• About this site|
• Site index|
• Contact us|
• Jobs

Changes in drought risk with climate change

Executive Summary

1. Introduction

2. Drought Risk under Current Climate

3. Drought Risk under Scenarios of Climate Change

4. Other Issues

5. References

6. Technical Appendix

1. Introduction

The purpose of this study is to provide quantitative measures of likely future changes in drought risk in New Zealand under climate change. This report is aimed at central and local government, water managers, and the agriculture sector, for whom the results are relevant when considering long-term management of water resources and land use. It was commissioned by the Climate Change Office of the Ministry for the Environment, and the Ministry of Agriculture and Forestry.

This section of the report describes the background to the drought risk study, introduces the index we use in this report to quantify 'drought', and discusses the use of scenarios to cover the range of possible future changes in drought risk. Section 2 introduces the reader to the historical variability of drought, in both space and time. Section 3 applies the climate change scenarios to project how drought risk might change under global warming. Sections 4 and 5 provide some discussion and a list of references. A technical appendix (Section 6) supplies more detail and additional discussion of issues raised in the main report.

1.1 Background to Drought Risk Study

Human activity is increasing the concentration of greenhouse gases in the atmosphere, and leading to global climate changes (IPCC, 2001). Scenarios of future climate change for New Zealand suggest that rainfall and temperature changes will differ between different parts of the country (Mullan et al., 2001; Wratt et al., 2003). These changes are expected to increase drought risk for much of New Zealand, and especially the drought-prone eastern regions of the country.

The study was undertaken in two phases. The first phase (see Appendix) was to develop a quantitative indicator of drought risk for this study and apply it to the recorded climate from recent decades to assess how variable and severe droughts can be under the 'current' climate. The second phase (the topic of this report) was to apply the drought risk indicator to a number of climate change scenarios to show a plausible range of effects that climate change may have on drought risk around the country.

1.2 Drought Index: Potential Evapotranspiration Deficit

A consensus emerged from the drought risk workshop held in Phase 1 of this study that a drought index based on potential evapotranspiration deficit (PED) would be suitable for assessing changes in drought risk. The method used for calculating PED is given in the Appendix (section 6.3) of this report.

PED is measured in millimetres (like rainfall), and can be thought of as the amount (depth) of water we would need to supply a crop, in addition to observed rainfall, to prevent loss of optimum production through water shortage. For example, a PED of 200 mm over a growing season could be overcome by applying 200 mm of water at appropriate times through irrigation. The total volume (in cubic metres) of water needed in that case would be: 200 times the paddock area in hectares, times the irrigation efficiency factor, times 10 (to convert to cubic metres).

PED is derived from a water balance model for the topsoil, which accounts for water gain from rainfall and loss from evapotranspiration (Coulter, 1973; Porteous et al., 1994). Evapotranspiration is the loss (or consumption) of water from an extended area of a short green crop (e.g., pasture) to the atmosphere through evapotranspiration (from the soil and other surfaces) and transpiration (from plant leaves and stems). Potential evapotranspiration (PET) refers to the maximum amount of water a crop can consume to meet both its physiological requirements and atmospheric demand when it is well supplied with water. When the crop is short of water at times of low rainfall, a gap develops between the potential water consumption (PET) and what the plant is actually consuming because of the dry weather. This gap is referred to as the potential evapotranspiration deficit, or PED.

In effect, PED is approximately equivalent to the amount of water that would need to be added by rainfall or irrigation to keep pasture growing at its daily potential rate. The Technical Appendix describes the relationship between PED and 'days of evapotranspiration deficit', a concept with which farmers are more familiar.

Our method for calculating historical values of PED at a particular location requires daily values of rainfall and potential evapotranspiration. We obtained these from a January 1972 - December 2003 data set prepared by NIWA (Tait et al., 2005). This uses daily measurements from New Zealand climate observing stations to estimate climate parameters on a 0.05° latitude by 0.05° longitude grid (approximately 5km by 4km) covering the whole country. Daily values of PED were accumulated over July to June years, beginning from zero on July 1st each year. These start and end points were chosen because PED accumulation is close to zero in the winter months most of the time.

We need to use an accumulated total (not just daily amounts of PED) because droughts are the result of dry conditions over a period of time. When discussing PED and its changes in this report, we use the July-June accumulated total unless otherwise stated.

1.3 Climate Change Scenarios

The standard approach to assessing future impacts of climate change is to develop 'scenarios' that take account of the range of estimated future emissions of greenhouse gases, and also the variation between models in the projected patterns for the New Zealand region. The global climate models predict trends in broad climate patterns across the Pacific, but do not take account of the effect of New Zealand's topography on the local climate. The local changes are inferred from the coarser-scale information of the global climate models by a statistical technique known as 'downscaling'.

Statistical downscaling starts with historical observations, and calculates "downscaling relationships" between broad regional climate patterns and these local climate observations. The downscaling relationships are then applied to the broad future regional patterns predicted by the global models, in order to provide more locally-detailed projections for New Zealand (e.g. Mullan et al., 2001). In the present study, we use the gridded New Zealand January 1972 - December 2003 data set described in Section 1.2 to build up historical relationships between monthly broad-scale climate patterns and local monthly rainfall and PET at locations on the 0.05° latitude by 0.05° longitude grid. These relationships are then applied to projected monthly regional climate patterns from a particular global climate model for the "2030s" (defined as the period 2020-2049) and the "2080s" (2070-2099). The result is a set of monthly rainfall and PET projections for each of these periods, downscaled from the global model to each location on the grid.

Two global climate models were chosen for developing these future scenarios: a CSIRO model (known as CSIRO Mark 2), and a model from the UK MetOffice Hadley Centre (known as HadCM2, but referred to as 'Hadley' in this report). These models have been used in previous New Zealand climate change work (eg, Wratt et al., 2003), and have similar global-average temperature changes. However, their downscaled climate changes for New Zealand are rather different. The downscaled Hadley model predicts that New Zealand's east will get even warmer and drier in future compared with the west. The CSIRO model, on the other hand, has a larger (but geographically more uniform) temperature increase over the country but a smaller change in the west to east rainfall difference.

The next step is to adjust the modelled climate changes to be consistent with global temperature projections from the Intergovernmental Panel on Climate Change Third Assessment report (IPCC, 2001), through a procedure outlined in Wratt et al. (2003). The IPCC concluded that by 2100 the global mean surface temperature could increase by between 1.4°C and 5.8°C. In the present study we produced two projections for each global climate model we used. The first corresponds to a global temperature change 25% of the way between the lower and upper bounds of the IPCC range, and the second to a global temperature change 75% of the way across this range (see Appendix 6.6 for details). This choice reflects the fact that some climate scientists consider the extremes of the IPCC range to be less likely than the intermediate values (e.g., Wigley and Raper, 2001). Although the low probability extreme values are driving the international debate about "dangerous" climate change, we did not wish to emphasise the extremes in this report.

To summarise: Four scenarios are developed from a combination of two climate models (which produce different patterns of change at the local scale) and two scalings (to account for differences in global emissions and temperature response). The four scenarios (Table 1.1) represent a range from a "low-medium" scenario (25% IPCC scaling, CSIRO model) to a "medium-high" scenario (75% IPCC scaling, Hadley model). However changes in drought risk which are smaller than those projected under our "low-medium" scenario are possible, particularly if substantial international action is taken to reduce greenhouse gas emissions. Similarly, changes greater than our "medium-high" scenario are also possible.

The scenarios provide monthly average changes of rainfall and PET at each location on the 0.05° New Zealand grid for the 2030s and the 2080s. We also have daily values at these grid points from the historical 1972-2003 analysis described in Section 1.2. We can therefore produce a hypothetical daily time series of future rainfall at each grid point by multiplying each daily rainfall from the historical record by a monthly adjustment factor equal to the ratio of the projected future rainfall to the observed rainfall for that month. This adjustment procedure leaves unchanged the number of wet days per month, and the year-to-year variability in monthly rainfall, relative to the present climate. Daily grid point values of PET for a particular month are assumed to equal the projected average PET value for that month (day to day PET variations within a month have little effect on accumulated PED).

Finally, these projected daily grid point values of rainfall and PET are used as input to a water balance calculation to obtain daily PED values at the grid points, which are accumulated over July-June years to produce annual PED projections for the 2030s or 2080s.

Figure 1.1 shows the change in summer precipitation and PET for the CSIRO and Hadley models, as determined for the 2080s with the 75% IPCC scaling. The pattern of change for 25% IPCC scaling is identical to that of 75% scaling, but the amount of change is smaller. The 2030s changes are, in general, similar but weaker than the 2080s, particularly in the case of the Hadley model. The CSIRO model tends to weaken the westerlies over New Zealand in the first 50 years (to 2030s) and thereafter strengthen them (Mullan et al. 2001), so the 2030s pattern can differ from that for the 2080s.

Summer is chosen for this example, as the hottest time of year with the highest evapotranspiration. The projected summer precipitation changes are similar in pattern, but quite different in size, for the two models. The Hadley model projects large precipitation decreases in the drought-prone eastern regions, in all seasons of the year. The CSIRO model has small precipitation decreases in the east, in spring and summer. In winter (not shown), the CSIRO model projects large precipitation increases in the east, but this is not a critical time of year for drought.

The PET changes are much more similar for the two models, in spite of the greater warming by the CSIRO model. In the Hadley model, the increase in windiness compensates for the smaller warming. Over the whole year for the 75% scaling projections (not shown), total PET increases by about 100mm or more in the currently driest eastern parts of New Zealand. Thus, in the east of the country, there is both a decrease in precipitation and an increase in evapotranspiration during the warmest seasons, which we would expect to aggravate the current tendency for droughts in this region.