Help & FAQ

The default and recommended equation set is SM2025 (Kjeldsen et al. 2025, aligned with UKFE R v2.0.0) — the current FEH statistical method for QMED, pooling, and growth curves.

The older SM2008 method is treated as legacy. It is no longer offered in the interface and is retained only so previously saved analyses that selected it still open. New assessments should use SM2025.

Yes. The SM2025 QMED chain is cross-validated against WINFAP 5 and agrees to within roughly 0–1.2%. The descriptor equation matches the UKFE R v2.0.0 reference coefficients, and the urban adjustment factor matches WINFAP to eight decimal places.

This is locked down by a suite of 97 QMED lock-down tests run against three WINFAP reference sites, so QMED cannot silently drift. See the FEH and UKFE formula audit for the full comparison matrix.

Donors are gauged catchments used to adjust the descriptor-based QMED towards local observed behaviour. They are selected by hydrological similarity using catchment centroids, with an urban-extent threshold applied to candidate donors.

You can change the donor set: exclude individual donors to recompute the adjustment factor, and rerun. The QMED methodology article documents the selection and weighting in detail.

A gauged assessment starts from an NRFA station that has an observed annual-maximum (AMAX) record. QMED is estimated directly from those observations, with descriptor regression as a cross-check, and single-site growth curves are available.

An ungauged assessment starts from a catchment descriptor export (FEH XML). There is no observed AMAX, so QMED comes from descriptor regression (refined by donor adjustment) and the growth curve comes from the pooling group only. Both routes converge at the pooling and growth-curve stage.

Yes. Individual descriptors (AREA, SAAR, FARL, BFIHOST, URBEXT, and others) can be edited in the Overrides panel, or replaced wholesale by importing a FEH XML export. You can also set a direct QMED override to test sensitivity.

Descriptors feed almost everything downstream — QMED regression, donor and pooling similarity, and the rainfall-runoff model parameters — so rerun the analysis after editing them.

Growth curves are fitted from pooled L-moment ratios. Hydrometric fits multiple distributions (GLO and GEV among them) and reports a goodness-of-fit Z statistic for each. The recommended distribution is the best-fitting one (lowest |Z|) for the pooling group — often GLO for UK catchments, which is the traditional FEH default, but not universally.

You can select a different distribution to compare its growth factors. See the growth-curve methodology.

These are pooling-group diagnostics from regional frequency analysis:

• Discordancy (D_i) — flags a site whose L-moment ratios are unusual relative to the rest of the group. A site above the critical value is worth reviewing and possibly excluding.
• Heterogeneity (H) — measures how consistent the group is. Broadly, H < 1 is acceptably homogeneous, 1–2 is possibly heterogeneous, and > 2 is heterogeneous.
• Z statistic — a goodness-of-fit score per candidate distribution; |Z| ≤ 1.64 is conventionally treated as an acceptable fit.

They surface as badges in the analysis tree so you can judge whether the pooling group is trustworthy before relying on the design flows.

Honestly: the rainfall-runoff engine currently ships with ReFH1 (Kjeldsen 2007) coefficients and has not yet been validated against ReFH2 reference output. It is offered as a research preview and an independent cross-check, not as a validated ReFH2 replacement.

The design rainfall it consumes also uses an approximate scaling where a full FEH DDF export is not supplied. Treat the statistical (FEH) design flows as the primary result and the rainfall-runoff peaks as a sense-check.

Design flows are produced for return periods from 2 up to 1000 years (2, 5, 10, 25, 50, 100, 1000 years, plus intermediate periods where configured).

Climate change allowances follow Environment Agency UKCP18 guidance for the 2020s, 2050s and 2080s epochs. Coverage is England only — the 11 EA river basin districts. Scotland (SEPA) and Wales (NRW) guidance is not yet implemented.

Production calculations run against the frozen NRFA Peak Flow Dataset version 14.0.1 (AMAX, peaks-over-threshold, and catchment descriptors), published by the UK Centre for Ecology & Hydrology.

The live NRFA website / API is used only for map browsing, never for the numbers in an assessment. The dataset version is recorded against every analysis. Full sourcing is on the Data & attributions page.

Yes. Every calculated result carries a version stamp recording the equation version (e.g. SM2025), the NRFA dataset version, and the timestamp it was calculated at. This lets you audit exactly which method and which data release produced any figure.

Method choices you make (QMED source, excluded donors, pooling exclusions, distribution selection) travel with the analysis and its exports, so a report reflects the decisions actually taken.

It depends on the change. Selecting a different QMED method or growth-curve fit recomputes the downstream design flows and climate change instantly on the client.

Structural changes — editing descriptors, or excluding pooling/donor stations — require a rerun of the pipeline, triggered with the “Rerun analysis” action. As a rule: after editing anything upstream, rerun so every downstream figure reflects it.

A FEH XML descriptor export is parsed and its descriptors become the source of truth for the assessment, replacing any fetched values. This is the normal starting point for an ungauged assessment.

Imported descriptors must pass validation before they are used in the SM2025 equations. Design rainfall imported from a CD3/FEH DDF export is used exactly as given; where no DDF export is supplied, an approximate rainfall scaling is used instead (which the rainfall-runoff preview inherits).

QMED is locked to within about 1.2% of WINFAP. Pooled L-skew can diverge by roughly 6–9% for some urbanised catchments, because of a per-station L-moment de-urbanisation difference — but the resulting design flows still track WINFAP to within about 2%.

Where a difference matters for a submission, compare against your own WINFAP run and use the diagnostics and audit trail to explain the method choices.

From the dashboard choose New Analysis, then pick the gauged route (enter an NRFA station ID, e.g. 39001 for the Thames) or the ungauged route (import a FEH XML catchment file). You can also open the Station Explorer map and click a station to run it directly.

New here? Launch the guided tour from the “How do I use Hydrometric?” card on the dashboard.

Design outputs can be exported as:

• PDF and DOCX reports
• Per-table CSV / Excel flow workbooks
• TUFLOW model inflows (bc_dbase CSVs with .tcf/.tgc/.tbc run templates)
• Flood Modeller inflows (QTBDY .ied event files with .ief templates)

A combined TUFLOW + Flood Modeller package is available in one ZIP. Export to Google Docs is behind a feature flag and may not be enabled on every deployment.

Work through the analysis, confirm your QMED, pooling, growth-curve and design-flow choices, then open the report preview from the analysis header and export a PDF or DOCX. The report includes the design flows, method choices, diagnostics, dataset version and method citations needed to support a flood estimation.

The DOCX export is editable, so you can drop the results into your own flood estimation report template before issuing it. Always review the output against your project’s requirements — Hydrometric produces the hydrology, not the surrounding assessment narrative.

Every stage of the analysis tree has an info icon with an explanation of what the step does, what feeds it, and how to change it — and, for the core stages, a link through to the full methodology article. The analysis workspace also has a persistent help button that brings you back to this page.

Deeper background lives in the Research articles.

Use the feedback button in the bottom-right corner of any page. Reports are filed directly to the Hydrometric issue tracker so the team can pick them up. Include the station ID or catchment and what you expected versus what happened.

The free plan includes 5 assessments. The Professional / unlimited plan removes the cap. Your remaining allowance is shown on the New Assessment screen.

Browsing stations, reading research, and viewing saved analyses do not count against the assessment allowance.

Yes. Analyses can be grouped into projects, and teams can share access so colleagues work from the same catchment versions. Bulk upload lets you import FEH Web ZIP archives into shared projects for team-wide access.

Team management and invitations are handled from the dashboard.