6. Technical Appendix
6.8 Potential Evapotranspiration Deficit: Current Climate
Figures 6.8.1 and 6.8.2 allow a comparison to be made between the station PED calculations of the Phase 1 report and the gridded PED calculations of Phase 2. This is an important check on the integrity of the data sets and calculations. Fig. 6.8.1 displays the time series of annual PED at Lincoln, using the actual station data, whereas Fig. 6.8.2 gives the corresponding nearest gridpoint data. Most of the annual PED accumulation comes during the summer season (true for all other sites as well), with substantial contributions in the spring and autumn seasons for some years. PED calculated from the gridded dataset covers a much shorter period, but the interannual variations and absolute magnitude agree well with the station calculations.
Figure 6.8.1 Accumulated June-May PED (mm) at Lincoln, 1881/82 to 2003/04, using data obtained from the NIWA Climate Database.
Three-month seasonal accumulations are separated according to colour on each bar. The series highlights 1988/89 and 1997/98 as the most severe droughts in the past 122 years. These two seasons were associated with a strong La Niña and El Niño event respectively. (Figure 1 in Phase 1 report, Porteous, 2004).
At this South Island east coast site, large PED accumulations can occur in both El Niño and La Niña years, and occasionally in ENSO-neutral years too such as 1980/81. The longer station time series for Lincoln also suggests there has been an increase in drought severity since the late 1970s, supporting the comments made in the previous section 6.7.2.
Figure 2.1 in the main report shows the accumulated PED for the severe drought in 1997/98. Figure 6.8.3 below shows similar maps for two other drought years. The left panel shows that 1982/83, also a very strong El Niño, did not have as great an effect in Canterbury and Otago as the 1997/98 El Niño. The right panel shows the 1988/89 La Niña had a much greater effect in the South Island than the North; this was the worst drought on record at the Lincoln climate site.
Figure 6.8.2 Accumulated July-June PED (mm) at the gridpoint closest to Lincoln, 1972/73 to 2002/03, extracted from the 0.05° gridded data set.
Figure 6.8.3 Annual PED accumulation (mm) for two severe drought years: 1982/83, a strong El Niño; and 1988/89, the worst drought on record at Lincoln.
In the main report, drought risk is assessed in terms of return period, or probabilities of exceeding a particular level of accumulated PED. Numerical methods are available for estimating the probability distribution from a sample of data (31 years of annual PED, in our case), and hence calculating the return period for specified exceedance levels. We follow the approach of Kim et al. (2003), who describe a non-parametric method for estimating the probability density function (PDF) by using weighted moving averages of the data in a small neighbourhood around the point of estimation, and who apply this method to estimating return periods of drought in Mexico.
Figure 2.3 in the main report shows the probability that in any one year PED will exceed 200mm and 600mm. Figure 6.8.4 below shows the corresponding probability for a PED exceedance of 400mm, derived from statistical analysis of the historical record.
6.9 Potential Evapotranspiration Deficit: Future Scenarios
Because there are four scenarios and two future timeframes (Table 1.1), discussions of future drought in the main report focus mainly on two key sites - Lincoln and Napier. The figures in this section show additional results mapped over the entire country. Figures 6.9.1 and 6.9.2 show the 1-in-20 year return periods for the CSIRO and Hadley scenarios, respectively.
Figure 6.9.1 PED (mm) with a 1 in 20 year return period (5% chance of occurrence in any one year) for the four future scenarios based on the CSIRO model.
Drought risk statistics exhibit strong gradients across the country, and thus it can be difficult to see just how the risk varies with the scenario. Differencing of the various maps can clarify these changing risks. Figures 6.9.3 and 6.9.4 show differences in some drought statistics between the present climate and the projected 2080s 75% climates, for the CSIRO and Hadley scenarios respectively.
The top left panel of Figure 6.9.4 shows the value of a 20-year return period PED increases by more than 150mm over most of the eastern part of New Zealand, for this most extreme of the eight future scenarios. Changes are relatively larger over the eastern half of the North Island.
Figure 6.9.5 shows the future return periods for what are 1-in-20 year PED exceedances under the current climate. The return period changes are generally smaller than those in Figure 3.4 (the analogous map for the 2080s), as expected, except for the northern North Island under the CSIRO scenarios. This is a consequence of the CSIRO model not increasing the westerlies smoothly with time. Most of the drying in the northern North Island occurs in the first 50 years, but most of the drying from Napier southwards (e.g., Figure 3.3) occurs in the second 50 years.
Figure 6.9.5 Future return periods (years) of current climate 1-in-20 year PED events, for four scenarios: CSIRO 2030s 25% and 75% scaling (upper panels) and Hadley 2030s 25% and 75% scaling (lower panels).
6.10 Change in Probability Distribution of Future Drought
Climate change scenarios project the likelihood that changes in mean climate will be accompanied by changes in the frequency of extreme events. This report has demonstrated that changes in mean rainfall and potential evapotranspiration can indeed lead to an increase in severe droughts in the currently drier regions of New Zealand. There is also the much talked about possibility of increased variability in a warmer climate. In this report, we have assumed no change in daily or interannual variation from the current climate for the driving climate parameters of rainfall and potential evapotranspiration. However, this does not rule out the possibility that the response parameter (PED) could become more variable.
Figure 6.10.1 shows the observed distribution of annual (July to June) potential evapotranspiration deficits at three selected sites, Ruakura, Masterton and Blenheim. The data suggest there is more variability at the drier site (Blenheim) with a relatively higher number of extremely dry seasons. That is, the drier Blenheim site has a greater interannual range in the PED drought index.
Figure 6.10.1 Frequency (%) of July to June seasons at various levels of potential evapotranspiration deficit, ranging from the wettest site at Ruakura, to progressively drier distributions at Masterton and Blenheim.
Might we expect a similar increase in variability in future drier climates? Figure 6.10.2 illustrates the scenarios changes at the Lincoln and Napier gridpoints, and suggests that indeed this could happen. Figure 6.10.2 focuses on the Hadley 75% scaling as the most extreme; all other scenarios show the same direction of change but are less pronounced. The two panels show the PED probability density function for the current climate and for future times, and indicate a substantial shift to the right (higher PED) by the 2080s. However, not only is there a change in the mean, but also a change in the variance (a 'broader' distribution). Any increase in variability in the underlying rainfall and PET parameters would enhance future PED variability further.
Figure 6.10.2 Distributional changes in PED at the Lincoln (left) and Napier (right) gridpoints, comparing historical distributions with those projected by Hadley model with IPCC 75%ile scaling.
The vertical axis for the probability density function (PDF) is normalised such that the area under each curve is unity. Thus the curves with a lower peak frequency compensate by having a larger interannual range.
6.11 Sensitivity of Results to Stomatal Resistance Changes
There is a well-documented carbon dioxide 'fertilization effect', whereby increased atmospheric concentration of CO2 increases the growth rates of agricultural crops, improves water efficiency and produces higher yields. This has been demonstrated in controlled environments and in field studies using free-air carbon dioxide enrichment (FACE) facilities. However, there is considerable uncertainty about whether these effects can be applied to long-term crop growth over large areas (see Smith et al., 2005, and references therein). For example, a recent review of agricultural ecosystem responses to elevated CO2 and global climate change concluded "agroecosystem responses will be dominated by those caused directly or indirectly by shifts in climate, associated with altered weather systems, and not by elevated CO2 per se" (Fuhrer, 2003). Essentially, this is because the associated temperature increase reduces the positive CO2-only effect, for a number of reasons.
In the New Zealand context, it has been found that soil moisture content under pasture varied little under different imposed CO2 levels. At the same time, biomass production of New Zealand pasture showed less stimulation to CO2 enrichment than other grassland ecosystems studied (Morgan et al., 2004). Other environmental factors, such as temperature, were not altered in these experiments.
McKenney and Rosenburg (1993) argued that stomatal resistance and plant leaf area are both expected to increase with the higher levels of atmospheric CO2 associated with climate change. Their work showed that responses in potential evapotranspiration (PET) to changes in these two vegetation characteristics are similar in magnitude but opposite in effect - increased stomatal resistance acts to reduce PET, whereas greater leaf area acts to increase it. Bunce (2004) noted that, for doubled CO2, stomatal conductance (inverse of resistance) decreased by anywhere from less than 15% in some crop species to more than 50% in others. However, he concluded that this would translate into less than 10% reduction in evapotranspiration, partly because of increases in temperature and decreases in humidity in the air around crop leaves.
The compensating effects of changes in stomatal resistance and leaf area on PET justify our default assumption of no direct CO2 effect on potential evapotranspiration in this study. However, because there is potentially such a large effect if stomatal resistance effects dominate over leaf area increases, we have carried out a short sensitivity study. We have taken the two extreme 2080s scenarios (CSIRO model with 25% scaling, and Hadley model with 75% scaling) and, after making the scenario PET adjustments as before, imposed a 5% reduction in PET. Note that this is a reduction on the total PET, not just the increment due to increased temperature or stronger winds in a future climate.
Figure 6.11.1 summarises the result in terms of change in current climate 1-in-20 year PED. The left panels are reproduced from the main report Figure 3.4, and show the future return period of events that currently occur on average once in 20 years. Under the default assumption of no CO2 effect, there are substantial reductions in return period, as discussed in the main report.
Figure 6.11.1 Future return periods (years) of current climate 1-in-20 year PED events, for four scenarios: CSIRO 2080s 25% scaling (top) and Hadley 2080s 75% scaling (bottom), and for standard scenario (left) and with ad hoc 5% PET reduction to represent increased stomatal resistance under CO2 enrichment (right).
The right panels show the future return periods after the 5% reduction in potential evapotranspiration. For the more benign CSIRO 25% scenario, reductions in return period are restricted to eastern margins of both Islands. Even here, the current 1-in-20 year event becomes no more frequent than about 1-in-15 years, in general. However, for the drier Hadley 75% scenario, there are still very substantial reductions in return period throughout eastern parts of New Zealand, from Northland to south Canterbury. Regions where there is a four-fold or more reduction in return period still occur on the eastern margins, although the areas affected contract.
6.12 Overseas studies of changes in drought under global warming
There have been a large number of international studies of changes in water availability under global warming. We have been unable to find a study that is truly comparable to ours in terms of using PED as a quantitative indicator of changes in drought risk, and calculating return period changes for drought. Several overseas studies are described briefly below, for the purpose of illustrating that comparable changes in water resources (runoff or some other measure) have been found elsewhere.
Döll et al. (1999) calculated global runoff for large drainage basins at 2075, based on one IPCC emissions scenario (IS92a) and two climate model patterns, and assuming climate variability remained constant. The 1-in-10 year dry runoff was computed to decrease by more than 50% in 2.5% and 6.7% of global land area, for the two models. At the same time, there was also an increase of more than 50% in 49.1% and 21.8% of land area, respectively, for the two models, demonstrating that the hydrologic situation became more extreme in many parts of the world.
A recent comprehensive integrated assessment study for the United States (Thomson et al., 2005) considered three model patterns of regional climate change, scaled to match global temperature increases of 1°C and 2.5°C (so there was no specific future date defined), and again assumed no change in interannual variability from the baseline period. The largest changes in water resource were noted in the current semi-arid regions of the western U.S., where changes in water yield, runoff and evapotranspiration exceeding ±50% of baseline levels were identified for the 2.5°C global warming case.
An earlier study by Rind et al. (1990) found even more dramatic changes in the United States from one climate model at about the time of CO2 doubling. Two drought indices were calculated: the Palmer Drought Severity Index (see appendix, section 6.2.3) and an index measuring excess potential evapotranspiration over precipitation. Both drought indices showed increased likelihood of drought as the climate warmed. In the latter half of the 21st century, when global temperature increases exceeded 4°C, the indices indicated that the "5% drought" (analogous to our 1-in-20 year PED) occurred the majority of the time (ie, a tenfold reduction in return period), as an average over the contiguous United States.
Kothavala (1999) used the Palmer Drought Severity Index to quantify changes in drought duration and severity over eastern Australia. A single climate model was used, and 30 years of data around the time of CO2 doubling were analysed. There was a three-fold increase in the number of months classified as severe or extreme drought in eastern Australia.
There is an intriguing parallel in return period changes between this report on droughts, and other studies on high intensity rainfall. We have found a reduction in return period by a factor of two (for CSIRO 25%) to four or more (for Hadley 75%) by the 2080s. Whetton et al. (1996) suggested that by 2070 there would be "no change through to a fourfold reduction in the return period" of daily heavy rainfall events in Australia and New Zealand, based on an equilibrium model study of CO2 doubling. Hennessy et al. (1997), in a similar equilibrium study with two models, confirmed the general finding of a shift in precipitation type to more intense convective events at many locations in middle and low latitudes. For a given intensity of daily precipitation, they found the average return period to shorten by a factor of 2 to 5 across Europe, USA and Australia.
6.13 Appendix References
Adams, P.D.; Horridge, M.; Masden, J.R.; Wittwer, G. (2002). Drought, Regions and the Australian Economy between 2001-02 and 2004-05, Australian Bulletin of Labour, 28 (4): 233-249.
Bunce, J.A. (2004). Carbon dioxide effects on stomatal responses to the environment and water use by crops under field conditions. Oecologia, 140: 1-10.
Byun, H.-R.; Wilhite, D. (1999). Objective quantification of drought severity and duration. Journal of Climate, 12: 2747-2756.
Cubasch, U.; Meehl, G.A.; Boer, G.J.; Stouffer, R.J.; Dix, M.; Noda, A.; Senior, C.A.; Raper, S.; Yap. K.S. (2001). Projections of future climate change. In: JT Houghton, Y Ding, DJ Griggs, et al (eds) Climate Change 2001: The scientific basis. Contribution of Working Group I to the Third Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge and New York: Cambridge University Press, 525-82.
Döll, P.; Kaspar, F.; Alcamo, J. (1999). Computation of global water availability and water use at the scale of large drainage Basins. Mathematische Geologie, 4: 111-118.
Fuhrer, J. (2003). Agroecosystem responses to combinations of elevated CO2, ozone, and global climate change. Agriculture, Ecosystems and Environment, 97: 1-20.
Gardiner, P. (2001). An assessment of the Asian crisis and droughts: 1991-2000. A New Zealand Institute of Economic Research report to the Ministry of Agriculture and Forestry, May 2001.
Gordon, N.D. (1986). The Southern Oscillation and New Zealand weather. Monthly Weather Review,114: 371-387.
Hennessy, K.J.; Gregory, J.M.; Mitchell, J.F.B. (1997). Changes in daily precipitation under enhanced greenhouse conditions. Climate Dynamics, 13: 667-680.
Kim, T.-W.; Valdes, J.B.; Yoo, C. (2003). Nonparametric approach for estimating return periods of droughts in arid regions. ASCE Journal of Hydrologic Engineering, 8(5): 237-246.
Kothavala, Z. (1999). The duration and severity of drought over eastern Australia simulated by a coupled ocean-atmosphere GCM with a transient increase in CO2. Environmental Modelling and Software, 14: 243-252.
Laughlin, G.; Clark, A. (2000). Drought science and drought policy in Australia: a risk management perspective. Expert Group meeting on Early Warning Systems for Drought Preparedness and Drought Management, Lisbon, September 2000. Bureau of Rural Sciences, Dept of Agriculture, Fisheries and Forestry-Australia, Kingston, ACT.
McKenney, M.S.; Rosenberg, N.J. (1993). Sensitivity of some potential evapotranspiration estimation methods to climate change. Agricultural and Forest Meteorology, 64: 81-110.
Mantua, N.J.; Hare, S.R.; Zhang, Y. (1997). A Pacific interdecadal climate oscillation with impacts on salmon production. Bulletin of the American Meteorological Society,78: 1069-1079.
Morgan, J.A.; Pataki, D.E.; Körner, C.; Clark, H.; et al. (2004). Water relations in grassland and desert ecosystems exposed to elevated atmospheric CO2. Oecologia, 140: 11-25.
Mullan, A,B. (1995). On the linearity and stability of Southern Oscillation-climate relationships for New Zealand. International J. Climatology,15: 1365-1386.
Palmer, W.C. (1965) Meteorological Drought. US Weather Bureau Technical Paper 45, 1-58.
Rind, D.; Goldberg,R.; Hansen, J.; Rosenzweig, C.; Ruedy, R. (1990). Potential evapotranspiration and the likelihood of future drought. Journal of Geophysical Research, 95: 9983-10004.
Salinger, M.J.; Mullan, A.B. (1999). New Zealand climate: temperature and precipitation variations and their links with atmospheric circulation 1930-94. International Journal of Climatology,19: 1049-1071.
Salinger, M.J.; Renwick, J.A.; Mullan, A.B. (2001). Interdecadal Pacific Oscillation and South Pacific climate. International Journal of Climatology,21: 1705-1721.
Shafer, B.A.; Desman, L.E. (1982). Development of a surface water supply index (SWSI) to assess the severity of drought conditions in snow pack runoff areas. Proc. Western Snow Conference, 164-175.
Smith, S.J.; Thomson, A.M.; Rosenberg, N.J.; Izaurralde, R.C.; Brown, R.A.; Wigley, T.M.L. (2005). Climate change impacts for the conterminous USA: An integrated assessment. Part I: Scenarios and context. Climatic Change, 69: 7-25.
Thomson, A.M.; Brown, R.A.; Rosenberg, N.J.; Srinivasan, R.; Izaurralde, R.C. (2005). Climate change impacts for the conterminous USA: An integrated assessment. Part 4: Water resources. Climatic Change, 69: 67-88.
Trenberth, K.R. (1976). Fluctuations and trends in indices of the Southern Hemisphere circulation. Quarterly Journal of the Royal Meteorological Society, 102: 65-107.
USDA (2000). Preparing for Drought in the 21st Century - Report of the National Drought Policy Commission, May 2000. United States Department of Agriculture.
Western Drought Coordination Council. (1998). How to reduce drought risk. Report of the Preparedness and Mitigation Working Group. National Drought Mitigation Centre, University of Nebraska-Lincoln, http://enso.unl.edu/wdcc/.
Whetton P.; Mullan, A.B.; Pittock, A.B. (1996). Climate change scenarios for Australia and New Zealand. In: Greenhouse 94, Coping with Climate Change, Eds: W.J. Bouma, G.I. Pearman and M. Manning. CSIRO/DAR, Melbourne, pp. 145-168.
