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

Chatham Islands Climate Change

Executive Summary

1. Introduction

2. Historical Trends and Patterns in Chatham Islands Climate

3. Expected Temperature and Rainfall Changes

4. Tides, storm surge and the effects of sea level rise

5. Analysis of Heavy Rainfall - Past and Future

6. Discussion and Recommendations

7. References

5. Analysis of Heavy Rainfall - Past and Future

Key points:

• What is an extreme rainfall in the current climate is likely to occur about twice as often by the end of the 21st century under a mid-range temperature change scenario.
• For the high temperature change scenario, an extreme rainfall in the current climate is likely to occur 3 to 4 times as often by the end of the 21st century.
• For the high temperature change scenario, an extreme rainfall in the current climate is likely to fall within about half the duration (e.g., 12 hours instead of 24 hours) by the end of the 21st century.

5.1 Historical high intensity rainfall

High intensity rainfall statistics have a wide variety of uses. These extend from a range of engineering construction work, where allowance has to be made for the disposal of rainfall from storm, to the influence of heavy rainfall on soil erosion and vegetation damage. For many purposes it is necessary to express, in probabilistic terms, the likelihood of various amounts of rainfall for a range of storm durations. To do this the concept of annual recurrence interval (ARI), or return period, is used.

5.1.1 Method

The ARI of a given storm event is the average number of years within which the event magnitude is expected to be at least equalled. This definition assumes that the storm events being considered are the maximum values of all such similar events in a year. For example the series of annual maximum one-hour rainfalls, sampled of several years, meets this criterion. Estimates of high intensity rainfall are often presented in tables that provide the rainfall amounts for various ARI and for a range of storm durations.

Annual maximum rainfalls from Waitangi, Chatham Islands, for 10 standard storm durations (10, 20, 30, 60-minutes, 2, 6, 12, 24, 48, 72-hours) are available for the period 1957 - 1992. To obtain the probabilistic estimates or average recurrence intervals from these data, it is necessary to fit a frequency distribution, such as a three-parameter generalised extreme-value (GEV) distribution (Jenkinson, 1955). This distribution has three characteristic types, known as EV1 or Gumbel, and EV2 and EV3. Once the distribution parameters have been estimated, by a method such as probability-weighted moments (Hosking et al., 1985), the average recurrence interval can be obtained from standard formulae. For each of the 10 durations, the Waitangi rainfall series were fitted to both a Gumbel and full generalised extreme-value distributions, and average recurrence intervals computed.

Figure 5.1 gives plots of rainfall depth for each duration against a plotting variable, known as a "reduced variate". The reduced variate, y_T, on the x-axis has the effect of "straightening" the lines on the plots, and can be used to graphically compare how well different distributions fit the data series. There is also a direct relationship between ARI and the reduced variate. For example, when y_T is 4.6, this corresponds to an ARI of approximately 100-years. In each diagram the rainfall maxima are also plotted. These data are ranked, and an estimate of the reduced variate is made from the ranking position of the rainfall maxima.

All the diagrams in this figure show that the difference between the two extreme value distributions is small, and statistically insignificant, based on a test to see whether the EV1 distribution is an acceptable alternative to the EV2 or EV3 distributions (Hosking et al., 1985). Thus, all these distributions are effectively equivalent when estimating design rainfalls. In this report the EV1 distribution is used to provide table of high intensity rainfall for Waitangi.

5.1.2 Results

Table 5.1 gives a table of rainfall depths for various durations and average recurrence intervals for Waitangi, Chatham Islands. To interpret this table, for example, a 12-hour storm rainfall of 71 mm could be expected to recur on average once every 50 years. Further, as to be expected, the high intensity rainfalls in the table increases monotonically with duration and with average recurrence interval.

5.2 Climate Change Scenarios

The Climate Change Guidance Manual provided a method showing how to adjust high intensity rainfalls for preliminary scenario studies (Section 5.2 and Appendix 4, MfE, 2004a). It was advocated that at least two sets of calculations be undertaken for low and high temperature change scenarios. Table 5.2 shows (for screening assessment scenario purposes) the recommended percentage adjustments per degree Celsius of warming to apply to high intensity rainfalls for various durations and average recurrence intervals. Note that the percentage changes in the table are mid-range estimates per degree Celsius and should be used in only preliminary scenario studies.

For the Chatham Islands, the projected temperature changes for 2030s and 2080s for a low, mid-range and high temperature change are given in Table 5.3 (from Table 3.1 in this report). Tables 5.4 and 5.5 provide the high intensity rainfall estimates for the projected temperature scenarios for the 2030s (Table 5.4) and 2080s (Table 5.5).

5.2.1 How to use the tables

As an example of how to use the above table to compute changes in high intensity rainfall, consider the 50-year 24-hour rainfall. From Table 5.1 the rainfall depth is 88mm, and the projected adjustment for global warming (Table 5.2) is 6.6 percent per degree Celsius warming. For a mid-range temperature adjustment for the 2030s of 0.7°C, the increase in high intensity rainfall is 4.6 percent (i.e. 0.7°C times 6.6%). This gives an estimate rainfall of 92mm for the 2030s scenario, (i.e. 1.046 times 88mm).

5.2.2 Implications

So what does this mean? The changes in rainfall extremes look relatively modest, but in all cases there is an increase with time and with temperature. The best way to interpret the results is to ask how extreme rainfall events might change in the future. For the example above, the current Waitangi estimate in Table 5.1 of 88 mm for a 50-year 24-hour rainfall suggests that for the mid-range scenario this amount of rainfall is expected to have a recurrence interval of about 35 years by the 2030s (Table 5.4b) and about 20 years by the 2080s (Table 5.5b). In other words, what is an extreme rainfall in the current climate is likely to occur at least twice as often by the end of the century. This can be taken as a useful 'rule of thumb' for other combinations in the tables too.

For the high temperature scenario at the 2080s (the 'worst case'), current extreme rainfalls are projected to occur 3 to 4 times more often. Alternatively, what is currently a 50-year 24-hour rainfall (Table 5.1) becomes a 50-year 12-hour rainfall (Table 5.5c). Obviously, this is likely to have implications for drainage and flooding.