Climate change risk management for the water sector

Climate projections such as those published on a global scale by the IPCC and on a regional or national scale by institutes such as the Netherlands’ KNMI have proven invaluable and continue to be an important pillar for understanding future boundaries and needs with respect to the water system. They feed a wide range of research and decisions on topics ranging from flood protection to drinking water demand. Subsequent generations of projections and scenarios (currently IPCC’s AR6 and the KNMI’23 scenarios) are consistent in the predicted direction of change and represent the evolving insights of climate science.  

High Impact events of perceived Low Likelihood

However, there is more to adapting to climate change than understanding how the climate may evolve. In his 2018 paper, Rowan Sutton argues that the climate science community does not pay sufficient attention to risk management. Risk is often understood to be the product of likelihood and impact (and sometimes exposure). According to Sutton, the climate science community, in particular those parts focusing on the physical aspects of climate change, focuses on the likelihood aspect and neglects the impact aspect.  

Risk management is standing practice with many water utilities and water authorities for the management and maintenance of infrastructure. However, the case of climate change related risk management is especially challenging, also for the water sector. Therefore, Sutton’s argument is particularly relevant for our industry. Sutton builds his argument on the pioneering climate change risk management paper by King et al. (2015). These authors propose the following key principles: 

1. Identify risks in relation to objectives (human perspective). 
2. Identify the biggest risk. 
3. Consider the full range of probabilities. 
4. Use the best available information. 
5. Take a holistic view (i.e., consider all relevant factors). 
6. Be explicit about value judgements.  

Sutton argues that the second and third of King’s principles stress “the importance of paying specific attention to high-impact events. Even if their likelihood is considered low”. So-called HILL (High Impact Low Likelihood) events, such as climate tipping point activations, “are a central concern in risk management”, but only play a minor role in reports and are absent from climate projections and scenarios. Their designation of low-likelihood events is debatable (unknown likelihood seems more appropriate) and their absence from climate scenarios removes them from the view of risk managers. 

Take all plausible developments into account

To translate this to the water sector, as a consequence, part of the risk spectrum is outside our sector’s field of view. It does not see and therefore cannot anticipate, climate “surprises” (key principles 3 and 5). In addition to this, the end users of climate projections and scenarios in the water sector may be tempted to stick to the mean/median climate projections for any particular Shared Socioeconomic Pathway (SSP). In doing so, the spread and uncertainty in the development of the global human society (i.e., the SSPs) is effectively considered to represent the complete spread of possible climate outcomes. This is obviously incorrect. All SSP climate projections by the IPCC come with a (relatively wide) confidence interval that reflects the variability between climate models and some national institutes do the same. If a model on either extreme of the distribution turns out to be closest to the real climate system, the mean or median projections are actually quite far off the true climate development. 

We also need to consider the following: 1) the climate science community’s understanding of and capacity to comprehensively model all aspects of our climate system on all relevant scales is impressive but nevertheless incomplete; 2) models still have difficulties in accurately predicting weather extremes associated with climate change. Therefore, the model uncertainty cannot be trusted to represent the full epistemic uncertainty. So, again, the water sector cannot see the risks associated with the full epistemic uncertainty of the climate projections). Therefore, it cannot anticipate the more marginal (but not necessarily irrelevant) climate projections (key principle 3).  

Domain experts should decide on risk appreciation of scenarios

One can ask oneself how much this matters. In the end, the scenarios that are put forward by the climate science community (IPCC, national institutes) are truly the best that current science has to offer. But it is the risk manager that needs to decide which risks are acceptable, which not and whether to take a risk-averse or a more risk-tolerant approach to climate change adaptation. For that, she needs to know all the risks (as put forward by King). More generally, this also impacts on the decision of how to deal with risk, e.g., whether robustness is the way to go (which may be acceptable when there is significant confidence on the bounds of the projections), or whether flexibility and adaptability is a preferable strategy. These decisions have such a strong domain component that it is really the domain experts that need to decide on the risk appreciation.  

Sutton’s message is one that the water sector should heed. The key principles put forward by King et al. should be embraced by our industry. Yet, there is more. Recent years have shown an increasing awareness of multi-hazards, consecutive disasters, compound effects, etc. (e.g., reported here). Managing these risks requires a yet more elaborate and comprehensive approach to climate change risk management (see, e.g., Simpson et al., 2023).  

More about climate scenarios, tipping points and extremes is written in the recent Trendalert Klimaatscenario’s, -kantelpunten en -extremen (climate scenarios, tipping points and extremes), available to organisations that are part of Joint Research Programme or Dutch Water Sector Iintelligence.