New Zealand climate varies significantly from year to year and from decade to decade. Human-induced long-term trends will be superimposed on these natural variations, and it is this combination that will provide the future climate extremes to which New Zealand society will be exposed.
New Zealand-wide temperatures can deviate from the long-term average by up to 1°C (plus or minus) on an annual basis, whereas regional precipitation can deviate by about 20% (plus or minus). The sign of the deviation will depend on whether it is a La Niña or an El Niño year, and also (for precipitation) on geographic location. Details can vary considerably from one event to another. These variations of the climate in individual years have amplitudes comparable to the mid-range projected changes expected over 30 to 50 years.
New Zealand also has decadal circulation and climate variations that appear to be related to the Interdecadal Pacific Oscillation (IPO). The predictability of the IPO, and how consistently it is reflected in local climate, is still a topic of active research.
The temperature increases projected for 2040 and 2090 lie generally above the linear extrapolation of the historical New Zealand temperature record (1908–2006).
The projected changes in New Zealand’s climate, as discussed in chapter 2, must be viewed within the context of natural yearly and decadal variability in circulation and climate. Chapter 3 provides this context by briefly summarising variations in current climate, and comparing the projected climate changes with past changes and variability.
New Zealand’s climate varies all the time, and this natural variability will be superimposed on the future long-term trends described in chapter 2. Much of the variation in New Zealand climate is random and short-lived, but some of the variations are quasi-cyclic38 in nature and some have long spans, lasting from seasons to years to decades. Figure 3.1 shows the historical national-average temperature for New Zealand, and sets the scene for discussing past and future changes and variability. This particular time series is derived by combining records from seven long-term climate stations (Auckland, Masterton, Wellington, Nelson, Hokitika, Christchurch, and Dunedin).
Records from all seven sites are available from 1908; since this date, New Zealand temperatures have increased by 0.90°C (ie, the linear trend between 1908 and 2006, as marked on Figure 3.1). A linear trend fitted to the New Zealand annual temperature record is statistically significant for data starting 1950 or earlier.39 Global temperature trends achieve significance over much shorter periods than is found with New Zealand data; this is understandable because, at the regional scale, circulation patterns such as the El Niño-Southern Oscillation move heat back and forth and increase the interannual variability. At the global scale, this natural climatic ‘noise’ is evened out, and the reduced variability means the global warming signal can be detected earlier.
Note: The figures show deviations from the climatological values for 1980–1999 for consistency with the baseline used for the projected values. Upper panel: annual mean temperatures (histogram bars, red if they are higher than the 1980–1999 mean) and the linear trend (solid black line). Lower panel: M1 Southerly Index (histogram bars, red when it is more northerly than the 1980–1999 mean; these periods generally match with relatively warm years in upper panel) and the same data smoothed (black line) to show the decadal M1 variation and highlight the trend towards more southerly airflow since the mid-1950s.
For the New Zealand region, a good measure of the circulation variability that influences temperature is given by the Trenberth M1 Index40 and a plot of the M1 Index has been included in Figure 3.1: more positive index values imply stronger mean southerly flows into the Tasman-New Zealand region. Hence, positive values (more southerly) tend to coincide with colder years in the New Zealand record, and negative values (more northerly) with warmer years. When the linear regression equation between the temperature and M1 data is calculated, then the residual temperature series (with the effect of northerly–southerly fluctuations removed) shows a significant warming trend much sooner (1986 or earlier). Moreover, Figure 3.1 shows that New Zealand has become warmer in spite of more southerly airflow in recent decades.
Figure 3.2: New Zealand temperature (in °C) – historical record, and schematic projections illustrating an example of future year-to-year variability (upper panel) and the full range of multi-decadal warming over the IPCC emissions scenarios (lower panel).
Note: Temperature is deviation in °C from the 1980–1999 mean. Upper panel: annual mean temperature, shown in red if it is higher than the 1980–1999 mean. For the observed period (1908–2006) the temperature shown is the average over seven long-term NIWA climate stations. For the schematic projection period (2007–2100) the temperature shown is the national-average of year-by-year downscaled temperatures from one of the Fourth Assessment models. Lower panel: Data from the upper panel displayed as a line plot (black) and a smoothed curve (blue line) showing the decadal variation in the observations and the example model projection. Vertical bars at 2040 and 2090 represent the full IPCC range of New Zealand warming in the six IPCC emission scenarios. The linear extrapolation to 2100 of the observed 1908–2006 trend line is also shown (black dashed line).
Figure 3.2 illustrates how natural variations will be superimposed on the long-term warming trend. It extends the historical temperature record by appending scenarios out to 2100.
In the upper panel, the historical period of 1908–2006 exactly reproduces Figure 3.1. Appended to this record is a time series of downscaled changes in New Zealand-average annual temperature, as simulated by one of the global climate models41. In this case, the downscaling has been applied year by year, instead of for the 20-year averages used in chapter 2. Natural variability (or at least the model’s simulation of it) causes large year-to-year fluctuations about the general warming trend, and these fluctuations appear to be of similar magnitude to those observed historically. For example, 2043 (in this particular simulation) has a temperature similar to the current 1980–1999 climate, although it would be an abnormally cold year in the context of the 2040s. This model projection in Figure 3.2 is presented as a schematic only: the exact sequencing of future cold and warm years is random relative to how the actual future climate will evolve. A different climate model would give different sequencing, and indeed so would the same model if the simulation was run on another computer or with another start year.42
The lower panel of Figure 3.2 presents the broader context across the range of models and emission scenarios considered in chapter 2. The historical temperature record is given by the (black) line plot, to which has been added a smoothed curve (blue line) representing the decadal variation (observed and projected) from the example in the top panel. Two further features have been added. The linear temperature trend observed over 1908–2006 has been extrapolated to 2100 (dashed black line). Coloured vertical bars show the full scenario range at 2040 and 2090,43 taking account of the different sensitivities of the various climate models and the six IPCC illustrative emission scenarios.
The extrapolation of past temperatures lies near the lower bound of the future projections. Only the combination of a low emissions scenario and a low sensitivity model gives a temperature at 2100 that is close to that extrapolated from the historical record. Note that the temperature projections in the lower panel of Figure 3.2 are decadally-smoothed curves. We would expect some individual years to fall outside the envelope, especially in the early years of the 21st century.
New Zealand’s climate varies naturally with fluctuations in the prevailing westerlies and in the strength of the subtropical high-pressure belt. Local climate changes often have a strong spatial pattern imposed on them by interactions between the circulation and the southwest/northeast alpine ranges. Many of the circulation fluctuations that affect New Zealand are short-lived or random. However, other changes are associated with large-scale patterns over the Southern Hemisphere or Pacific Ocean. There are a number of key natural oscillations that operate over timescales of seasons to decades. This section focuses particularly on the El Niño-Southern Oscillation (ENSO, operating on the interannual timeframe) and the Interdecadal Pacific Oscillation (IPO, which persists in one phase for two or three decades).
Other factors that affect New Zealand’s climate include large volcanic eruptions in the tropics (leading to cooling for a year or more),44 and possibly solar variations over a range of timescales. On the extremely long timescale of thousands of years, there are the well-documented ice age cycles caused by systematic and predictable variations in the earth’s orbit, but these do not concern us here.45 On timescales shorter than 1 year, the most significant oscillation affecting New Zealand is the Antarctic Oscillation (also known as the ‘High Latitude Mode’46). This oscillation appears to change sign on a month-to-month basis with very little predictability. Recent work has identified a long-term trend towards a stronger positive phase in the Antarctic Oscillation (meaning stronger westerlies at 50°S), which has been related to trends in stratospheric ozone depletion. This trend is also reproduced in climate model studies driven by greenhouse gas increases, so it is likely that both ozone and carbon dioxide contribute to the changes observed in high-latitude circulation.47
The Southern Oscillation, or more generally ENSO, is a tropical Pacific-wide oscillation that affects pressure, winds, sea-surface temperature (SST) and rainfall. In the El Niño phase, the easterly trade winds weaken and SSTs in the eastern tropical Pacific can become several degrees warmer than normal. There is a systematic eastward shift of convection out into the Pacific. Australia then experiences higher pressures and droughts, while New Zealand experiences stronger than normal southwesterly airflow. This generally results in lower seasonal temperatures for New Zealand, and drier conditions in the northeast of the country. The La Niña phase is essentially the opposite in the tropical Pacific, and New Zealand experiences more northeasterly flows, higher temperatures and wetter conditions in the north and east of the North Island. Pressures tend to be higher than normal over the South Island, and this can lead to drought conditions in the south. Thus, drought can occur in New Zealand in both El Niño and La Niña phases. Figure 3.3 shows average rainfall anomalies (that is, the differences from the multi-year averages) in New Zealand associated with El Niño and La Niña summers. Individual ENSO episodes can differ substantially from the average pattern.
Figure 3.4 (upper panel) shows a time series of the Southern Oscillation Index (SOI), a common measure of the intensity and state of ENSO events derived from the pressure difference between Tahiti and Darwin. Persistence of the SOI below about –1 coincides with El Niño events, and periods above +1 with La Niña events. Because the tropical Pacific SST anomalies persist for up to a year, there is substantial predictability in how ENSO events affect New Zealand’s climate, and there has been considerable research to identify local impacts.48 The ENSO cycle varies between about 3 and 7 years and there is large variability in the intensity of individual events.
Figure 3.3: Differences between the long-term average rainfall and that in ENSO years (the rainfall anomaly), in percent, for summer (December, January, February). The ENSO rainfall is the average of the 10 strongest ENSO events between 1960–2007. (The insert box shows the ENSO years, where 1964 is December 1963 to February 1964, etc.)
There has been an increase in the frequency of El Niño events since the late 1970s, and there has been much debate about whether this is a consequence of global warming.49 The issue is still not settled (see chapter 2). Another possible explanation for increased El Niño activity in the last two decades is minor natural variability in climate. The IPO has been shown to be associated with decadal climate variability over parts of the Pacific Basin,50 and to modulate interannual ENSO climate variability over Australia51 and New Zealand.52 A time series of the IPO, derived from a UK MetOffice analysis of global SST patterns is shown in Figure 3.4 (lower panel). Three phases of the IPO have been identified during the 20th century: a positive phase (1925–1943), a negative phase (1944–1977), and another positive phase (1978–1999). The pattern associated with the positive phase is higher SSTs in the tropical Pacific (more El Niño-like) and colder conditions in the North Pacific. Around New Zealand, the SSTs tend to be lower, and westerly winds stronger.
Long-lived fluctuations in New Zealand climate show some association with IPO changes. The increase in New Zealand temperatures around 1950 (Figure 3.1) occurred shortly after the change from positive to negative-phase IPO (Figure 3.4). The switch from negative to positive IPO in the late 1970s coincided with significant rainfall changes. Figure 3.5 maps annual rainfall changes between negative and positive IPO periods centred on 1978, and Figure 3.6 shows the corresponding rainfall time series for the southwest part of New Zealand. In the later (positive IPO) period, rainfall increased in the southwest of the South Island, but decreased in the north and east of the North Island, relative to the earlier (negative IPO) period.
Figure 3.5: Percentage change in average annual rainfall for 1978–1998 compared to 1960–1977.
Figure 3.5 shows the percentage change in average annual rainfall for 1978 to 1998 compared to the period from 1960 to 1977 (as a percentage of the 1961-1990 average). The south and west of the South Island was more than eight percent wetter; the north of the South Island was up to eight percent wetter; the north and east of the North Island was up to eight percent drier; and in the remaining areas of New Zealand the differences were close to zero.
Note: In 1978–1998 the IPO was in its positive phase, compared to preceding 18 years, when the IPO was negative. Any local rainfall response due to global warming would also be contained within this pattern of rainfall trends.
Two main circulation changes affecting New Zealand in 1930–1994 have been identified as occurring around 1950 and 1975.53 The period 1930–1950 was one of more south to southwest flow. Temperatures in all regions were lower in this period. Wetter conditions occurred in North Canterbury, particularly in summer, and drier conditions in the north and west of the South Island. In 1951–1975 (corresponding approximately with the negative IPO phase), there was increased airflow from the east and northeast, and temperatures in all regions increased. Conditions became wetter in the north of the North Island, particularly in autumn, and drier in the southeast of the South Island, particularly in summer. From 1976 onwards, west to southwest flow was more frequent, with little additional warming relative to the 1951–1975 period. There were significant rainfall trends, with summers becoming drier in the east of the North Island and wetter in the southeast of the South Island, and winters becoming wetter in the north of the South Island.54
What are the implications for future climate of a shift in the IPO phase? Analysis of the sea temperature data for up to about the year 2000 suggested the IPO had switched to the negative phase in 1998. For several years subsequent to 2000, it was unclear whether the 1998 shift was a true shift to the negative phase.55 However, the 2007 UK MetOffice analysis (Figure 3.4) seems to confirm that the negative phase is now established.56 If this IPO phase persists, then more La Niña (and less El Niño) activity could be expected compared to the 1978–1999 period. Weaker westerlies are likely, along with an implied weakening in the west–east rainfall gradient across the country. This gradient would act to partially counter the projected anthropogenic trend of increasing westerlies for perhaps the next 20–30 years or so, but at the same time increase the rate of New Zealand temperature and sea-level rise above the trend expected from global warming.
Trends in New Zealand’s historic daily temperature and rainfall extremes have been calculated.57 Some of the trends (such as a decrease in frequency of frosts) are in agreement with trends in the global climate model projections, but most of the observed past changes have marked temporal variations and can be related qualitatively to regional decadal circulation changes as outlined in section 3.2.1 above.
Note: Upper panel: Average number of frosts per year in three Regional Council regions, 1972-2006, from NIWA gridded daily minimum temperature data set. Lower panel: Number of frosts per year at Dannevirke (NIWA station number D06212) in the eastern part of the Manawatu-Wanganui region, 1951–2006. (Dannevirke is taken for this example because it is the station nearest to the eastern Wellington region with a long continuous record – see text.) The straight trend lines in both panels are least-squares fits to the annual values.
Higher mean temperatures obviously increase the probability of extreme warm days and decrease the probability of extreme cold days. The IPCC also notes that climate models forecast a decrease in diurnal temperature range at many locations;58 that is, the nighttime minimum increases faster than the daytime maximum. The evidence for increasing numbers of very warm days in New Zealand is not consistent, with regionally varying patterns that can be related to circulation fluctuations. However, there is clear evidence of a decreasing number of frost days at many New Zealand sites, as can be seen in Figure 3.7. For example, both Canterbury and Marlborough regions show about 20 fewer frosts per year now than in the early 1970s. However, the change in frost occurrence is not uniform, and shorter periods can give counter-intuitive results. Figure 3.7 (upper panel) actually shows an increase since 1972 in the frost incidence in the Wellington regional council region, which is due to increases in the eastern (Wairarapa) side of the lower North Island. This tendency for increased frost since 1972 is also evident in other east coast sites, including Gisborne, south Canterbury and coastal Otago. However, even for these sites, there has been a decrease in frosts on the longer timescale (eg, lower panel of Figure 3.7, record from 1951). Analyses of New Zealand temperature records have shown that trends in maximum and minimum temperatures are strongly linked to atmospheric circulation changes; slight changes in prevailing airflow direction can affect the frequencies of frosts and hot days in localised areas.59
Historical changes in New Zealand’s extreme rainfall have also been documented.60 The variations in extremes are quantified by measures such as the annual 95th percentile amount of rainfall, or number of days per year with rain exceeding the long-term, mean 95th percentile. Changes in extreme daily rainfalls are strongly related to changes in mean rainfall. Station 1-day rainfall extremes were highly correlated to westerly circulation across the country. Thus, increases in mean and extreme daily rainfall over 1930–2004 were found in the west of the country, and decreases were found in the north and east of New Zealand.
Note: AEP is ‘annual exceedance probability’. The fitted lines are Gumbel, Extreme Value Type 1, distributions fitted using Probability Weighted Moments. Figure from McKerchar and Henderson (2003).
Analysis of historical records of extreme river flows, both very low flows and floods,61 has shown very marked changes in the frequency of extreme flows with the phase of the IPO in some parts of New Zealand. A decrease in flood size has occurred since 1978 in the Bay of Plenty, and increases in flood size have occurred in the South Island for most rivers with headwaters draining from the main divide of the Southern Alps and in Southland. An example is given in Figure 3.8, which shows an analysis of flood return periods for Lake Te Anau. A high flow with an estimated return period of 50 years during 1947–1977 (a period of negative phase IPO, the lower line in the figure) has a return period of approximately 7 years during 1978–1994 (a period of positive phase IPO, upper line).
Whilst there appears to be a public perception that ‘increased storminess’ is likely under climate change, the evidence of changes in storminess in New Zealand to date is far from conclusive. The concept of ‘storminess’ is itself somewhat ambiguous: it could refer to the number of storms, or to storm intensity – which, in turn, could be judged on the basis of strong winds or heavy rainfall. Storms can also approach New Zealand from the subtropics and from mid-latitudes (extra-tropics), and different trends are possible in those two regions.
The IPCC Third Assessment reported that:62
Changes globally in tropical and extra-tropical storm intensity and frequency are dominated by inter-decadal to multi-decadal variations, with no significant trend evident over the 20th century. Conflicting analyses make it difficult to draw definitive conclusions about changes in storm activity, especially in the extra-tropics.
Since 2001, there have been a number of articles published about increases in intense tropical cyclones. These results are still controversial, and the Fourth Assessment Report was cautious in its conclusions:63
There is observational evidence for an increase in intense tropical cyclone activity in the North Atlantic since 1970 … There are also suggestions of increased intense tropical cyclone activity in some other regions where concerns over data quality are greater.’
The only conclusion about mid-latitude circulation changes that was sufficiently clear-cut to bring forward into the Summary for Policymakers was:
Mid-latitude westerly winds have strengthened in both hemispheres since the 1960s.
A limited number of studies have described observed changes in the Southern Hemisphere relevant to New Zealand, and most of these have been summarised in the IPCC Fourth Assessment Report.64 Tropical cyclones that develop in the southwest Pacific could affect New Zealand. From 1971 to 2004, tropical cyclones in this region averaged nine per year, with no observed trend in either frequency65 or intensity.66 In any case, only about one cyclone per year moves south of 30°S and comes close enough to New Zealand to have a direct impact. Thus, there has been no increase in New Zealand’s storminess from this source to date.
Trends in the frequency and strength of extreme winds across New Zealand since the 1960s have been examined.67 These show an increase in the number and strength of extreme westerly wind episodes to the south of the country, but only a slight increase over New Zealand itself. At the same time, extreme easterlies decreased across New Zealand. These local changes relate well to the observed increase in southern hemisphere westerly wind-flow during the latter part of the 20th century noted earlier. These trends are also consistent with climate model simulations for an increasing trend in what is known as the ‘positive phase of the Southern Annular Mode’, which has been linked both to increases in greenhouse gases and to the size of the ozone hole over Antarctica. Thus, there is some evidence of an increase in westerly ‘storminess’ (ie, strong westerly wind episodes) in the late 20th century in the New Zealand region.
In terms of increasing heavy rainfall, it has already been noted that an increase has been observed in extreme precipitation in western parts of the North and South Islands, but also decreases in extremes in eastern regions.68 While a contribution to these trends from global warming cannot be ruled out, the simplest explanation is natural decadal variability (in this instance, the IPO described above).
Several recent studies have been made of trends in southern hemisphere extra-tropical cyclones. Over the period 1979–1999, there has been about a 50% increase in the number of explosively deepening cyclones (the so-called ‘weather bombs’) per year.69 These rapidly deepening systems occur mainly to the south of 50°S, but can form in the western Tasman Sea in the winter. However, they are a small percentage (around 1% or less, depending on location) of the total number of cyclones.
Changes in the number of Southern Hemisphere cyclones have also been documented.70 Over the 40 year period 1958–1997, there has been a general reduction in the mean cyclone density over most regions south of 40°S, with the greatest reductions near 60°S but little change in the Tasman Sea. At the same time, systems have become more intense, on average, in the Australian Bight and the Tasman Sea, and weaker over the eastern Pacific. Just why the reduction in overall numbers should be occurring is not well understood, although one modelling study71 has suggested that under more moist conditions (as would occur in a warmer atmosphere) cyclonic eddies transfer energy poleward more efficiently, and thus fewer cyclones would be ‘required’ to effect the same energy transport.
Observations dating back to the early to mid-19th century show that sea level is rising around New Zealand. The historic rate of rise has been around 1.6 mm/year, based on analysis of tide-gauge data from the four main ports (Auckland, Wellington, Lyttelton and Dunedin).72 This value also lies midway in the range of estimated global sea-level rise of between 1 mm and 2.5 mm per year since the early 1800s.
There is no sign yet of any definitive acceleration in the rise of sea level from any New Zealand sea-level gauges, although detection of such an acceleration has been claimed73 for the global record. However, the Fourth Assessment Report of the IPCC concludes that it is unclear whether the recent faster rate of sea-level rise is a trend or is just a reflection of natural decadal variability.
The IPO, which spends 20–30 years in each phase, appears to have switched around 1999–2000 to the negative phase. If this current negative phase dominates for the next two decades, it is likely to bring more La Niña episodes than seen over the last 20 years, and produce a faster rate of sea-level rise locally than that experienced over the previous positive phase of IPO from 1978 to 1999. This pattern of more rapid sea-level rise during negative phases of the IPO has been demonstrated from the Port of Auckland tide-gauge record, and has been observed again in 2000–2005.74 Other records show that a similar trend is occurring around the southern North Island; therefore, the next 20–30 years should see a faster rise in sea level than that attributable to the mean long-term trend of 1.6 mm per year. This local acceleration of sea-level rise is irrespective of any changes in the rate of sea-level rise attributable to global warming.
38 There are very few true cycles in climate records, apart from the daily (diurnal) and annual cycle, in the sense of having a clearly defined period and being predictable for many cycles into the future. For this reason, the term ‘oscillation’ is often used in preference to ‘cycle’.
39 Linear trends for shorter periods (1951–2006, 1952–2006, etc) have a p-value exceeding 0.05, and are, therefore, not significant at the 5% level.
40 M1 is defined as the normalised pressure difference between Hobart and Chatham Islands. See Trenberth 1976; Jones et al 1999.
41 The gfdl_cm21 model. (It projects a warming of about 2°C by 2090. Other models have similar realistic interannual variability, but different rates of warming. See Appendix 2.) Changes are relative to the model average over 1980–1999, and are for the national average calculated over all land points of the gridded dataset used for the downscaling (Appendix 3), and not just at grid points co-located with the seven-station dataset used for the historical record.
42 Chaotic behaviour in the atmosphere not only affects weather forecasts over the next few days, but also feeds into interannual and multi-decadal variability over the climate timescale.
43 As in chapter 2, the nominal years ‘2040’ and ‘2090’ actually represent an average over the 20-year periods 2030–2049 and 2080–2099, respectively.
44 Salinger (1998) quantified the effects of large tropical volcanic eruptions on New Zealand climate; on average, they lead to a cooling of 0.6–0.8°C over 1–2 years. The impact of the May 1991 Mt Pinatubo eruption is clearly evident in the New Zealand temperature record during 1992.
45 Hays et al. (1976) predicted a cooling trend over the next several thousand years and glacial conditions in 20,000 years time. However, this prediction of very long-term cooling was based on only natural changes in radiative forcing (due to Earth orbit changes), with anthropogenic effects explicitly excluded.
46 First described by Kidson (1988).
47 Thompson and Solomon 2002; Cai et al 2003.
48 For example: Gordon 1986; Mullan 1995.
49 Trenberth and Hoar 1996.
50 Mantua et al 1997.
51 Power et al 1999.
52 Salinger et al 2001.
53 Salinger and Mullan 1999.
54 Mullan et al 2001b.
55 The negative phase is also often called the ‘cold’ phase (and the positive phase, the ‘warm’ phase) because of the sign of the associated decadal anomaly in tropical Pacific sea surface temperature.
56 Parker et al 2007.
57 Salinger and Griffiths 2001.
58 Cubasch et al 2001.
59 Salinger and Griffiths 2001; Salinger 1995.
60 Salinger and Griffiths 2001; Griffiths 2007.
61 McKerchar and Henderson 2003.
62 IPCC 2001a.
63 IPCC 2007a.
64 Trenberth et al 2007 (Chapter 3 of Fourth Assessment, Working Group I); Hennessy et al 2007 (Chapter 11 of Fourth Assessment, Working Group II).
65 Burgess 2005.
66 Diamond 2006.
67 Salinger et al (2005), whose analysis was based primarily on daily pressure gradients rather than winds directly.
68 Griffiths 2007.
69 Lim and Simmonds 2002.
70 Simmonds and Keay 2000.
71 Zhang and Wang 1997.
72 Hannah 2004.
73 Church and White 2006.
74 Tait et al 2002. See also Coastal Hazards and Climate Change. (Ministry for the Environment 2008)