Subtropical Cyclone Signal Climatology — Station Records and Return Periods

1. Warning signal framework

The Tropical Cyclone Warning Signal system in use across the region provides a graduated public alert scale keyed to sustained wind speed near the surface and to the expected trajectory of the circulation centre relative to the reference point. The scale has remained structurally stable since the 1973 revision, with the principal subsequent change being the 2002 redefinition of the directional signals, discussed in Section 6.

Signals are issued on the basis of sustained wind averaged over a ten-minute interval at a designated reference anemometer, not on instantaneous gust values. This distinction is material when comparing the record against archives that adopt a one-minute averaging convention; the ratio between the two conventions is discussed in Section 4.

Table 1. Warning signal definitions and associated sustained wind thresholds.
Signal	Designation	Sustained wind (km/h)	Operational meaning
T1	Standby	—	Circulation centre within 800 km; no immediate wind threat
T3	Strong wind	41–62	Strong winds expected or blowing generally
T8	Gale or storm	63–117	Gale-force winds expected from the quadrant indicated
T9	Increasing gale	63–117	Gale force winds expected to increase significantly
T10	Hurricane	≥118	Hurricane-force winds expected or blowing

Note on signal numbering. The non-consecutive numbering (1, 3, 8, 9, 10) is a historical artefact of the pre-1973 scale, in which signals 2 and 4 through 7 carried directional meanings later consolidated into the four T8 sub-signals. The numbering has been retained deliberately for continuity of public recognition.

2. Seasonal frequency and duration

Over the fifty-one year record, the mean annual count of signal issuances shows no statistically significant monotonic trend at the 95 per cent confidence level, although the interannual variance is substantial and correlates moderately with the boreal summer phase of the dominant basin-scale oscillation index.

Table 2. Mean annual issuance count and mean hoisted duration by signal, 1975–2025.
Signal	Mean issuances/yr	Std. deviation	Mean duration (h)	Max duration (h)
T1	6.2	2.14	14.1	61.5
T3	3.8	1.67	9.6	44.0
T8	1.1	1.02	6.2	21.5
T9	0.21	0.46	3.4	9.0
T10	0.16	0.42	4.8	11.0

2.1 Monthly distribution

Issuance is strongly seasonal. Ninety-four per cent of all T8-or-above issuances fall between June and October inclusive, with a pronounced September maximum. The shoulder months of May and November together account for approximately five per cent; the remaining fraction is distributed across isolated early- and late-season events, of which the December 1974 and April 2008 cases remain the most frequently cited outliers.

Table 3. Monthly distribution of T8-or-above issuances as a percentage of the annual total.
Month	Share (%)	Month	Share (%)
January–April	0.3	July	21.7
May	2.6	August	24.9
June	11.4	September	26.8
November	2.4	October	9.6
December	0.3

3. Station rainfall return periods

Return period estimates below are derived by fitting a generalised extreme value distribution to annual maximum series at each reference station, using the method of L-moments. Confidence intervals are obtained by parametric bootstrap with 10,000 resamples. Stations with fewer than thirty complete years of record have been excluded from the fit and are reported separately in the supplementary series.

Table 4. Estimated rainfall depth (mm) by duration and return period, composite reference station.
Duration	2-yr	10-yr	50-yr	100-yr	200-yr
15 min	24	37	49	54	59
1 h	53	84	113	126	139
3 h	89	146	201	226	252
6 h	118	198	277	313	350
12 h	151	256	362	411	461
24 h	188	320	456	519	584

Caution on extrapolation. The 200-year estimates are reported for completeness but rest on an extrapolation well beyond the length of the underlying record. Users requiring design values at that exceedance level should consult the regional pooled analysis rather than the single-station fit given here.

4. Wind field and gust factor observations

The gust factor — the ratio of peak three-second gust to ten-minute sustained wind — is not constant across the record and varies systematically with terrain exposure, measurement height, and the strength of the parent circulation. Values compiled from exposed coastal anemometers cluster considerably more tightly than those from stations with partial upwind obstruction.

Table 5. Observed gust factors by exposure class and sustained wind band.
Exposure class	41–62 km/h	63–87 km/h	88–117 km/h	≥118 km/h
Open coastal (E1)	1.31	1.28	1.26	1.24
Open inland (E2)	1.42	1.38	1.35	1.33
Partially sheltered (E3)	1.58	1.53	1.49	1.46
Urban / obstructed (E4)	1.79	1.71	1.65	1.61

4.1 Averaging convention conversion

For comparison against archives using a one-minute sustained convention, an approximate multiplicative factor of 1.14 may be applied to convert ten-minute values to one-minute equivalents over open water. This factor should not be applied over land without adjustment for surface roughness; over built terrain a value nearer 1.09 is more defensible. Conversion introduces an uncertainty of order five per cent and should be avoided entirely where the original observations remain accessible.

5. Instrumentation and siting notes

Reference anemometry across the network transitioned from cup-and-vane assemblies to ultrasonic sensors in a staggered programme between 1998 and 2006. Parallel operation was maintained at four stations for a minimum of twenty-four months to characterise the instrument difference; the resulting adjustment coefficients are applied to pre-transition values in the homogenised series but not in the raw archive.

Anemometer height. Standard exposure is 10 m above ground in open terrain. Six stations operate at non-standard heights owing to siting constraints; values from these are reduced to the 10 m equivalent using a logarithmic profile with a station-specific roughness length.
Sampling rate. Ultrasonic sensors sample at 4 Hz with a three-second running peak retained for gust reporting. Legacy cup assemblies sampled effectively at 1 Hz, which biases the recorded gust low by approximately two per cent.
Rain gauge type. Tipping-bucket gauges of 0.5 mm resolution are standard. Undercatch correction for wind-driven rain follows the standard field intercomparison relation and is applied to all values in the homogenised series.
Data logging. One-minute resolution has been archived network-wide since 1996. Prior to that date, hourly summaries are the finest resolution available for most stations, with a small number of manuscript records digitised at higher resolution under a retrospective programme.

6. Record homogeneity and known discontinuities

Users are advised that the series is not homogeneous across its full length. The following discontinuities are documented and, where quantifiable, corrected in the homogenised product:

Table 6. Documented discontinuities in the reference series.
Year	Nature of change	Effect on series
1982	Reference station relocation, 1.4 km	Mean wind speed step of −3.1%
1989	Surrounding development, upwind fetch reduced	Gradual gust factor drift, E2→E3 reclassification
1998–2006	Ultrasonic sensor transition	Gust bias +2.0%, corrected
2002	Directional signal redefinition	T8 sub-signal counts not comparable across boundary
2014	Undercatch correction methodology revised	Rainfall totals +1.2% mean, seasonally variable

The 2002 redefinition is the most consequential for signal-count analyses. Prior to that year, the four directional T8 sub-signals were issued on a criterion that combined observed wind direction with forecast centre track; afterwards, the criterion was simplified to forecast quadrant alone. Aggregate T8 counts remain comparable across the boundary, but sub-signal counts do not, and analyses spanning the discontinuity should aggregate to the parent signal before comparison.

7. References and data availability

The homogenised series, the raw archive, and the station metadata catalogue are maintained as a single versioned dataset. Citation should reference the version identifier, which changes whenever a homogenisation coefficient is revised.

Regional Observational Data Series, Technical Note 14: Homogenisation Procedures for Surface Wind Records, third revision.
Regional Observational Data Series, Technical Note 19: Extreme Value Analysis of Short-Duration Rainfall, second revision.
Field Intercomparison of Rainfall Intensity Gauges, final report of the working group on precipitation measurement.
Guide to Meteorological Instruments and Methods of Observation, Part I, Chapters 5 and 6.
Station Metadata Catalogue, current version, with siting photographs and exposure classifications for all reference stations.

Enquiries regarding data access, including requests for sub-hourly records predating 1996, should be directed to the data custodian through the standard request procedure. Bulk extracts are provided in CSV and NetCDF formats.