Research outputs
Completed outputs are listed separately from planned ones. Registration, DOI and manuscript links appear only after they actually exist.
Study protocol
FrozenThe current internal protocol specifies the primary estimand, study window, completeness rules, QC, missingness, bootstrap procedure and sensitivity analyses before the primary period begins.
DAH-PM25-2026-27-v1-1-1 · frozen 8 Sept 2026
Open protocol →External registration
Registered under embargoThe prospective protocol has an independently timestamped OSF registration.
OSF Registries
Open OSF registration ↗Research data
2 frozen releasesNamed releases freeze the station roster, normalized snapshots, monitor-hour table, provenance and QC state so a later analysis can refer to a stable dataset boundary.
9,225 monitor-hours · 71 physical sites
Open data releases →Software
VersionedThe application, data pipeline and research code are versioned separately from frozen datasets. A public archival software DOI will be added when the repository release is connected to an archival service.
Software DOI: not yet assigned
Dataset DOI
PlannedA DOI-bearing dataset package is planned after the primary study window, with a data dictionary, station registry, QC ledger, provenance, weather covariates and reproducibility files.
DOI: not yet assigned
Manuscript
PlannedThe primary manuscript will report the prespecified network-relative site contrasts after the primary study period. No manuscript or preprint is presented as complete at this stage.
Primary window ends 31 Mar 2027
Reproducibility chain · v2.29.0
The research-object generator freezes the data, protocol, cohort and audit trail. The analysis engine consumes only that checksum-verified object and applies the frozen primary rules: physical-site aggregation, shared network-hour eligibility, network-relative residuals, site completeness, heterogeneity, block-bootstrap uncertainty, monthly persistence and every prespecified sensitivity analysis.
Before the primary period, the analyzer can run only with --pilot-dry-run, which marks outputs as code-validation artifacts. Primary mode requires the frozen baseline cohort embedded in the object; final reporting can additionally require the complete prespecified study window.
Immutable results · v2.29.0
Result pages read checksum-verifiable analysis files embedded in the deployment. They do not query the live database and do not silently recompute when new observations arrive.
Open frozen results →Evidence chain
Defines the analysis before the primary data period.
Collects hourly source data with provenance and explicit missingness.
Create stable data boundaries with checksums.
Applies the prespecified rules to eligible primary-period observations.
Cites the protocol, release and software used for the published result.
Current study identity
DAH-PM25-2026-27-v1-1-1
SHA-256 22076bd8e22dba94f8f6ca11bb66f0ac09fe4e9dc6801a1a6ebd76682835672a