Skip to content

Diagnostics

The Diagnostics tab answers one question: is this dataset fit for frequency analysis? It runs on demand — load data, then click ▶ Run Diagnostics.

Controls

Record completeness

  • Exclude incomplete years — previews the record with calendar years below the threshold removed (a maximum drawn from a partial year is biased low).
  • Threshold (%) — the minimum share of expected records, at the detected resolution, for a year to count as usable.
  • Review flagged years… — keep specific below-threshold years (e.g. when a gap falls in a dry season). Run diagnostics first.
  • A record-adequacy readout reports the effective record length and a return-period reliability ceiling (roughly twice the record).

Checks (all on by default)

  • Gaps & timestamp integrity
  • Value plausibility (against the world-record rainfall envelope)
  • Stationarity & independence (Mann–Kendall trend with Sen's slope, Pettitt change point, lag-1 autocorrelation)

Each check's method, thresholds, and sources are summarised in the Methods overview.

Results

  • Six verdict cards — Record, Completeness, Timestamps, Values, Stationarity, Independence — coloured green / orange / red.
  • A coverage heatmap (with Pop out for report sizing).
  • A Diagnostics Report panel; the per-check boxes control what it includes.

Notes

Diagnostics does not configure the other tabs

The completeness setting here shapes only this tab's coverage preview and report. The Analysis and IDF tabs carry their own independent completeness controls.

See also: Analysis · Methods & References