The best estimates of New Zealand temperatures are for an expected increase of about 1°C by 2040, and 2°C by 2090. However owing to the different emission scenarios and model climate sensitivities, the projections of future warming cover a wide range: 0.2–2.0°C by 2040 and 0.7–5.1°C by 2090.
Projected rainfall and wind patterns show a more marked seasonality than was evident in models used in the IPCC Third Assessment, 2001. Westerlies are projected to increase in winter and spring, along with more rainfall in the west of both the North and the South Island and drier conditions in the east and north. Conversely, the models suggest a decreased frequency of westerly conditions in summer and autumn, with drier conditions in the west of the North Island and possible rainfall increases in Gisborne and Hawke’s Bay.
Other changes expected are: decreased frost risk, increased frequency of high temperatures, increased frequency of extreme daily rainfalls, decreased seasonal snow cover, and a possible increase in strong winds.
Temperature rise is expected to speed up. The rate of temperature increase from these projections is expected to be higher than a linear extrapolation of the historical New Zealand temperature record for the 20th century.
Projected New Zealand climate changes are based on results from 12 global climate models, with additional information on extremes and other physical climate elements provided from a regional climate model.
This chapter outlines the changes in New Zealand’s climate that are expected to result from global human-induced emissions of greenhouse gases and aerosols. Most of the projections are based on results from General Circulation Model simulations prepared for the IPCC Fourth Assessment, 20079. Model changes are statistically downscaled10 to provide spatial detail over New Zealand.
Human-induced climate change should be considered within the context of the natural variability of the climate system, and this aspect is discussed in chapter 3. Chapter 5 provides advice on typical changes that local government should take into account when assessing risk. Appendix 2 provides technical details on the General Circulation Models used and the scaling applied to generate future projected ranges that appear in the tables of this chapter. Appendix 3 gives further information on topics such as the downscaling approach, the level of agreement (or otherwise) between the model projections, and changes in extreme precipitation.
Projected values of a particular climate element for use in an impacts assessment are available from Tables 5.1 and 5.2 of chapter 5. Those tables provide guidance on values for use in scenario analyses, and refer users back to particular parts of the current chapter.
Climate is often thought of as only the long-term averages of weather elements, but it actually also includes the range of likely values and the occurrence of extremes. Indeed, it is recognised that the largest impacts of climate change will probably be felt through changes in these extremes. Changes in extremes cannot be reliably derived directly from General Circulation Model outputs, owing to the coarse spatial resolution of the models. However, as illustrated in Box 2.1, small changes in average values (for example, in average annual temperature) can result in large changes in the frequency with which climate extremes occur (for example, for frosts and very high temperatures, and similarly for heavy rainfall, floods and drought). Thus, projections of changes in the average value of a climate element can also help us estimate how the frequency of extremes might change, although this might require additional assumptions about the shape of the distribution. Information on the distribution of daily extremes for New Zealand’s future climate can now be supplemented with simulations by the NIWA regional climate model. This model is currently run at a 30-km grid spacing over New Zealand, which is an improvement in resolution over the typical global model (100- to 300-km spacing). Computational constraints mean that the regional model can presently be run only from a single global model (ukmo_hadcm3, see Appendix 2) and for a limited number of emissions scenarios.
Predictions of future climate depend on projections of future concentrations of greenhouse gases and aerosols. These depend on projections of emissions which, in turn, depend on changes in population, economic growth, technology, energy availability, and national and international policies. The IPCC developed 40 different future emissions pathways or scenarios11 as a basis for projecting future climate changes. These SRES scenarios formed the basis of much of the climate projection work done for the IPCC’s Third and Fourth Assessments.
Figure 2.1 indicates a range of possible future global temperatures, and reflects the range of plausible emissions scenarios and the range of General Circulation Model predictions for given scenarios. The scenario labelled ‘A1B’, which gives an intermediate level of warming by the end of the century, has more General Circulation Model output data available than any other scenario, and is the scenario used to derive most of the projections discussed in this manual. To cover the full spread across all the IPCC emission scenarios, New Zealand projections from the A1B scenario were rescaled using the known differences on the global scale between the A1B and other scenarios (this gives the vertical grey bars in Figure 2.1).
The IPCC made subtle changes between the Third and Fourth Assessment Reports in the way it expressed the climate projections. The Third Assessment Report stated12: “The globally averaged surface temperature is projected to increase by 1.4 to 5.8°C over the period 1990 to 2100.” These results are for the full range of 40 SRES scenarios, based on a number of climate models. In the Fourth Assessment Report, the projections were expressed as changes between 1980–1999 and 2080–2099, and projections were given separately for six illustrative scenarios (see Appendix 2) that spanned the range of all 40 SRES scenarios. For each of the six scenarios, a best estimate was provided, as well as the likely range. The full range in global temperature increase over the six illustrative scenarios used in the Fourth Assessment Report was 1.1–6.4°C.
Note: Solid lines are multi-model global averages of surface warming (relative to 1980–1999) for scenarios B1, A1B and A2, shown as continuations of the 20th century simulations. The coloured shading denotes the ±1 standard deviation range of individual model annual averages. The grey bars at right indicate the best estimate (solid horizontal line within each grey bar) and the ‘likely range’ for all six SRES illustrative scenarios. Source: IPCC 2007a (figure SPM.5).
The SRES scenarios cover the key greenhouse gases (carbon dioxide, methane, nitrous oxide and CFCs) and the sulphate aerosols. They do not account for explicit climate change policy actions to reduce greenhouse gas emissions, such as might be taken under the Kyoto Protocol. However, some scenarios assume a reduction in world population after a mid-century peak, and the rapid and widespread introduction of clean and resource-efficient technologies. The SRES scenarios also do not account for any unexpected climate ‘surprises’, such as increased methane emissions from permafrost melting or undersea methane clathrates.13
Local impacts of climate are likely to depend more on changes in the frequency of extreme events (such as heavy rainfall, drought or very high temperatures) than on changes in average conditions. However, these two aspects of climate – averages and extremes – are closely connected. The figure below is a simplified illustration of how a small change in average conditions can lead to a large change in the frequency with which extremes occur. (In the real world, the curves will not be so smooth or symmetrical.)
Suppose the dashed line represents the current frequency of hourly temperatures over a year, and the heavy line a possible future distribution. The shaded area under a curve represents how often temperatures occur above a particular threshold (orange, red) or below a threshold (blue).
Suppose that in this case the cold area (blue) represents hourly temperatures below freezing, the orange area represents temperatures above 30°C, the red area represents temperatures above 35°C, and the change in mean (average) temperature shown by the arrow is 3°C. So, in this case, an apparently modest change in average temperature is accompanied by a total cessation of frosts, occurrence of higher temperatures than hitherto experienced (red area) and a substantial increased frequency of temperatures above 30°C.
This relationship between averages and extremes has important implications for adaptation (as noted by Warrick 2002). For example, a particular farming operation might already be well adapted to temperatures ranging between –2°C and 30°C, and able to cope occasionally with temperatures between 30°C and 35°C. While the changes in the mean temperature lie well within the ‘autonomous’ (easily coped with) adaptation region, changes into the red area are outside the ‘coping’ region and damage occurs. This example is overly simplistic (in the real world the shape, width and height of the curve might also change), but it serves to illustrate the importance of extremes.
Figure Box 2.1: Effect of climate change on average and extreme temperatures. The horizontal axis represents temperature (Source: figure 4.1–IPCC Synthesis Report, IPCC 2001b). Note that the horizontal axis is not to scale, and the diagram is illustrative only.
To identify likely future climate changes across New Zealand, projected changes from General Circulation Models are statistically downscaled. This method is used to translate the coarse-scale information available from General Circulation Models to the local scale. Historical observations are used to develop regression equations that relate local climate fluctuations to changes at the larger scale. These historical observations are then replaced in the regression equations by the modelled changes to produce the fine-scale projections (see Appendix 3 for more information). Downscaled changes were prepared for a 0.05 degrees latitude and longitude grid (approximately 5 km by 4 km) covering New Zealand.
The New Zealand downscaled projections follow the approach of the Fourth Assessment Report. That is, changes are relative to 1980–1999, which we abbreviate as ‘1990’ for convenience. Changes are calculated for two future periods: 2030–2049 (‘2040’) and 2080–2099 (‘2090’). Thus, the New Zealand projections are for changes over time periods of 50 and 100 years from the baseline climate (centred on 1990). Figure 2.2 provides a schematic for the time horizons of the climate projections. Also shown in Figure 2.2, for reference, are the averaging periods referred to as the ‘2030s’ (2020–2049 average) and the ‘2080s’ (2070–2099 average) used in the previous edition of this Guidance Manual.
Note: Curve (blue line) shows a smoothly varying climate parameter, such as temperature or sea level, relative to a base level defined as the average over 1980–1999 (first horizontal red line; ‘1990’). Future 20-year averages are indicated by the other red lines at 2040 (2030–2049 average) and 2090 (2080–2099). Dotted orange lines show projection horizons used in the previous Guidance Manual (Ministry for the Environment 2004), identified as the ‘2030s’ (2020–2049 average) and the ‘2080s’ (2070–2099 average).
Councils may also be interested in projections for other decades during the 21st century. Initial projections for these non-tabulated decades can be obtained by interpolating linearly between the values for 1990, 2040 and 2090. For example, a projection for 2050 (relative to 1990) would be the change at 2040 plus 20% of the change between 2040 and 2090. Different start dates (eg, council data more recent than 1999) could also be accommodated by linear interpolation, although it is important to use a time average rather than an individual year.
Downscaling is applied to the projections obtained from 12 General Circulation Models when emissions follow the A1B middle-of-the-road emissions scenario (Figure 2.1). A range of possible values for each climate variable (temperature, rainfall, etc) is provided. The range for each variable reflects not only the range of greenhouse gas futures represented by the six SRES illustrative scenarios, but also the range of climate model predictions for individual emission scenarios. The other five SRES emissions scenarios are catered for by re-scaling the A1B results for New Zealand according to the ratio of global temperature increases, as documented in the IPCC Fourth Assessment Report (see Appendix 2 for details).
Like the IPCC, we are unable to indicate whether any one emission scenario is more likely than another, but do provide the average across all models and all emission scenarios. The extreme ends of the ranges may be slightly less likely than the central values, since they generally result from the one climate model that gives the most extreme projection, rather than reflecting the consensus from a number of models. Eliminating the most extreme models as outliers causes little change to the average from the remaining models, but can, on occasion, greatly reduce the range of the projected changes (see Appendix 3).
Table 2.1 summarises the main features of these New Zealand climate projections. More detail on the changes is given in the figures and tables later in this chapter. Quantitative estimates of the changes in parameters relevant to local government functions and services, and advice on how to construct relevant scenarios to estimate the importance of those changes, are given in chapter 5 (specifically Tables 5.1 and 5.2).
Each estimate in Table 2.1 is the best current scientific estimate of the direction and magnitude of change a given climate variable could undergo. The degree of confidence placed by NIWA scientists on the projections is indicated by the number of stars in brackets:
**** Very confident, at least 9 out of 10 chance of being correct. Very confident means that it is considered very unlikely that the estimate will be substantially revised as scientific knowledge progresses.
** Moderate confidence, which means that the estimate is more likely than not to be correct in terms of indicated direction and approximate magnitude of the change.
* Low confidence, but the best estimate possible at present from the most recent information. Such estimates could be revised considerably in the future.
Hence, a higher degree of caution should be employed where investment decisions are based on the low-confidence estimates.
The following sections, along with material in Appendix 3, provide more detail on the projected changes summarised in Table 2.1.
Downscaled projections of the changes in mean temperature14 over New Zealand are shown in Table 2.2 (for 2040) and Table 2.3 (for 2090), and in Figure 2.3 (changes in annual average temperature) and Figures 2.4–2.5 (seasonal changes).
The tables indicate the range not only across the models analysed, but also across the various emissions scenarios. The A1B projections were rescaled by the quoted IPCC global temperature changes to cover the other five illustrative scenarios. The values given in Tables 2.2 and 2.3 are averages over all grid points within each regional council region.
The figures depict the pattern of temperature change as an average over 12 climate models for the A1B emissions scenario15. There is considerable pattern variation among the climate models, so we also present changes in the annual average for each of the 12 models separately (Figures A3.2 and A3.3 in Appendix 3).
Averaging over all models and all six illustrative emissions scenarios gives a New Zealand-average warming of 0.2–2.0°C by 2040 and 0.7–5.1°C by 2090. For just the A1B scenario alone, the projected warming is 0.3–1.4°C by 2040 and 1.1–3.4°C by 2090, with a 12-model average (or ‘best estimate’) of 0.9°C and 2.1°C for 2040 and 2090, respectively. For comparison, the IPCC quotes a best estimate of 2.8°C for the global temperature increase by 2090 under the A1B scenario, with a likely range of 1.7–4.4°C. The projected New Zealand temperature changes are in all cases smaller than the globally averaged changes for the corresponding SRES scenarios (see also Table A2.1 in Appendix 2).
The pattern of warming in the annual average is fairly uniform over the country, although slightly greater over the North Island than the South. Also, the warming accelerates with time under this emissions scenario: ie, the 2090 warming is more than twice the 2040 warming. Figures 2.4 and 2.5 map projected seasonal mean changes at 2040 and 2090 for the A1B scenario. In the summer and autumn seasons, the North Island and northwest of the South Island show the greatest warming, whereas in the winter season the South Island has the greatest warming. Spring shows the least warming of all seasons. Further discussion of agreement between the various models can be found in Appendix 3 (section A3.3 and Figures A3.2 and A3.3).
Table 2.2: Projected changes in seasonal and annual mean temperature (in °C) from 1990 to 2040, by regional council area. The average change, and the lower and upper limits (in brackets), over the six illustrative scenarios are given.
Table 2.3: Projected changes in seasonal and annual mean temperature (in °C) from 1990 to 2090, by regional council area. The average change, and the lower and upper limits (in brackets), over the six illustrative scenarios are given.
Downscaled rainfall projections are shown in Figure 2.3 (changes in annual average) and Figures 2.6 and 2.7 (seasonal changes), and in Table 2.4 (for 2040) and Table 2.5 (for 2090). Maps of the changes in annual average rainfall given by individual models are presented in Appendix 3.
There are often systematic variations in the projected rainfall within regional council regions (for example, wetter in the west and drier in the east for Canterbury). Thus, it is not very useful to tabulate averages for each region as was done for temperature. Instead, rainfall projections have been tabulated for specific places. Councils may need to carefully examine these regional gradients in rainfall changes, when considering issues related to river levels. For example, in coastal Canterbury, rainfall is projected to decrease, but large alpine-fed rivers could have increased flows because of greater rainfall in the headwaters.
Tables 2.4 and 2.5 give the estimated range in precipitation change over the six illustrative SRES scenarios, for selected sites within each region. The average change over all 12 models and six scenarios is also given. Two sites per region (for Canterbury, three sites) are included in the tables whenever there is a marked spatial variation across a region. Figure 2.2 maps the projected annual mean precipitation change for the A1B scenarios for the period from 1990 to 2040 and 2090. Figures 2.6 and 2.7 show the seasonal projections, again as an average over the 12 models for just the A1B scenario. As might be expected, there is much more spatial structure in the rainfall changes than in the temperature changes, and also a larger spread between models. For most sites, rainfall can show either an increase or a decrease, depending on which model is chosen. Appendix 3 gives further discussion on the level of model agreement.
The annual average rainfall change has a pattern of increases in the west (up to 5% by 2040 and 10% by 2090) and decreases in the east and north (exceeding 5% in places by 2090). Figures 2.6 and 2.7 show that this annual pattern of being wetter in the west and drier in the east is driven by that pattern occurring in the winter and spring seasons. In summer and autumn, the pattern is quite different. Indeed, for the North Island in particular, the pattern is reversed, with it being drier in the west and wetter in the east (although the percentage changes are smaller than for the winter and spring seasons, and winter has the largest total precipitation). These distinct seasonal differences are a new result, not apparent in the smaller sample of models used in the previous edition of this Manual. There is still a lot of variability between models, although some regions show more agreement between models than others on the sign of the projected precipitation change (see Appendix 3 for further discussion).
Figure 2.3: Projected changes in annual mean temperature (in °C) and in annual mean rainfall (in %), relative to 1990: average over 12 climate models for A1B emission scenario. Note the different temperature scales for 2040 and 2090.
9 Meehl et al 2007.
10 Mullan et al 2001a.
11 IPCC Special Report on Emissions Scenarios: Nakicenovic and Swart 2000. See also Appendix 1.
12 IPCC 2001a.
13 Clathrates, also called ‘gas hydrates’, are crystalline solids that look like ice, and that occur when water molecules form a cage-like structure around smaller ‘guest molecules’ such as methane. Clathrates occur naturally in cold environments, such as the deep ocean.
14 ‘Mean’ temperature is the average of daily minimum and maximum temperatures. Simulations by NIWA’s regional climate model suggest that minimum and maximum temperatures both increase at very nearly the same rate, and so no distinction between them is made for New Zealand temperatures.
15 Note that in this 2008 edition of the Guidance Manual, the maps are specific to a single SRES emissions scenario (A1B), unlike in the previous edition where changes were scaled to cover all SRES scenarios. However, the Tables (2.2 and 2.3) do incorporate the full range of projected changes.