Seasonal forecasts of tropical cyclone numbers in the
Australian/southwest
Verification of 2003/04 and 2004/05 season forecasts and
forecast for 2005/06 season
contributed by Katrina A McDonnell1,
Neil J Holbrook1, and Hakeem Shaik2
1Department of Physical Geography,
2Bureau of Meteorology,
Introduction
Recent statistical
model forecasts of the number of tropical cyclones expected to form (following
the conventional definition, tropical cyclone formation (genesis) is identified
to occur at the time and location where the wind speed first exceeds 34 knots
(17.5 ms-1)) in the Australian/southwest Pacific Ocean region
(6°-20°S, 105°-170°E) during the tropical cyclone season (November-May) are
verified here, and forecasts for the upcoming 2005/06 season are
provided. The model has been designed and developed based on a Poisson
regression approach using the September lead (i.e., immediately prior to
the upcoming tropical cyclone season) saturated equivalent potential
temperature gradient (EPT) and September lead Southern Oscillation index (SOI)
as predictor variables (McDonnell and Holbrook 2004a,b). In physical
terms, the EPT depends on the atmospheric vertical temperature profile and is a
measure of the potential for cumulonimbus convection from a lapse-rate
stability viewpoint. With this model, skilful hindcasts
(i.e., "predicting" back in time) and forecasts of the numbers of
tropical cyclones formed in the Australian/southwest Pacific Ocean region have
been shown to be possible (McDonnell and Holbrook 2004a,b).
For the
development and application of this model, cyclone genesis occurrence points
are determined from the Australian/southwest Pacific Ocean region tropical
storm data (obtained from the Australian Bureau of Meteorology via
http://www.bom.gov.au/climate/how/) and binned into monthly (November-May)
cells in time and 2° latitude x 5° longitude boxes in space across the region
(see McDonnell and Holbrook 2004a). The EPT field used in the model is
calculated from the September average (prior to the upcoming tropical cyclone
season) temperature fields at 1000hPa and 500hPa from National Center for
Environmental Prediction (NCEP)-NCAR reanalysis data but spatially coarsened
onto the same 2° latitude x 5° longitude grid [the September 2005 EPT field
used in the model to forecast the 2005/06 season tropical cyclone numbers
presented here was calculated from the 2005 September-average gridded temperatures at 1000hPa and 500hPa from the Climate
Data Assimilation System (CDAS)-NCEP/NCAR Reanalysis data]. The
temporally- and spatially-varying EPT (at the 2° latitude x 5° longitude grid
scale) offers the potential for probabilistic forecasting of not only the number
of tropical cyclones formed, but also in what part of the season and in what
location they might form. We note, however, that the current model
configuration is such that it is most reliable at the broader subregional (defined as the Western (105°-125°E), Northern
(125°-145°E) and Eastern (145°-170°E) subregions) and
subseasonal (defined as early (November-December),
mid (January-March) and late (April-May) season) scales or larger as defined in
McDonnell and Holbrook (2004b). At these smaller subregional
scales, the temporal variability of the seasonal number of tropical cyclones
formed in the Eastern (Coral Sea) subregion, where El
Niño-Southern Oscillation (and hence SOI) influences may be expected to be
strongest, is hindcast with the most skill.
Forecast
verification of 2003/04 season tropical cyclone numbers
Figure 1 shows the
observed seasonal number of tropical cyclones formed and the cross-validated hindcasts for the years 1960/61-1992/93. Also shown
are independent forecasts (hindcasts) of the 11 tropical
cyclone seasons 1993/94-2003/04, plus results from last year's forecast of
tropical cyclone numbers for the 2004/05 season and the forecast provided here
for the upcoming 2005/06 season (McDonnell and Holbrook 2004c,d), using both
forecast models. We note that the Poisson regression model using the
September lead EPT forecast 7 (6.8) tropical cyclones to form across the
broader Australian/southwest Pacific Ocean region and 2 (2.3) to form in the
smaller
Forecast
verification of 2004/05 season tropical cyclone numbers
To coincide with
the commencement of the 2004/05 tropical cyclone season McDonnell and Holbrook
(2004c,d) provided published model forecasts of the
number of tropical cyclones expected to form in the Australian/southwest
Pacific Ocean region and the smaller Coral Sea subregion
using two versions of their Poisson regression model: (i)
a September lead SOI predictor model, and (ii) a September lead EPT predictor model. The
monthly-averaged SOI for September 2004 was -2.8. Using their SOI
predictor Poisson regression model, McDonnell and Holbrook forecast a total of
8 (7.9) tropical cyclones to form in the Australian/southwest
Using the SOI and
EPT predictor Poisson regression models, totals of 3 (2.8) and 2 (2.4) tropical
cyclones were respectively forecasted to form in the
Six tropical
cyclones were eventually observed to form within the Australian/southwest
Pacific Ocean region during the 2004/05 season, one of which occurred within
the
Forecasts of
2005/06 season tropical cyclone numbers
The
monthly-averaged SOI for September 2005 is +3.9. Using the SOI predictor
Poisson regression model we forecast a total of 9 (8.6) tropical cyclones to
form in the Australian/southwest
As with the
2004/05 forecasts, the Poisson regression model using the spatially-varying EPT
predictor variable produces more conservative forecasts of the number of
tropical cyclones expected to form across both regions in the upcoming
season. The present Poisson regression formulation does not currently
provide useful forecasts at the 2° x 5° grid scale due to the simple representation
of space in the model. Nevertheless, these Poisson regression models have
been shown to provide skilful aggregates of expected occurrences on subregional scales, and in time. We caution, however,
that the forecasts reported here are experimental in nature. The reader is
advised that the methods and forecasts are subject to future change and
improvement.
References
McDonnell, K. A. and N. J. Holbrook, 2004a: A Poisson
regression model of tropical cyclogenesis for the
Australian-southwest
McDonnell, K. A. and N. J. Holbrook, 2004b: A Poisson
regression model approach to predicting tropical cyclogenesis
in the Australian/southwest
McDonnell, K. A. and N. J. Holbrook, 2004c: 2004/05
tropical cyclone forecasts for the Australian-southwest
McDonnell, K. A. and N. J. Holbrook, 2004d: Tropical
cyclone seasonal forecasts for the Australian-southwest
Table
1. Summary record of tropical cyclones
that formed in the vicinity of the Australian/southwest
|
Occurrence
Dates |
Formation
Latitude (o) |
Formation
Longitude (o) |
Maximum
wind speed (knots) |
|
|
Australian
region (105° E - 165° E) |
|
|||
|
Raymond |
2
-2 Jan |
14.0S |
123.6E |
40 |
|
Tim |
23
-25 Jan |
15.3S |
109.9E |
45 |
|
|
6
-7 Feb |
14.0S |
139.7E |
85 |
|
Vivienne |
8 -8 Feb |
15.3S |
115.9E |
45 |
|
Ingrid |
6
- 16 Mar |
13.6S |
148.7E |
120 |
|
Willy |
10-14
Mar |
14.0S |
117.5E |
75 |
|
|
|
|||
|
Kerry |
5
-14 Jan |
13.4S |
171.5E |
75 |
Table
2. Poisson regression
coefficients (βj) and standard errors
(SE) estimated over the period 1960/61-2004/05, for both the September lead SOI
model and the September lead EPT model.
|
SOI
model βj |
SE |
EPT
model βj |
SE |
|
|
Intercept |
-4.58 |
0.07 |
-4.63 |
0.07 |
|
SOIsep |
0.17 |
0.06 |
- |
- |
|
EPTsep |
- |
- |
0.42 |
0.06 |
|
Month |
-0.50 |
0.07 |
-0.50 |
0.07 |
|
Cyclone
year |
0.06 |
0.06 |
0.08 |
0.06 |
|
Latitude |
-0.47 |
0.06 |
-0.72 |
0.06 |
|
Longitude |
-0.29 |
0.05 |
-0.32 |
0.06 |
Figure
1. The observed (solid
line) and cross-validated hindcast (dashed line)
seasonal numbers of Australian/southwest Pacific Ocean tropical cyclones formed
during the period 1960/61-1992/93 using the (a) SOI, and (b) EPT predictor
Poisson regression models. Independent tropical cyclone genesis
observations (o) for 1993/94-2004/05 are also shown, along with the forecasts
(x) for 1993/94-2005/06 using both models.