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

Climate Change and Infrastructure Design Case study: Wairau Catchment, North Shore City

Executive Summary

1. Introduction

2 Data and Methods

3 Results and Discussion

4 Implications

5 Conclusions and Limitations

6 Further Work

7 References

3 Results and Discussion

3.1 Comparison of Results

The results from the 4 dynamic simulations and the "Present" and "Future 3" design storms are compared below. For consistency, all comparisons are made at the 10% AEP level (Annual Exceedence Probability, i.e. a 1 in 10 year storm event). Below is a summary of the comparisons made.

Table 3-1: Modelling Comparisons Made

3.1.1 Catchment Runoff

Catchment runoff is the amount of water leaving a catchment, and is calculated using the hydrological model only - the transformation of rainfall into a runoff hydrograph. Catchment runoff results are reported as a 'peak' or maximum runoff for an entire storm event or time series.

These were directly compared for the design storms modelled, and it was found that the peak runoff from the catchments was between 36% and 68% greater as a result of the Future 3 design storm, when compared with the Present design storm, with the larger differences being in the pervious catchments (due to the initial soil abstraction of rainfall being set under the TP108 methodology). Table 3-2 below provides results from three of the modelled catchments.

Table 3-2: Peak Catchment Runoff Comparison - Design Storms

Because the dynamic modelling results were from the full 100 or 110-year series (and thus the peak catchment runoff was the maximum for the entire time series), and because the rainfall events provoking the 10% AEP event in the catchment are different for all parts of the catchment, direct comparisons of catchment runoff from these models cannot be made.

3.1.2 Flows in Channels

The design storm modelling approach produces theoretical maximum flows with a certain annual exceedance probability at a given point in the system

Each time-series contains 150 years of randomly generated rainfall events, based on historical patterns. However, there is no certainty that a 1% AEP event did actually occur within the time series, although there is more confidence that the smaller frequency events (eg a 10% AEP flow) have occurred on a number of occasions.

MOUSE LTS software analyses extreme event data and identifies the 10% AEP flow by ranking the data. Probability distributions are usually applied to ranked rainfall and flow data to relate the magnitude of extreme events to their frequency occurrence. Analysis undertaken with flow results at the gauge showed that the MOUSE LTS estimates of event frequency did not produce consistent results. Annual extreme flows for the time series data were analysed using three different probability distributions commonly used in New Zealand; Gumbel (Extreme Value Type I), Log-Pearson Type III, and Log-normal. The Log-Pearson III and Gumbel distributions produced very similar results, and the results of the Log-Pearson III analyses are presented in Table 3-3 below. Water levels corresponding to the 10% AEP flow have been estimated based on events of similar size occurring in the time series'.

Table 3-3: Comparison of 10% AEP flows and Water Levels - Design storm and Dynamic Models

* Approximation from MOUSE results

Present Climate vs Future Climate - Design Storm Model

The 'Future 3' design storm predicts flows approximately 30% to 40% greater than those resulting from the 'Present' design storm in both the mid and lower- catchment channels. This is comparable to the 40% difference in 24-hour rainfall volume during the design storms. The associated water level rise in the mid-catchment is 2%. At the outlet, water levels rise by 10%.

Present Climate vs Future Climate - Dynamic Model

Climate change comparisons showed that the dynamic model predicted a maximum increase under climate change of 4% for the 1 in 10 year flows at the outlet of the stormwater system. Channel flows increase from Future 1 to Future 3, with Future 3 producing the largest changes in flow. Of interest are the lower than expected flows of Future 2 - the increase in dry periods in the climate scenario are possibly providing additional storage within the catchment. The extended dry periods in the Future 3 scenario, however, are causing high intensity rainfall events, which are not buffered to the same extent as the lower intensity events in the Future 2 scenario. Water level rise associated with these changes in flow is negligible.

Design Storm vs Time Series

When comparing the design storm flows to the dynamic model flows, the 'present' time series dynamic model produces channel flows that are lower than the 10% AEP flows from the 'present' design storm model. The dynamic model predicts flows at the gauge that are 10% lower than those predicted with the design storm model at the gauge, and 15% lower at the outlet.

Water levels associated with the changes in flow change only slightly in the channels, due to the large cross sectional areas in the channels and flood plains providing additional storage.

The "Future 3" dynamic model predicts flows that are 29% and 37% lower (for the mid and lower-catchment channels respectively) than the 'Future 3' design storm model 10% AEP flows. This is a much greater difference than the present design storm / present dynamic model comparison, and it is possible that additional storage in the catchment soils and network (ponds and channels) is available as a result of the increased dry periods in the 'Future 3' time series, 'buffering' the increased rainfall. Because the increased dry periods in the Future 3 time series lead to higher intensities during storms, grouping these together to form a design storm does not account for the dry periods, and this may need to be reviewed with respect to adjusting the design storm.

The different approach in modelling rainfall and run-off in the design storm and dynamic time series models makes it difficult to fully resolve the differences in run-off for the present climate, and the changes under future scenarios. Use of dynamic time series modelling under climate change scenarios requires significant scientific expertise; results from dynamic time series modelling should not necessarily be treated as representing a more realistic representation of present or future conditions.

3.1.3 Flooded Nodes and Links

A number of nodes were investigated with respect to the comparison of the number of times the water level rises above an identified 'critical level'. This level has been set at either the top of the lined channel, or the top of the bank, depending on the location (proximity to houses, shape of the channel).

Figures 3-1 and 3-2 below identify the locations where the water level at a location in the network exceeds the critical level assigned to that point, under both the present climate scenario, and the future climate scenario.

Because all of the nodes which flooded under the present climate scenario also flooded under the future climate scenario, results are presented as:

• Areas that would need upgrading, regardless of the climate scenario used; and
• 'Additional' areas that would only need upgrading if the future climate scenario were applied

Present vs Future Climate

In both the design storm and dynamic modelling examples, the Future scenario yielded a higher number of flooding nodes than the Present scenario. Under the present design storm, 45 locations had a water level greater than the critical level. This rose to 55 locations when the Future design storm was modelled.

The difference between the present dynamic model and the Future 3 dynamic model was not so great, with an additional two points in the system identified as flooding in the Future, over and above those susceptible under the Present dynamic model.

3.2 Number of Floods vs Design Requirements

The flooding locations can be investigated further by looking at the number of times a particular link or node in the model overflows. If a link is to carry the flow resulting from a 10% AEP design storm, and overflows 10 times during a 100-year series for the same climatic conditions, it could be identified as having a 1 in 10 year level of service. If, on the other hand, it only overflows twice during the 100-year series, the link could potentially be over-designed (i.e. the design storm used may have been too conservative). Table 3-4 below displays a number of example locations throughout the model where there is a significant change in the flooding occurrences under the different climate change scenarios.

Table 3-4: Flooded Location Examples in the Lower Catchment for a 10% AEP Event

*LOS = Level Of Service for that location (Protection required against rain events of a given frequency)

The results from the first node (J-13c) show that during the present design storm, the channel would not overflow (it is in fact, sized almost exactly to accommodate the flow from this design storm). Additionally, when tested with the present time series, the water level in the channel would only top the banks in two storms. Because the required level of service is the 10% storm, it would be acceptable for the water to overtop the banks 10 times in 100 years. This link is therefore capable of maintaining the current level of service under the future climate change scenarios, and is possibly over-designed for the current climate.

Node J-13b, on the other hand, accommodates the 'present' 10% AEP design storm flows, and additionally appears to be within the acceptable flooding limits for the 'present' climate scenario for the 5% AEP level of service required at that link, however this shows that the 10% 'present' design storm simulates a 20% AEP flow under the present dynamic model at this point. Additionally, under future climate change scenarios, the number of floods in 100 years would most likely mean that the level of service would not be maintained.

Node J-20e would appear to be under-designed - Figures 3-1 and 3-2 above highlight the number of locations in the Wairau catchment where this is the case.