Tropical cyclone forecast verification - southern hemisphere 2017-18

1. Introduction

A summary of tropical cyclone activity in the Southern Hemisphere for the 2017-18 season is presented below together with an assessment of the performance of the Met Office global model in predicting the tracks of these tropical cyclones.

Tropical cyclones are experienced in the South Indian Ocean, tropical seas to the west, north and east of Australia and the western and central South Pacific Ocean. For the purpose of tropical cyclone verification the Southern Hemisphere is divided in to two basins; the South-West Indian (west of 90° E) and Australian (east of 90° E). Mean error statistics for each basin are presented together with a table of statistics for the whole Southern Hemisphere. The global model produces a six-day forecast every 12 hours. Verification is performed at 12-hour intervals up to forecast time T+168, although statistics are only presented at 24-hour intervals in this report.

The global model resolution in operation during the season was 0.140625° x 0.09375° x 70 levels. This is equivalent to a horizontal resolution of 16 km × 10 km at the equator.

Tropical cyclone forecast verification - measures of error

Advisory positions from RSMCs La Réunion and Nadi, Fiji, Bureau of Meteorology, Australia and JTWC Hawaii are used as verifying observations of storm location. Best track data from these centres will be obtained once they become available. Past experience shows that use of best track rather than real time data usually only makes minor differences to seasonal track error statistics. Some mean error statistics for last season are also included for the purposes of a comparison. Forecast tracks are only verified when a depression reaches tropical storm status.

2. Tropical cyclone activity

Tropical cyclone activity
SWI AUS SAT TOTAL
Tropical depressions (<35 knots) 0 (0) 0 (2) 0 (0) 0 (2)
Tropical storms (35-63 knots) 3 (3) 9 (8) 1 (2) 13 (13)
Hurricanes/typhoons (>63 knots) 4 (2) 6 (5) 0 (0) 10 (7)
Total 7 (5) 15 (15) 1 (0) 23 (22)

Basin name abbreviations:
SWI : South-West Indian (west of 90° E)
AUS : Australian (east of 90° E)
SAT : South Atlantic

The number in brackets indicates the figure for the 2016-17 season.

N.B. 10-minute averaged wind speeds from RSMCs used when available. Data from JTWC (1-minute averaged) is only used when other data is unavailable and maximum wind speeds are scaled to make them equivalent to the RSMC 10-minute averages.

3. Summary of all southern hemisphere storms

3.1 South-west Indian basin storms

Table of mean error statistics
T+0 T+24 T+48 T+72 T+96 T+120 T+144 T+168
Cases verified 99 83 67 53 39 27 18 13
Detection rate (%) 100 100 100 100 100 100 100 100
AT error (km) 9 0 -20 -45 -52 -21 129 190
CT error (km) 1 0 -1 20 89 220 313 326
Track skill (%) ***** 66 71 74 ***** ***** ***** *****
* 2016-17 skill (%) ***** 48 69 74 ***** ***** ***** *****
DPE (km) 35 61 107 156 232 386 610 733
* 2016-17 DPE (km) 37 90 153 201 276 411 366 799
Central pressure bias (mb) -1.9 -2.0 -3.5 -7.5 -12.3 -13.2 -6.5 1.4

* DPE and skill for all south-west Indian storms in 2016-17 season

Plot of the observed tracks of all storms in the south-west Indian basin
Forecast positional errors in the south-west Indian basin
Forecast skill in the south-west Indian basin

There was much greater activity in this region this season than in the previous season. At lead times up to 96 hours track forecast errors were the lowest ever recorded. Skill scores against CLIPER were similar to the last few seasons. Track biases were mostly small with a slight polewards bias at longer lead times. The central pressure bias was negative at all lead times indicating model predictions were too intense by this measure.

3.2 Australian basin storms

Table of mean error statistics
T+0 T+24 T+48 T+72 T+96 T+120 T+144 T+168
Cases verified 113 86 60 38 27 19 16 14
Detection rate (%) 100 100 100 100 100 100 100 100
AT error (km) 1 -29 -45 16 57 87 97 -38
CT error (km) -1 1 6 27 93 212 78 62
Track skill (%) ***** 68 67 73 ***** ***** ***** *****
* 2016-17 skill (%) ***** 59 58 55 ***** ***** ***** *****
DPE (km) 38 88 144 190 220 215 226 325
* 2016-17 DPE (km) 61 124 209 308 468 538 442 434
Central pressure bias (mb) 2.1 0.6 -4.6 -9.3 -10.6 -10.8 -9.9 -10.8

* DPE and skill for all Australian storms in 2016-17 season

Plot of the observed tracks of all storms in the eastern Australian basin
Plot of the observed tracks of all storms in the western Australian basin
Forecast positional errors in the Australian basin
Forecast skill in the Australian basin

This basin was more active this season than last. Track forecast errors were lower than last season at all lead times and track forecast skill against CLIPER was higher than last season. Track forecast biases were small at all lead times. The central pressure bias was negative at most lead times.

3.3 South Atlantic basin storms

There was one short-lived South Atlantic subtropical storm this season. No verification is shown.

3.4 Combined Statistics for whole Southern Hemisphere

Table of mean error statistics
T+0 T+24 T+48 T+72 T+96 T+120 T+144 T+168
Cases verified 212 169 127 91 66 46 34 27
Detection rate (%) 100 100 100 100 100 100 100 100
AT error (km) 5 -15 -32 -20 -7 24 114 71
CT error (km) 0 0 2 23 91 179 202 189
Track skill (%) ***** 67 69 74 ***** ***** ***** *****
* 2016-17 skill (%) ***** 55 63 66 ***** ***** ***** *****
DPE (km) 37 75 125 171 227 315 429 521
* 2016-17 DPE (km) 53 111 188 266 372 474 404 617
Central pressure bias (mb) 0.2 -0.6 -4.0 -8.3 -11.6 -12.2 -8.1 -4.9

* DPE and skill for all southern hemisphere storms in 2016-17 season

The southern hemisphere had much increased activity compared to last season.

The 2017-18 season sees the completion of 30 years of tropical cyclone track forecast verification for the Met Office Global Model. This season track forecast errors were the lowest ever achieved at all lead times except 144 and 156 hours. Skill scores against CLIPER were the second highest ever achieved. Track forecast biases were low at short lead times with a slight polewards bias at longer lead times.

The forecast positional error graph shows the huge improvement in track forecast errors over the 30 years of model evaluation in the southern hemisphere. This is illustrated by the fact that in the 2017-18 season track forecast errors at 120 hours were lower than the errors in 24-hour forecasts 30 years earlier.

Forecast positional errors for the whole southern hemisphere
Forecast skill for the whole southern hemisphere
Along-track errors for the whole southern hemisphere
Cross-track errors for the whole southern hemisphere

There was no bias in central pressure in the analysis. The bias became negative in the forecast indicating model central pressures were lower than observed. 10m winds were too low at short lead times, but recovered to near zero at long lead times.

Central pressure forecast bias for the whole southern hemisphere
Peak 10m wind forecast bias for the whole southern hemisphere

Further tropical cyclone information

The Met Office tropical cyclone web pages contain information on tropical cyclone forecasting at the Met Office. Updates of tropical cyclone activity and forecasts are made monthly, together with observed and forecast track information of recent storms, track and intensity prediction error statistics, lists of names and real-time tropical cyclone forecast guidance.

For further information on tropical cyclone forecasting at the Met Office please email us.