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

A methodology to assess the impacts of climate change on flood risk in New Zealand

Executive Summary

1. Introduction

2. Planning and decision-making context

3. Methodology

4. Assumptions and uncertainties

5. Further reading

Appendix: Results for the Buller Catchment example

Appendix: Results for the Buller Catchment example

This appendix reports on the use of the above approach in assessing the impact of climate change on flooding in the Buller catchment. Here we will show the decisions we made in choosing the weather events and climate change scenarios, and the results achieved by the weather, rainfall-to-river flow, and inundation models to calculate the flood risk from current and future 1% and 2% AEP events.

Select base event(s)

The events we selected were chosen to be representative of a range of rainfall generating weather systems, and hence are not necessarily the worst ever experienced. They were also chosen, somewhat pragmatically, as events for which we had sufficient data for modelling. The three chosen started on 12 November 1999, 14 August 2000, and 8 December 2001, and generally between 3 and 7 days of weather was modelled. Rainfall totals for the events were around 91 mm, 95 mm and 537 mm respectively. These totals are not extreme events and would have return periods of less than 1 year.

Select climate change scenarios

We chose target years of 2030 and 2080, with mid-low and mid-high points to cover uncertainty, from Table 2 of Preparing for climate change (see Further Reading, main report). More details can be found on pages 11 and 12 of Climate Change Effects and Impacts Assessment. The relevant values are:

We chose the roughly 25 and 75 percentiles of the total annual average temperature range, which gave temperature increases that we approximated to 0.5°C and 1°C for 2030, 1°C and 2.7°C for 2080. Choosing 1°C for both time scales saves significantly on the cost of the modelling. 2030 and 2080 were chosen as time scales that were representative of the design lifetime of much of the infrastructure a council would manage.

Screening results

The results of the screening method, drawn from Table 7 (shown below, and which indicates the percentage increase in rainfall per 1°C) of Preparing for climate change, suggest that for a 6-hour duration 1 in 50 year event (2% AEP) and for 2030, mid-low and mid-high rainfall multipliers should be 3.5 (0.5°C increase), and 7.1 % (1°C increase) and by 2080 should be 7.1 and 19.2% for increases of 1°C and 2.7°C respectively.

As there were insufficient representative gauges in the catchment, rainfall information produced by the weather model for current conditions was used as the base rainfall information. This rainfall was increased by the factors suggested in the MfE guideline tables. The current and future rainfall estimates were then used as input into the Topnet model of the catchment.

Increasing the rainfall for our 3 storms led to increases in river flow as shown in Table 1. Average changes are shown in Table 2.

Results of the screening test

An inspection of the flow model output shows that, for the current climate scenario, the models have reproduced well the measured peak flows. These good results come from modelling the weather at sufficient resolution, and from a refined hydrological model.

This screening test suggests that for the Buller catchment, climate change is likely to produce increases in peak flow of up to (mid-high scenario) 20% by the 2080s. As this is likely to cause a significantly larger flooding risk, a more detailed analysis would be justifiable.

Using the weather model to determine the rainfall increase for the projected temperature rise

These systems were modelled with the RAMS mesoscale model at 5 km resolution. The model was initialised with United Kingdom MetOffice unified global model analyses at 60 km resolution. Rainfall was output at hourly intervals.

The temperature of the air and the sea surface were adjusted in the weather model in accordance with the climate change scenarios. The weather model height fields were adjusted to achieve hydrostatic balance. This is necessary for the model to be able to adjust to the new climate.

The Buller catchment is interesting, in that it lies behind the Paparoa ranges, and is hence in the lee of these coastal hills. This complicated the response to the climate change scenarios, in that the Paparoas appeared to take the "brunt" of the change, with little and sometimes even less rainfall spilling over into the Buller catchment. This led to the response of the model for storms being complicated, with the rainfall showing little change for 0.5°C and 1°C scenarios. Bigger changes were seen for the 2.7°C scenario, particularly on the larger storm. This has meant that the interpretation of the results is difficult, and has led to wider standard errors than would be expected for catchments that don't sit in the eastward lee of a major mountain range.

From Table 5, it is apparent, compared to the screening method increases, that the weather modelling examples have a significantly higher increase in rainfall for the mid-high scenario (2.7°C increase) for 2080 events. While Table 3 shows the were only small changes for the 0.5 and 1.0°C cases, particularly for the lighter events, the 2.7°C event had rainfall changes ranging from 18 to 48%.

In this analysis we have combined 3 weather events by simply averaging the increases in peak flow. Given the range of results across the three storms, and given that the more intense storm showed significantly higher increases, it is apparent from Table 4 that we have sampled too few storms to be confident of our results. Investigating several more storms, particularly intense ones, may help improve the confidence in these results.

A more detailed analysis of the weather model output was undertaken to test the validity of this work, and some of the assumptions that have been made.

One implicit assumption is that the remodelling of the storm for a warmer climate would not lead to a significantly different storm developing, and hence invalidate the comparisons. Figure 1 to Figure 6 show the changes in rainfall for each of the global warming scenarios for each of the events. The rainfall shows consistent changes in intensity or pattern of the time series, with no significant extra developments.

These figures also show that the Buller catchment, lying as it does in the lee of the Paparoas, has no significant overall increase in rainfall for the first two climate change scenarios, but there are large changes in the rainfall at other locations. For example, the area to the south of the catchment, near Mt Cook, shows large increases in rainfall with the first two scenarios for the first two events. Indeed, there are increases in rainfall on the upwind side of the Paparoas, but this appears not to spill over into the Buller catchment.

Results of the weather modelling have shown that, while the screening approach is a first step, it is quite simplistic. It does not take into account any of the topographic and catchment characteristic effects, and only accounts for the greater moisture holding capacity of the air. The weather modelling enables us to approximately estimate the positive feedback that the increase in moisture will have on storms and explicitly models the effects of the topography. In this example, the effect of the Paparoas, while not influencing the screening approach, has been shown to be significant in the weather modelling.

Translating projected rainfall into river flow

The hourly rainfall from the weather models was used as input into the Topnet (see Bandaragoda et al, 2004, in further reading, main report) model of the Buller catchment. This is a sophisticated spatially-distributed model that is capable of replicating the hydrological characteristics of this catchment.

Figure 7 shows the change in the rainfall over the Buller catchment for a 2.7°C increase in temperature. Figure 8 shows the consequences of this change, assuming all other things remain unchanged, on the likely flood at the Te Kuha river flow measurement station just upstream of Westport.

Table 5: A comparison of the screening and weather modelled changes in rainfall.

The results shown in Figure 7 and Figure 8 are for a large storm. Similar calculations were carried out for two other, smaller, storms and these results indicate a less dramatic increase in the flood peak magnitude. A summary of the results is shown in Table 6. These results show that the model has replicated well the flow observed for the current climate, giving confidence that the results from the climate change scenarios are valid.