A3.2.3 Agreement on projected precipitation change
Agreement between the models on regional precipitation changes may not be that apparent from Tables 2.4 and 2.5, but actually does occur in some seasons and parts of New Zealand, as illustrated by Figure A3.8, at least if we restrict ourselves to considering just the sign of the precipitation change.
In Figure A3.8, the darkest red colour means 11 or 12 models predict decreases in the seasonal precipitation under the A1B scenario. The darkest blue colour means just 0, 1 or 2 models have decreases (or alternatively, 10–12 models agree on increased precipitation). It is clear that there is strong agreement that there will be wetter conditions in the western South Island, and simultaneously drier conditions in the north and east of the North Island, in the winter season (and also in spring, not shown). For the summer season (and also autumn, not shown), there is lower agreement between the 12 models on the sign of the precipitation change, but there is a tendency for wetter conditions in the eastern North Island.
Figure A3.8: Level of agreement between models on summer (left) and winter (right) precipitation changes to 2090. Each map gives the number of models, out of 12, that indicate decreased precipitation in the season shown.
This seasonality of projected precipitation changes can be understood to some degree by examining the pressure changes. Figure A3.9 illustrates changes at 2090 in the summer and winter pressure fields, averaged over all 12 climate models used in the downscaling. Under a global warming scenario, precipitation increases in the tropics, and the overturning Hadley circulation intensifies and expands to higher latitudes (Meehl et al 2007; Yin 2005). This results in higher pressures in the descending branch of the Hadley cell, which in the present climate is located just north of New Zealand in the annual mean, but moves down over the North Island in summer. The circulation changes simulated for 2090 suggest that in the summer season (Figure A3.9, left panel) the high pressure belt is sufficiently far south that there is more of an easterly component to the wind flow over the North Island. The southward movement of the anticyclone belt does of course vary from model to model, but is consistent with increased rainfall in the east of the North Island.
The situation is different in the winter season, when the Hadley cells move northward with the sun. Even with a future southwards expansion, the axis of highest pressure is still to the north of New Zealand. There is, therefore, an increase in the north–south pressure gradient across the country, which drives stronger or more persistent westerly winds. This circulation response is very consistent across the global climate models, and is the reason for the strong agreement on increased rainfall in the west of the South Island and decreases in the east of the North Island in the winter season.
Note: 1990–2090 changes are shown for one site within each of 15 regional council areas (the grid point is co-located with the first named city for each region, as listed in Table 2.4). Thus, the Northland (Nth) site is Kaitaia, the Canterbury (Can) site is Christchurch, etc. Vertical coloured bars show the range over all 12 models, and stars the 12 individual model values.
Figure A3.10 indicates the model spread in annual precipitation for a selection of sites across all 15 regional council regions. If the lowest and highest models are again considered as outliers, the projected range from the remaining 10 models is (of course) reduced, but not as dramatically as it is for temperature. In most regions, the projected rainfall still covers both increases and decreases. Nevertheless, four of the regional sites in Figure A3.10 show full agreement among models on the sign of the rainfall change, in the annual mean, after removing the low and high outliers. Thus, the Northland and Auckland sites have decreased annual rainfall indicated by all (10) models, and the West Coast and Southland sites have increased annual rainfall from all models.
A3.3 Changes in Extreme Precipitation
The Fourth Assessment Report, relying on both observational and modelling studies, declared that more intense precipitation events were ‘very likely’ to increase over many areas of the globe (IPCC 2007: table SPM.1). This might not necessarily apply to a region as small as that of New Zealand, particularly to those parts where the annual mean rainfall is projected to decrease. However, a warmer atmosphere can hold more moisture (about 8% more for every 1°C increase in air temperature), and so there is potential for heavier extreme rainfalls.
An alternative way of viewing these systematic increases in average rainfall intensity is to say that a reduction in the return period of heavy rainfall events is expected. Estimates of projected changes in return periods for New Zealand were provided by Whetton et al (1996), who suggested that: by 2030, there would be ‘no change through to a halving of the return period of heavy rainfall events’, and by 2070, ‘no change through to a fourfold reduction of the return period’. This statement was based on analysing daily precipitation time series from a regional climate model, driven by the CSIRO equilibrium General Circulation Model. Very recent results from the NIWA regional climate model are described in chapter 2.
Further useful information on how daily precipitation extremes could change for New Zealand is available in the paper by Semenov and Bengtsson (2002). These authors present global maps of changes in total rainfall and in the 95th percentile daily value. They also analyse changes in the rainfall distribution. The ‘normal’ distribution (Figure Box 2.1), which is symmetrical about the mean, is not appropriate for rainfall, which is commonly represented by the so-called ‘gamma’ distribution (see Figure A3.12). The parameters of the gamma distribution are known as the ‘shape factor’ (‘alpha’, α) and the scale factor (‘beta’, β). For alpha less than 1, as is always the situation for New Zealand, the higher the rainfall amount, the less frequently it occurs. If alpha increases above 1, then there is a peak mode or most likely rainfall amount. As alpha increases further, the gamma distribution tends to the same shape as the normal distribution. The mean rainfall averaged over days when it rains (the so-called ‘rainfall intensity’) is simply the product of these two factors (ie, αβ). The gamma distribution applies only to rain days. The other relevant factor is the likelihood of it raining at all (probability of a wet day, Pw). The change in mean rainfall (section 2.2.2) is, therefore, the change in the triple product Pwαβ.
Semenov and Bengtsson (2002) mapped the changes in alpha and beta, as simulated by their model (which corresponds to the MPI model used by Mullan et al 2001). Outside the tropical and subtropical oceans, alpha generally decreases (by up to 10%), whereas beta generally increases (by up to 40%). Figures A3.11 and A3.12 illustrate how this projection might be applied at a particular site. Thirty years of observed daily winter rainfall data at Auckland (Whenuapai) are used to compute the distributional parameters alpha (= 0.735) and beta (= 10.19). By systematically varying alpha and beta from their observed values, we can generate a surface that represents how the daily winter rainfall might change by 2100. In this case, Figure A3.11 shows how the 95th percentile winter daily rainfall amount for the present climate (= 25.0 mm) could change. Over the (shaded) range of alpha and beta parameters suggested by Semenov and Bengtsson (2002), Figure A3.11 shows that this winter extreme rainfall could vary from about 6% less than present (10% decrease in alpha, with no change in beta) to 40% more than present (no change in alpha, with 40% increase in beta). Either of these changes could be made consistent with the scenario of mean rainfall change by adjusting the number of wet days. A similar study of rainfall distributional changes (Voss et al 2002) found virtually no change in the alpha parameter, but consistent increases in beta, when averaged over large areas of the globe.
Note: Shaded region shows the range of changes by 2100 in alpha and beta projected for the New Zealand region by Semenov and Bengtsson (2002). Hatched shading indicates increases in the 95th percentile value, and solid shading indicates decreases.
Figure A3.12 plots the gamma distribution for this 2100 extreme case of no change in α but a 40% increase in β. The figure shows that a daily winter rainfall amount of about 100 mm, which was exceeded only once in the 30 years of observed record, would become at least 10 times more likely under this particular scenario.
Note: Lines show the probability (on a logarithmic scale) of a particular daily rainfall amount for present winter daily rainfall at Auckland (solid line), and for a changed climate with a 40% increase in beta (dashed line). An increase of one unit on the vertical axis corresponds to a 10-fold increase in probability of occurrence.
Figure A3.13 shows the extreme rainfall data translated into return periods. The data were generated by random sampling from the gamma distribution, firstly using the α,β parameters fitted to winter observations at Auckland, and repeated with a 40% increase in β. Return periods were estimated by fitting an EV1 (Extreme Value type 1) distribution to the highest daily rainfall for each of 30 winters. The result (Figure A3.13) indicates that a rainfall amount with a return period of 50 years under the present climate has a return period of about 7 years at 2100 in this worst case. This sevenfold reduction in return period is broadly consistent with the earlier estimates of Whetton et al (1996).
Note: Comparing the present climate (blue line, below) with the extreme change at 2100 (red line, above), as inferred from the gamma distributions changes of Semenov and Bengtsson (2002). The dotted lines shows how the 50-year return period under the present climate, associated with a rainfall amount of about 72 mm, changes to a return period of about 7 years for this worst case scenario.
IPCC. 2007. Summary for Policymakers. In: Solomon S, Qin D, Manning M, Chen Z, Marquis M, Averyt KB, Tignor M, Miller HL (eds.) Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press: Cambridge, UK and New York. http://www.ipcc.ch/pdf/assessment-report/ar4/wg1/ar4-wg1-spm.pdf (2 April 2008).
Ministry for the Environment. 2004. Climate Change Effects and Impacts Assessment: A Guidance Manual for Local Government in New Zealand. Wratt D, Mullan B, Salinger J, Allen S, Morgan T, Kenny G with MfE. Ministry for the Environment, Wellington, 153 p. http://www.mfe.govt.nz/publications/climate/effects-impacts-may04/index.html (2 April 2008).
Mullan AB, Wratt DS, Renwick JA. 2001. Transient model scenarios of climate changes for New Zealand. Weather and Climate 21: 3–34.
Semenov VA, Bengtsson L. 2002. Secular trends in daily precipitation characteristics: greenhouse gas simulation with a coupled AOGCM. Climate Dynamics 19: 123–40.
Tait A, Henderson R, Turner R, Zheng X. 2006. Thin plate smoothing spline interpolation of daily rainfall for New Zealand using a climatological rainfall surface. International Journal of Climatology 26: 2097–2115.
Trenberth KE. 1976. Fluctuations and trends in indices of the Southern Hemisphere circulation. Quarterly Journal of the Royal Meteorological Society 102: 65–75.
Voss R, May W, Roeckner E. 2002. Enhanced resolution modelling study on anthropogenic climate change: changes in extremes of the hydrological cycle. International Journal of Climatology 22: 755–777.
Whetton P, Mullan AB, Pittock AB. 1996. Climate change scenarios for Australia and New Zealand. In: Bouma, WJ, Pearman GI, Manning M (eds). Greenhouse ’94: Coping with climate change. CSIRO/DAR: Melbourne. 145–68.
Yin JH. 2005. A consistent poleward shift in the storm tracks in simulations of the 21st century climate. Geophysical Research. Letters 32 L18701, doi:10.1029/2005GL023684.