On tropical cyclone activity in the Southern Hemisphere: Trends and the ENSO connection

Abstract

[1] A collective list of historical El Niño and La Niña events has been developed, based on an examination of different indices describing the El Niño-Southern Oscillation (ENSO) phenomenon. Based on this list, tropical cyclone (TC) data from a newly created TC archive for the Southern Hemisphere (SH) have been stratified accordingly and significant changes in TC occurrences depending on warm or cold phases of ENSO have been identified. TC trends in the SH (area south of the equator, 30°E to 120°W) have been examined. For the 1981/82 to 2005/06 TC seasons, there are no apparent trends in the total numbers and cyclone days of TCs, nor in numbers and cyclone days of severe TCs with minimum central pressure of 970 hPa or lower. However, significant positive trends in occurrences and cyclone days of severe TCs with minimum central pressure of 945 hPa or lower have been identified.

1. Introduction

[2] The El Niño-Southern Oscillation (ENSO) is the most important coupled ocean-atmosphere phenomenon affecting inter-annual climate variability on a global scale. ENSO consists of three phases: El Niño (the warm phase), La Niña (the cold phase) and the neutral phase. Despite the fact that ENSO has been the subject of numerous studies over recent decades, there is as yet no international agreement on a definition of ENSO and its phases. Usually, ENSO definitions are based on a quantitative analysis of (i) the intensity of the Southern Oscillation (SO) using indices linked with atmospheric pressure gradients and/or (ii) sea surface temperature (SST) anomalies in the Pacific (equatorial or near-equatorial regions). It is common practice to describe the strength of the SO in terms of a Southern Oscillation Index (SOI), for example the Troup Index [Troup, 1965], or mean sea level pressure at certain stations such as Darwin (Australia), Papeete (Tahiti), etc. Area-averaged SST anomalies (SSTAs) in various regions such as the NINO3, NINO3.4 and NINO4 regions are commonly used indices to describe oceanic conditions in the Pacific region. However, ENSO is a coupled ocean-atmosphere phenomenon, and combining both atmospheric and oceanic responses in a composite index seems to be a more appropriate index-based approach to a comprehensive description of the phenomenon than using atmospheric and oceanic indices alone. Lack of consensus on a definition of El Niño (La Niña) events typically results in certain warm (cold) episodes being included or excluded from lists of ENSO events compiled by different scientists.

[3] It has been shown that ENSO is a significant contributor to the year-to-year variability in tropical cyclone (TC) activity in most ocean basins [Nicholls, 1984; Chan, 2000; Landsea, 2000; Saunders et al., 2000; Kuleshov and de Hoedt, 2003]. Once again, the above-mentioned absence of agreement on defining ENSO warm and cold phases leads to investigators employing different criteria in the delineation of El Niño and La Niña events. Thus, NINO3.4 (5°N-5°S, 120°–170°W) SSTA thresholds of above +0.3°C (below −0.3°C) for identifying El Niño (La Niña) were used by Saunders et al. [2000] to investigate ENSO spatial impacts on occurrence and landfall of TCs in the Atlantic and Northwest (NW) Pacific basins. On the other hand, while studying TC activity over the NW Pacific, Chan [2000] referred to El Niño (La Niña) years if SSTAs in the NINO3.4 region rose above +0.5°C (fell below −0.5°C) sometime during that year. Consequently, even though significant ENSO impact on TC activity has been demonstrated, quantitative comparison of the results from different studies is somewhat difficult.

[4] In this paper, some key ENSO indices are examined in an attempt to develop a collective list of historical El Niño and La Niña events which may be recommended for subsequent studies on TC activity in the Southern Hemisphere (SH). Based on this list, TC seasons are stratified accordingly and changes in TC occurrences depending on warm and cold phases of ENSO are examined. Further, TC trends in the SH as the whole domain as well as in two sub-regions, the South Indian Ocean (SIO) (west of 135°E) and the South Pacific Ocean (SPO) (east of 135°E), are derived.

2. Data and Indices Examined

[5] The NINO3.4 SSTA and the SOI are the two most commonly used indices in defining ENSO phases. The SOI data used in this study were obtained from the Australian Bureau of Meteorology (www.bom.gov.au/climate/current/). Values for the NINO3.4 SSTA (3-month running mean) were obtained from the Climate Prediction Center, NOAA (ftp.cpc.ncep.noaa.gov/wd52dg/data/indices/).

[6] The Multivariate ENSO Index (MEI) [Wolter, 1987] was also used to quantify the start and end of ENSO events (www.cdc.noaa.gov/people/klaus.wolter/MEI/table.html). Another multivariate ENSO index, based on the first principal component of monthly Darwin mean sea level pressure (MSLP), Tahiti MSLP, and the NINO3, NINO3.4 and NINO4 SST indices, was developed and examined. Its base period is 1950–1999. We will denote this (standardised monthly) index the 5VAR index. It has a distinct advantage over the MEI in that it can be consistently extended further back in time.

[7] A new TC archive for the SH (the SHTC) has been prepared at the Australian Bureau of Meteorology's National Climate Centre. The data for the Australian region (90°E to 160°E) has been provided by the Australian Bureau of Meteorology, for the area from 30°E to 90°E by Météo-France (La Réunion) and for the area east of 160°E by the Meteorological Services of Fiji and New Zealand. TC tracks from these three archives have been merged in one homogeneous archive, ensuring consistency of trajectories and intensities when TCs cross regional borders. Record lengths of the TC archives and their quality vary. Australian region TC records are relatively complete only after meteorological satellites came in operational use in the late 1960s [Holland, 1981]. Solow and Nicholls [1990] further discussed changes in observations of TC activity in the Australian region over time arising from changes in observational capabilities. Examining the SHTC data, we confirm the findings of these previous studies and conclude that TC records in the SIO and SPO can be considered homogeneous from the 1981/82 TC season. The SHTC archive (available at www.bom.gov.au/climate/change/) now consists of TC best track data for TC seasons from 1969/70 to 2005/06. The data from the SHTC archive were used in this study to calculate trends and analyse the influence of ENSO on TC activity.

3. Evaluation of the MEI and the 5VAR Index for Identifying ENSO Phases

[8] Based on findings from early studies [Trenberth, 1997; Reid, 2003], the following assumptions have been made in this study: the Pacific exhibits El Niño conditions approximately 31% of the time, La Niña conditions 23%, and neutral conditions 46%. Bi-monthly MEI values (in multiples of the standard deviation) for the period from Dec 1949/Jan 1950 to Dec 2005/Jan 2006 were ranked from lowest to highest. Applying the above assumptions to the period 1950 to 2005 (56 years), a warm (cold) ENSO episode was identified if the MEI ranks for at least five consecutive months were greater than or equal to 39≈(1–0.31) × 56 (less than or equal to 13 ≈ 0.23 × 56). Another list of ENSO episodes was derived in an analogous fashion using the 5VAR index. The El Niño and La Niña events in these two lists, as identified by the MEI and 5VAR index ranks, were compared with the lists of ENSO events from Rasmusson and Carpenter [1982], Kiladis and van Loon [1988], Trenberth [1997], Larkin and Harrison [2001, 2002] and Reid [2003].

[9] For the SH, the TC year was considered as the 12 month period from July to June inclusive. In general, the main historically-recognized ENSO events included in the lists of examined publications were identified by the MEI rank and the 5VAR index rank using the selected thresholds. As a result, in this study the following TC years are considered as El Niño seasons: 1951/52, 1957/58, 1963/64, 1965/66, 1969/70, 1972/73, 1976/77, 1977/78, 1979/80, 1982/83, 1986/87, 1987/88, 1991/92, 1992/93, 1993/94, 1994/95, 1997/98, 2002/03 and 2004/05. The corresponding list of La Niña seasons is 1950/51, 1954/55, 1955/56, 1956/57, 1961/62, 1964/65, 1970/71, 1973/74, 1974/75, 1975/76, 1988/89, 1998/99 and 1999/2000.

4. TC Occurrences in the SH

[10] Based on the above considerations, the SHTC data were stratified between warm and cold ENSO phases to construct maps of average annual number of TCs in El Niño (Figure 1) and La Niña (Figure 2) years. A comparison of the maps demonstrates substantial shifts in the positions of maximum TC occurrences as well as changes in the intensity of the occurrence maxima. On average, around 25 and 29 TCs annually occur in the SH during El Niño and La Niña years, respectively. In general, TC activity is higher in the SIO than in the SPO. This difference is especially pronounced in La Niña years, with an average annual number of around 18 TCs (11 TCs) in the SIO (SPO).

Figure 1

Average annual number of TCs in the Southern Hemisphere in El Niño years in 2° × 2° boxes.

Figure 2

Average annual number of TCs in the Southern Hemisphere in La Niña years in 2° × 2° boxes.

[11] Tropical cyclogenesis (defined here as the point at which central pressure attains a threshold of 1000 hPa or lower) in El Niño (Figure 3) and La Niña (Figure 4) years has also been examined. In the SIO, one preferred area of cyclogenesis stays centred around 120°E during El Niño and La Niña years. However, another area of cyclogenesis which is located between 60°E and 85°E in El Niño years shifts eastwards to between 85°E and 105°E in La Niña years. In the SPO, the focus for cyclogenesis shifts eastwards in El Niño years in comparison to La Niña years, which is consistent with Basher and Zheng [1995].

Figure 3

Average annual number of TC genesis in the Southern Hemisphere in El Niño years in 2° × 2° boxes.

Figure 4

Average annual number of TC genesis in the Southern Hemisphere in La Niña years in 2° × 2° boxes.

5. Trends in TC Activity

[12] Concern about the enhanced greenhouse effect affecting TC frequency and intensity has grown over recent decades. Recently, trends in global TC activity for the period 1970 to 2004 have been examined by Webster et al. [2005]. They concluded that no global trend has yet emerged in the total number of tropical storms and hurricanes. However, they found a substantial change in the intensity distribution of TCs globally throughout the 37-year period. They conducted analyses for both hemispheres, but their SH analysis was limited to two areas (50°E to 115°E and 5°S to 20°S in the SIO and 155°E to 180° and 5°S to 20°S in the SPO).

[13] The new SHTC archive consists of the TC best track data for the area south of the equator, 30°E to 120°W, and it allows us to examine TC trends in the SH over the last 37 years (1969/70 to 2005/06 TC seasons). There are significant inter-annual variations in TC annual totals ranging from 16 to 34 TCs per year. Downward trends (statistically not significant) in the total annual number of TCs in the SH and in both sub-regions (SIO and SPO), have been identified. As TC activity in the SH is typically higher in La Niña years compared to El Niño years (on average 29 versus 25 TCs), we attribute this downward trend at least in part to the uneven temporal distribution of warm and cold ENSO episodes. Four La Niña (four El Niño) events occurred in the decade between 1969 and 1978, and three La Niña (eleven El Niño) events have occurred since 1979. In the Australian region, such an attribution of the decline in TC numbers to trends in ENSO was made by Nicholls et al. [1998], who also suggested that part of the decline was artificial and due to changes in the rules for classifying TCs. The SOI, the NINO3.4 SSTA index and the 5VAR index have been examined to establish trends in ENSO since 1969; the trends demonstrate that El Niño (La Niña) events have been more (less) frequent in recent decades.

[14] The SHTC archive has records starting from the 1969/70 TC season, but the period of records suitable for analysis of TC intensities is shorter: records of intensity for the SIO provided by Météo-France, La Réunion are complete only from the 1981/82 TC season. Therefore the series of the number of severe TCs with minimum central pressure of 970 hPa or lower (not shown) and of 945 hPa or lower (Figure 5), which approximately corresponds to TCs with intensity of winds greater that 59 m s⁻¹ (hurricanes categories 4 and 5), are from the 1981/82 TC season.

Figure 5

Occurrences of TCs with central pressure of 945 hPa or lower in the SH (30°E-120°W), the SIO (west of 135°E) and the SPO (east of 135°E), for the 1981/82 to 2005/06 TC seasons, with linear trends superimposed

[15] There are no apparent trends in total annual occurrences of TCs in the SH for the 1981/82 to 2005/06 TC seasons, nor in severe TCs with minimum central pressure of 970 hPa or lower (i.e., the calculated trends are not statistically significant). However, upward trends have been identified for occurrences of severe TCs with minimum central pressure of 945 hPa or lower (all trends are statistically significant; the SH trend is significant at the 1% level; the SIO and SPO trends are significant at the 5% level). These results, applicable for the whole SH, are in agreement with the findings of Webster et al. [2005] for the two smaller sub-areas of the SH which they studied, and confirm the earlier finding of a trend towards more intense TCs in the Australian region [Nicholls et al., 1998]. They are also consistent with Knutson and Tuleya [2004] who, investigating the impact of CO₂-induced warming on simulated hurricane intensity, found that greenhouse gas-induced warming may lead to a gradually increased risk in occurrences of category 5 hurricanes.

[16] The percentage of severe TCs with minimum central pressure of 945 hPa or lower in the SH has increased dramatically from around 10% in the early 1980s to around 30% in the mid-2000s, an increase which is statistically significant at the 1% level. Upward trends have also been identified in total TC days and proportion of TC days (that is, the ratio of severe TC days to all TC days) (both trends are statistically significant at the 1% level) for severe cyclones with minimum central pressure of 945 hPa or lower.

6. Summary

[17] In this paper, different indices which describe ENSO (the SOI, the SSTA NINO3.4 index, the MEI and the 5VAR index) have been examined, and a collective list of historical El Niño and La Niña events has been developed.

[18] Based on this list, TC data from a newly created TC archive for the SH have been stratified, and significant changes in TC occurrences (in terms of geographical distribution and intensity of maxima) depending on warm and cold phases of ENSO have been identified. The TC climatology for El Niño and La Niña years shows a geographical shift in the positions of maxima of TC occurrences in the SH. In the SIO, TC occurrences in El Niño (La Niña) years in the area between the east African coast and 75°E (between around 75°E and 135°E) are higher than in La Niña (El Niño) years. In the SPO, TC occurrences in El Niño (La Niña) years in the area east of around 170°E (between around 140°E and 170°E) are higher than in La Niña (El Niño) years.

[19] TC trends in the SH (specifically the area south of the equator, 30°E to 120°W), the SIO and the SPO have been examined. For the 1981/82 to 2005/06 TC seasons, there are no apparent trends in the total numbers and cyclone days of TCs, nor in numbers and cyclone days of severe TCs with minimum central pressure of 970 hPa or lower. However, significant positive trends in occurrences and cyclone days of severe TCs with minimum central pressure of 945 hPa or lower have been identified.

Acknowledgments

[20] The authors acknowledge the financial support of the Australian Greenhouse Office for the project “Climate Change and the Southern Hemisphere Tropical Cyclones.”