I believe in climate change. I am very concerned, however, that the projections of (the current generation of) global and regional climate models are not hydrologically believable. Comparisons of their predictions with the normal control period (1961-1990) show that they often exhibit strong bias. They are inadequate predictors of the control period in many parts of the world, even where there are ensemble predictions of regional scale, dynamic downscaling models. In the UK, the UKCP09 outputs at show many different aspects of the predicted changes at 25km scale. They do not show any comparison of the predictions with the control period (despite posing an official question asking for these to be made available. Can I encourage others to log on and pose the same question!). Indeed, the UKCP09 weather generator (for which predicted realizations of weather at the 5 km scale can be obtained, uses the regional future climate projections only by making local bias corrections and RCM model derived change factors into the future. We expect, of course, that with more research money devoted to climate modelling, more computer power devoted to climate modelling at finer grid scales, better land surface parameterisations in the models (that is the hydrology and hydrologists should still wince about how it is being represented!), and improved understanding of other process representations in the models, the projections of the next generation of climate models might well be better. But in the meantime, there is an awful lot of research time, effort and money being devoted to impact studies based on the projections of the current generation of models. The question is whether any of this work is fit for the purpose of adapting to, or managing for, the future?
Let us assume for the sake of argument (and to provoke a response) that it is not. I should not then consider the projections of climate models to be an adequate basis for impact studies (so that a lot of research time, effort and money is being wasted). This is not to deny that there might be an anthropogenic effect on climate. I believe in climate change. I am also worried about the possibility that the climate system, as nonlinear dynamic system, might be subject to mode of behaviour shifts instigated by variability that is not being predicted by the current generation of GCMs. We know that there have been rapid modal shifts in the past, before any significant anthropogenic greenhouse gas inputs to the atmosphere. That suggests that we should plan to adapt to the possibility of change, despite the fact that we might have little faith in climate model projections. How should we then proceed?
The wrong reaction is to do nothing just because the climate projections have little credibility (or large uncertainty). It would be better to be precautionary by taking action. The question is how to be precautionary, given a lack of believable impact predictions? This depends on how risk averse or risk accepting we are prepared to be. Being risk averse will generally require more expensive measures than being risk accepting. But we can consider how expensive the required adaptation might be for different scenarios of future change, more or less extreme, quite independently of any climate model projections. In that way it is possible to plan a response to different magnitudes of change in terms of costs (and benefits).
Ideally we would wish to evaluate the probability associated with each magnitude of change. UKCP09 is presented in this way, with quantiles of change factors for various model predicted variables mapped across the country, conditional on assumed emissions scenarios. It is important, however, to remember that these projections are not a representation of the odds of climate actually turning out that way – they are rather the empirical probabilities of the ensemble model projections (with some Gaussian interpolation to compensate for the limited number of ensemble members in a high dimensional model space). This difference is important. Making use of these probabilities is to treat them as if the model was correct and the range of potential outcomes was complete. This is not the case.
To ignore those probabilistic estimates and deal with the magnitudes of change factors directly (without the need for climate simulations) therefore precludes a complete risk-based strategy but places the focus directly on what is considered to be affordable in being precautionary.
A particular case in point is protection against flooding. If a changing climate is intensifying the hydrological cycle we expect the frequency of floods of a given magnitude to be changing (even if, given the nature of extremes this has proven to be difficult to demonstrate from the available observations). There have been a number of studies that have invoked the change factors produced by climate models to examine how flow frequencies might change. This is straightforward to do so if it can be assumed that the parameters calibrated to represent catchment response might not change with changing inputs. It is much more difficult to do so if it is thought that the change in inputs or land use and management might require that parameters sets be changed to represent new sets of conditions.
But it is known that climate models do rather poorly in representing extremes, particularly of rainfalls, under control period conditions. They get the wrong result, and are known to get the wrong result, presumably for the wrong reasons (whether that be the result of scale effects, sub-grid rain, snow and cloud parameterisations, the simplicity of land surface parameterisations, inadequate representations of heat exchange with the oceans, etc). What is clear, however, is that we should not be assuming stationarity in estimating flood characteristics and we therefore need to plan for change.
A number of strategies are possible so as not to exacerbate the problem: avoiding new developments on flood plains; improving flood defences; flood proofing of existing buildings; breaching of existing defences to make more storage; building flood detention basins. In most cases these solutions will be robust in the sense of not precluding future adaptive management strategies but they all have a greater or lesser cost. So what is the cost-benefit of protecting against different levels of change. How precautionary are we prepared to pay to be?
This is, essentially a political decision. The science comes in estimating costs and benefits rather than in estimating the magnitudes of change. Even if some of the evident problems of the current generation of climate models will be (hopefully) less apparent in the next generation, the path towards have a realistic model still seems long and tortuous. So should we continue to do local bias corrections and use change factors in impact studies just because the funders of research and decision makers are asking for “evidence” of how great the impacts of change might be; or should we change the nature of the game into something more overtly political before the “evidence” becomes to be seen as based on insubstantial foundations.
In case you are worried about your current research funding, this need not lead to less impact studies. Indeed a wider range of potential outcomes might need to be considered rather than the just the latest grand ensemble of predicted change factors. It is just that the evidence does not now depend on climate models but rather on the range of potential future conditions that decision makers want to consider.
Are there any difficulties in this approach? Yes. Climate models provide projections in space and time constrained by energy, momentum and mass balances. These projections are consistent in so far as the approximations of the numerical solutions allow. It could be argued therefore that any study of the potential impacts of change that only supposes the magnitude of change and consequent impacts will be inherently subjective, unscientific and providing inadequate evidence. This would be to totally misunderstand the arguments presented above. However, there is an associated problem of providing suitable scenarios for patterns of change factors in catchments that are complex in their patterns of precipitations and other characteristics (we might note that this is already an issue in current practice, since the need to make (often large) bias corrections in the local application of climate change projections is already inconsistent with the energy, mass and momentum balances of the original models).
So we could still accept that climate model projections (together with any necessary bias corrections) are just one way of producing plausible patterns of change factors into the future, but then modify the patterns of change factors in assessing costs and benefits for precautionary action. In doing so, we should also take account of any uncertainty in the impact modelling. Where the disbenefits are a nonlinear function of the projected change this might make an important difference to the decision that might be made.
Do we reduce the strength of the arguments to induce a reaction in politicians to mitigate the effects of climate change by such a strategy? Perhaps – even if they might believe in climate change, the potential costs of adaptation might be considered politically unacceptable, particularly in a time of economic recession and increasing unemployment. Governments have made that argument in the past, most notably both Bush administrations in the USA. However, to do nothing is then to be risk accepting (or even irresponsible) to a possibly dangerous degree.
In addition, there are other factors that might also affect future hydrological responses (urbanisation, agricultural intensification, deforestation/afforestation, river training and re-naturalisation,…..). There are certainly model-based predictions of the effects of potential changes in different factors, mostly deterministic in nature, even though we know that process representations of such factors are subject to considerable uncertainty. In this case, the sensitivity of response to assumed future change is generally evaluated (also, like the climate case, a form of scenario analysis) but mostly without an assessment of the cost of possible adaptation strategies. Such changes could also be evaluated in the form of the cost-benefit strategy to be precautionary suggested here.
People will not agree about how to assess appropriate costs and benefits for different types of impact and mitigation strategies, particularly in respect of future socioeconomic scenarios affecting future risk assessment. More science and understanding is required to reduce the uncertainties in doing so. That does not, however, preclude the use of such an approach. I believe in climate change, and the potential impacts of other catchment changes, but I would suggest that we need better, and more scientifically honest, ways of deciding how precautionary to be in planning for the future.
Keith Beven
Lancaster Environmental Centre, Lancaster University, Lancaster LA1 4YQ, UK
Email:
A Report of “CCN at Hydropredict 2010” can be found I’ve not only come across great ideas in these chats, i’ve connected with like-minded teachers across the world and learned more about what’s working for them and their students
