Managing Uncertainty

There are two main types of uncertainty that can affect the inclusion of hazards in land use planning:

1. Uncertainties in the hazard modelling; and
2. Uncertainties in the decision making process, as shown in Figure 1. When levels of uncertainty are deemed to be significant, the precautionary planning approach should be implemented.

Figure 1. Uncertainties in modelling and decision making (van Asselt 2000, p91)

Figure 1. Uncertainties in modelling and decision making (van Asselt 2000, p91)

Uncertainty in hazard modelling
It is important to be aware of uncertainties in hazard modelling, to ensure that the limitations and assumptions of the modelling are well understood, taken into consideration (see following sub-section on mapping uncertainty), and the modelling data and quality are retained.

Uncertainty is encountered at various steps of the modelling process. There are four types of modelling uncertainties as outlined below and in Figure 1 (van Asselt 2000):

1. Technical: from the quality or appropriateness of the input data used to describe the system, from aggregation (temporal and spatial) and simplification, as well as from lack of parameters from data and approximations;
2. Methodological: due to uncertainty in equations and model structures;
3. Epistemological: uncertainty in levels of confidence and model validity; and
4. Model operation uncertainties: due to hidden flaws in technical equipment, and/or accumulation of uncertainties propagated through the model.

For example, uncertainties in tsunami inundation modelling include:

- uncertainties around earthquake parameters;
- the quality of the information about: water interaction with ground roughness (including buildings and land use types);
- quality of digital elevation model (map contours vs. LIDAR); quality of bathymetry; real shape of ocean displacement (e.g. fault offset or bulge); and
- reflections and refractions of waves across the ocean.

Uncertainties from the modelling software can be reduced through validation of the modelling software using benchmark cases or common validation standards.

Various types of uncertainty in decision making may play a role in the process of deciding whether to incorporate tsunami modelling into land use planning. For example, political uncertainty may arise as the decision maker struggles with the political acceptability of options (van Asselt 2000). To overcome this, decision makers need to be provided with an opportunity to learn and understand the importance of information and modelling, and the role it can play in reducing future risks to communities. [see Engagement]

Mapping uncertainty for land use planning
For other natural hazards i.e. active faults and landslides, uncertainty is already included in planning maps.

Active faults
Examples of mapping uncertainty include using ‘well defined’, ‘constrained’ and ‘distributed’ mapping of active faults (Kerr, Nathan et al. 2003, see Figure 2); and landslide ‘core’ and ‘fringe’ areas (Saunders and Glassey 2007, see Figure 3).

Figure 2. Example of uncertainty cones in active fault mapping, with colours depicting type of fault and level of uncertainties (Kerr, Nathan et al. 2003

Figure 2. Example of uncertainty cones in active fault mapping, with colours depicting type of fault and level of uncertainties (Kerr, Nathan et al. 2003

Landslides
Many locations in New Zealand have been developed on areas of slope instability. While it is best to avoid these areas, in many cases the hazard was not known about before development occurred. Figure 3 shows an example of one of these areas in Nelson, where the core area of a landslide has been mapped where the hazard is known (red), surrounded by a fringe area where the edge of the active slump has not been able to be accurately defined (green) (Nelson City Council). Rules stipulate that new residential units within the core are non-complying; within the fringe area they are discretionary (Rule REr.77.3).

Figure 3. The Tahunanui Slope Risk Area, commonly known as the Tahunanui Slump, is presented on the Planning Maps within the Nelson Resource Management Plan (Nelson City Council).

Figure 3. The Tahunanui Slope Risk Area, commonly known as the Tahunanui Slump, is presented on the Planning Maps within the Nelson Resource Management Plan (Nelson City Council).

Tsunami
For tsunami, a similar approach can be taken as for landslides, as shown in Figures 4 and 5. Figure 4 presents a cross section of modelled probabilistic tsunami wave heights at the coast, and associated levels of (un)certainties. The middle hashed zone is bounded by the lower and upper levels of a chosen level of confidence. Figure 5 presents a birds-eye view of the zones shown in Figure 4.

Confidence levels are expressed as percentages. On a graph or a map they define a confidence interval either side of an average value. In Figures 4 and 5 this average value lies in the middle of the hashed ‘Uncertain tsunami inundation’ zone. The confidence interval is the size of the hashed ‘uncertain tsunami zone’. For 99% confidence, 1% of the time the true value will lie outside of the interval, while at 95% confidence, 5% of the time the true value will lie outside. Choosing a higher confidence (e.g. 99% instead of 95%) will make the hashed zone larger; the lower limit will become closer to the coast and the upper limit will be further inland.

Figure 4. Cross section of modelled probabilistic tsunami wave heights at the coast, and associated levels of (un)certainties (to a chosen level of confidence – i.e. a confidence interval).

Figure 4. Cross section of modelled probabilistic tsunami wave heights at the coast, and associated levels of (un)certainties (to a chosen level of confidence – i.e. a confidence interval).

Figure 5. Map view of the tsunami inundation shown in Figure 4.

Figure 5. Map view of the tsunami inundation shown in Figure 4.

The confidence interval used depends on how certain one needs to be that the following two situations will not occur:
(a) a section of the ‘high certainty of tsunami inundation’ zone is actually not at risk from tsunami; or
(b) a section of the ‘high certainty of no inundation’ zone is actually at risk from tsunami.

Coastal erosion
Coastal erosion zones can be mapped using lines to represent risk zones. For example, Tauranga District Council has mapped three coastal erosion policy areas: the current erosion risk zone (red); the 50 year risk zone (yellow); and the 100 year risk zone (blue). An example of this is shown below.

References
Kerr, J., S. Nathan, et al. (2003). Planning for development of land on or close to active faults. Wellington, Ministry for the Environment.
Nelson City Council. "Nelson Resource Management Plan Map 13." Retrieved 8 October, 2009, from http://www.nelsoncitycouncil.co.nz/assets/Environment/Downloads/RMP-PDFs/planning-maps/urban/MAP13LH.pdf.
Saunders, W. S. A. and P. Glassey (2007). Guidelines for assessing planning policy and consent requirements for landslide-prone land. GNS Science Miscellaneous Series 7. Lower Hutt, GNS Science: 76.
van Asselt, M. B. A. (2000). Perspectives on uncertainty and risk: the PRIMA approach to decision support. Dordrecht, Kluwer Academic Publishers.