting a higher minimum wind speed or less stable atmosphere — and documented the decision either way | ☐ | ☐ |
| Continuous logging in place | Temperature, humidity, wind speed, and wind direction are logged continuously at the site with automatic retention, not spot-checked | ☐ | ☐ |
| Instrument siting documented | Siting rationale recorded, following standard meteorological exposure conventions, with justification that data is applicable to the stationary source | ☐ | ☐ |
| Retention horizon meets five years | Meteorological records are retained at least five years to cover both the § 68.22 three-year lookback and the § 68.42 five-year accident history window | ☐ | ☐ |
| Incident procedure captures weather | Emergency response and incident investigation procedures specify capture of meteorological conditions at time of event, per § 68.42(b) | ☐ | ☐ |
| Accident history weather fields complete | Every release in the current five-year accident history has weather conditions populated in RMP*eSubmit | ☐ | ☐ |
| Six-month reporting tracked | Process exists to submit accident history data within six months of a reportable release or at the next RMP update, whichever is earlier | ☐ | ☐ |
| Natural hazard evaluation in PHA | PHA or hazard review addresses credible weather-driven scenarios — flooding, freeze, high wind, lightning — informed by site data (F) | ☐ | ☐ |
| Power loss scenarios considered | PHA addresses loss of power associated with natural hazards and its effect on monitoring and safeguards (F) | ☐ | ☐ |
| §68.39 rationale documented | For each meteorological parameter, the assumption, the value used, the data source, and the rationale for selection are recorded in writing per § 68.39 | ☐ | ☐ |
| Surface roughness selection justified | Urban versus rural classification under § 68.22(e) is documented with supporting rationale | ☐ | ☐ |
| Representativeness narrative on file | If off-site data is used, a written assessment covers distance, terrain similarity, anemometer exposure and height, climatic similarity, and data completeness | ☐ | ☐ |
| Five-year OCA review scheduled | § 68.36(a) review cycle is calendared, and the meteorological basis is re-examined as part of it | ☐ | ☐ |
| Factor-of-two trigger monitored | Process exists to identify changes that might move endpoint distance by a factor of two or more, requiring revised analysis within six months under § 68.36(b) | ☐ | ☐ |
| Modeling inputs archived | Dispersion model input files (RMP*Comp, ALOHA, or other) are retained with date stamps alongside their outputs | ☐ | ☐ |
| Current rulemaking status verified | Someone owns tracking the Common Sense Approach rulemaking and reporting status changes to the RMP coordinator (F) | ☐ | ☐ |
Frequently Asked Questions: EPA RMP Weather Documentation Requirements
Does EPA's Risk Management Program require weather monitoring?
What meteorological conditions does 40 CFR 68.22 require for worst-case modeling?
Can a facility use its own weather data instead of EPA's default wind speed?
Does the RMP five-year accident history require weather data?
Does RMP require facilities to evaluate natural hazards and climate risk?
What is the difference between worst-case and alternative release scenario weather assumptions?
How long must RMP-related meteorological records be retained?
What does 40 CFR 68.39 require facilities to document about weather assumptions?
How often must RMP offsite consequence analyses be reviewed and updated?
Would data from a nearby airport weather station satisfy RMP requirements?
Additional Resources
- 40 CFR Part 68 — Chemical Accident Prevention Provisions, full text at ecfr.gov
- 40 CFR § 68.22 — Offsite consequence analysis parameters
- 40 CFR § 68.36 — Review and update of offsite consequence analyses
- 40 CFR § 68.39 — Documentation requirements for offsite consequence analyses
- 40 CFR § 68.42 — Five-year accident history
- EPA — When to Revise OCA guidance — epa.gov/rmp/when-revise-oca
- EPA RMP Offsite Consequence Analysis Guidance — epa.gov/rmp
- EPA — Risk Management Program Safer Communities by Chemical Accident Prevention Final Rule (89 FR 17622, March 11, 2024)
- EPA — Common Sense Approach to Chemical Accident Prevention Proposed Rule (February 2026) — epa.gov/rmp
- EPA RMP*eSubmit — the submission system for risk management plans and accident history data
Related cyclonePORT resources:
- Industrial Weather Station — Product Overview — cycloneport.com
- Weather Monitoring for Oil, Gas & Mining Operations — cycloneport.com
- MSHA Lightning Requirements for Surface Mining and Blasting Operations — cycloneport.com
- OSHA Lightning Safety Requirements for Outdoor Workers: The Complete 2026 Guide — cycloneport.com
