As we have seen, the extent of the effect of climate-induced changes on inundation mainly depends on changes in flood flows and changes in sea level. We will now look at tools for estimating potential changes in inundation. Although ‘off the shelf’ flood-modelling software is widely available, the complex techniques described here should only be applied by experienced personnel with a full understanding of fluvial hydraulics.
The flood-related effects of climate-induced changes in sediment transport capacity and adjustments in channel morphology are not covered in detail in this manual. Climate-related sediment movement depends on flood size and frequency, and can be affected by sea-level rise and flood protection works. The implications of sea-level rise are particularly significant because a higher sea level causes a flatter river gradient near the coast. This reduces the velocity of floodwaters in the river channel, encourages silting and aggradation and further restricts river conveyance. As noted in section 2.4.4, changes to channel morphology include both aggradation and degradation, the extent of which can vary in time and space. Such changes are very site specific and will warrant specialised investigations.
5.1.1 Riverine inundation
In situations where flood-plain maps for a range of flood sizes are already available, estimating increased inundation due to increased river flows may be a straightforward process of reinterpreting the return period assigned to each existing inundation map (eg, a 100-year ARI map may become a 50-year ARI map in a particular case, if rules of thumb can be developed, as described in section 4.2.3). In all cases, the event to which the recurrence interval applies must be clearly specified. Usually, probabilities of exceedance relate to the water source of the flood, such as rainfall or river flow. The probability of different levels of inundation and joint probabilities of two or more events can be very complex and will require specialised studies (see the Westport example in chapter 6).
Where flood maps are not available, or are inappropriate for future scenarios, eg, where the shape of inflow flood hydrographs is expected to change significantly, new hydraulic modelling will be necessary.
As discussed in chapter 2, over the short term, inter-decadal climate variations may have a more pronounced effect on inundation than climate trends, and studies of regional flood frequencies in relation to climate patterns (such as the IPO) may be required.
5.1.2 Sea-level rise and storm surge
Coastal riverine communities must prepare for flooding that results from the effects of sea-level rise and storm surge. Predicted sea-level rise due to climate change will cause lower freeboard on coastal flood protection structures, increased inland influence of tides, and a flattening of river slopes in coastal reaches (see the Coastal Hazards and Climate Change manual, Ministry for the Environment, 2008b). The reduction of a river’s slope reduces the energy of the flood flow, increases the depth of flow and reduces the sediment-transporting capacity of the flow. Any increase in the frequency of floods also increases the chance of floods occurring during adverse tidal conditions. These effects are complex, but the consequences can be calculated using hydrodynamic models. There is also the potential for increased frequency and magnitude of wind set-up and storm surge, which result when high winds and decreased barometric pressure raise the local ocean level.
In calculating potential future sea levels, note that the present sea level may be higher than the level at the time the local datum was established (eg, the Lyttelton mean sea-level datum was set in 1937).
The Coastal Hazards and Climate Change manual makes recommendations about the sea-level heights that should be considered in risk assessments. Tables 2.2 and 2.3 on page 21 of that guidance manual should be referred to for more detail on this issue, and for discussion on other coastal hazard drivers such as waves and storm surge.
5.1.3 Top water levels and freeboard
Common to both inland and coastal flooding design is the determination of top water levels (TWLs) and freeboard. TWLs are modelled heights that floodwaters are expected to reach during a flood. Freeboard is an allowance for limited knowledge that could not be included in the modelling, such as limited ground survey data and differences in model assumptions. In a sense, TWLs account for certainties while freeboard accounts for uncertainties. For the purposes of flooding design, climate change must therefore be disaggregated into what is known, which then informs TWL modelling, and what is not known, which is folded into the freeboard allowance. The ‘knowns’ are simply the products of advanced modelling methods to predict flood flows; the ‘unknowns’ are the uncertainties stemming from the scenarios and results from preceding modelling efforts.
Freeboard allows for the uncertainties in hydrological predictions, wave action, modelling accuracy, topographical accuracy, final flood defence levels and the quality of the digital elevation models. The increase in flood levels associated with climate change is in addition to freeboard, because the uncertainty freeboard incorporates is not reduced in future climate scenarios. Therefore, freeboard should not contain the ‘core’ component of climate change impacts, but may be increased to account for climate change uncertainties.
5.2 Screening methods
The simplest screening method is to assess whether land has been inundated in the past. If it has, and no flood remediation works have been carried out, it is clear that increased river flows and sea levels would cause increased inundation of these areas. More precise methods can be used to estimate whether thorough investigations are warranted. However, coastal flooding is particularly difficult to forecast because of the overlapping of two climate change effects: altered river flow and increased sea level. This combination means you need to be particularly cautious when interpreting screening studies in these cases.
The screening methods in this chapter can be used to assess whether a flood hazard is expected to change significantly, and to decide whether further investigations of the extent, depths and velocities of floodwaters are required.
5.2.1 Non-coastal river reaches
The following techniques can be used to roughly assess the effects of river floods on inundation. The techniques assume that climate change-adjusted river flows have been estimated using information derived from the methods described in chapter 4.
Where flood-plain maps for a range of flood sizes are already available, a useful indication of increased inundation due to increased river flows can be made by re-interpreting the return period assigned to each existing inundation map. For example, if the screening methods described in chapter 4 have enabled a return period to be interpreted for flood flows (perhaps the 100-year ARI flood flow has become the 50-year ARI flood flow, as above), the corresponding inundation map would be an appropriate first indication of inundation (ie, the old 100-year ARI map becomes the 50-year ARI map).
Where there are river flow recording stations in an area of interest, existing rating curves can be used to convert future river flows into corresponding river levels. Alternatively, in uniform reaches of the river, information on channel size and roughness can be used to convert river flows into river levels using a flow resistance equation (eg, Mannings equation; see section 6.2). These levels can be propagated across a digital representation of the local terrain using GIS techniques to obtain a rough estimate of the extent of flood inundation using GIS software such as ARCMAP or MapInfo. In ARCMAP (with the Spatial Analyst extension) the cut/fill procedure can be used.
Such methods can provide a quick assessment of whether flood inundation is likely, but they do have several drawbacks, including:
- there is usually no rating data for extreme flood flows
- the volume of floodwater from the river may not be sufficient to fill the flood plain before the river level falls
- the slope of the flood plain may prevent water from ponding on the flood plain
- no indication of out-of channel flood velocity is available.
5.2.2 Coastal river reaches
A raised sea level can be propagated across a digital representation of coastal terrain using GIS techniques to obtain a rough estimate of the extent of inundation, as described above. However, the interaction of river flood flows with varying sea levels is very complex and cannot be estimated using general GIS techniques, and more advanced hydraulic methods must be used.
5.3 Advanced methods
Advanced methods for assessing the depth and extent of inundation vary in their complexity but are all based on fluid hydraulics. Specifically, they vary in the dimensionality with which they represent reality; one-dimensional models approximate river flow along a single line, while three-dimensional models consider flow complexities, both across a channel and to depth.
Climate change is accounted for in each of the following approaches by (i) altering the flow that enters the modelled domain, and, in the case of coastal inundation, (ii) altering the hydraulic conditions under which water flows out of the modelled domain.
5.3.1 One-dimensional (1-D) numerical models
With 1-D models the river and its flood plains are represented by many cross-section slices. Each cross-section has a flat water surface and constant average velocity. Cross-sections are spaced closely enough to capture the main features of the topography. Flow circulation patterns cannot be resolved. Flow paths are determined by the model designer. Buildings cannot be represented in 1-D models and their effects are incorporated by increasing the cross-section roughness. Models of complicated situations take time to set up, but they run quickly on a desktop PC, so that Monte Carlo simulations are possible to help quantify uncertainty. Predicted cross-section flood levels can be interpolated to give a map showing the extent of inundation. Examples of 1-D models include AULOS, MIKE-11 and HEC-RAS.
5.3.2 Two-dimensional (2-D) models
A river and its flood plains are described by three-dimensional digital representations of the ground surface roughness and elevation. Water levels and velocities can vary in all horizontal directions. Plan-form flow circulation patterns are reproduced. At any given point, water velocities are the same at the water surface as at the bed (depth-averaged flow). Flow paths are determined by the ground topography and roughness. Models of complicated situations may take several days to run on a desktop PC. The model results indicate local flood depths and velocities at each node of the digital elevation model. Some models use 1-D equations for the river channel and 2-D equations for the flood plains. Examples of 2-D models are RiCOM, Hydro-2de, River2d and MIKE21.
5.3.3 Three-dimensional (3-D) models
In these models a river and its flood plains are described digitally as for a 2-D model, but calculated water velocities may vary in all three dimensions and secondary currents can be reproduced. These models are more commonly used for specific, detailed investigations, such as the flow around structures. Large or complex models require a supercomputer to reduce the run time. Examples of 3-D models include FLUENT, CFX and FLOW-3D.
- Where flood maps exist, screening-level estimates of increased inundation due to increased river flows can be a straightforward process of reinterpreting the return period assigned to each flood inundation map.
- For predicting effects where there is coastal interaction (sea-level rise or storm surge), new hydraulic modelling is necessary.
- Coastal riverine communities are most at risk and require specific hydraulic investigations.
- For non-coastal flood-prone communities, over the short term inter-decadal climate variations may have a more pronounced effect on inundation than climate trend. Studies of regional flood frequencies in relation to applicable climate patterns (eg, IPO) are required.
- A range of methods have been presented which provide estimates of how altered river flow predictions may affect inundation levels. The aim of each method is to convert extreme flow data into an estimate of flood height and spatial extent.
- As with the methods presented in chapters 3 and 4, each method differs in its complexity, data requirements and the reliability of results, as well as user experience needs. All are available to practitioners, subject to their experience.
- As was recommended in chapters 3 and 4, you should consider three factors when deciding how to develop predictions of inundation:
- what flow, channel and coastal data is available?
- what accuracy and precision do we need?
- do we have access to the expertise and computing facilities to undertake the analysis and modelling?
- The summaries in table 7 will help you choose the most appropriate method.
|Method (example)||Description||Advantages||Disadvantages||Data and climate change requirements|
|Screening methods||A general approach based on past, more complex studies.||Easy to use; based on comprehensive analysis.||Limited to inferences from past studies.||Case-by-case considerations.|
|1-D flow models (HEC-RAS, AULOS, MOUSE, MIKE-Urban, MIKE-11, Infoworks CS)||Produce flow depths and velocities down a 1-D channel.||Low data requirements and relatively rapid computation.||Linear flow paths determined by the modeller; lacks 2- and 3-D flow patterns.||Inflow hydrograph; downstream hydraulic conditions; river and flood-plain cross-sections; roughness coefficients; calibration observations.|
|2-D flow models (MIKE-21, Infoworks RS, RiCOM, Hydro-2de, River2d)||Produce flow depths and velocities across a complex 2-D terrain.||Simulate variable flow depth and flow velocity laterally.||More computationally intensive than 1-D; lack 3-D flow patterns.||As with 1-D models; digital elevation model of the river and flood plain.|
|3-D flow models (FLUENT, CFX, FLOW-3D, MIKE 3)||Produce flow depths and velocities around 3-D structures.||Simulate vertical flow patterns.||More computationally intensive than both 1-D and 2-D.||As with 2-D models; 3-D representation of structures.|
Note: Data requirements specifically related to climate change are underlined.