Case for support: Climate futures science for adaptation

The Climate Futures Information Gap

There is growing demand for information about future climate change, to prepare for unavoidable changes in risk. Climate modelling and climate impacts research generate many simulations of future climate and impacts on crops, water resources, infrastructure, and beyond. Whilst there has been impressive progress in modelling, there is an increasing recognition that these simulations alone are insufficient to inform decision-making. There has been critique of a “top down” approach whereby data and model outputs are provided to decision-makers, with little understanding for their decision-making context, meaning the data provided may not be relevant or usable [1]. There is a growing emphasis on “decision-led” approaches [1], involving iterative engagement with decision- makers, to understand their priorities and processes, pinpoint where climate change is important, and assess whether potential adaptation options are climate resilient.

These “decision-led” approaches to climate services show promise, however there is one challenging step in the process which warrants greater attention. In any effort to inform decision- making about future climate, it is necessary to characterise and communicate future changes in hazard (shown in purple in Fig. 1). Climate models can be used to explore change, however their future simulations cannot be validated. They can be used to test what might happen under certain conditions, but they cannot make deterministic predictions. It is also difficult, some argue impossible, to evaluate the probability of a future climate model projection [2]. Therefore, in using model projections to describe future risk, scientists or climate service practitioners must make important and difficult choices about how to characterise and communicate future change.

Current options for characterising future climate change include taking an average across an “ensemble” of different climate model projections (“ensemble mean”), identifying the outcomes which are most common among different models (“consensus”), quantifying the range of modelled changes, or providing many future scenarios for “stress-testing” adaptation options [1,3]. Given model biases, there are also efforts to “constrain” model ensembles, or remove less credible projections, using evaluation against observations to weight the results, or searching for features of current climate which relate to future change to apply an “emerging constraint” [4]. In any of these approaches, future outcomes are primarily determined by the models being included. Much analysis is based on “multi-model ensembles” which include state-of-the-art models, but do not systematically explore uncertainty. Due to biases, imperfect representation of important processes, and natural variability, it is possible that the real future will lie outside the modelled range [5].

Current options for communicating future climate change include information products, often using timeseries, boxplots, and maps to visualise change, and processes of communication, including training and engagement. Making this communication useful for decision-making is challenging. It can be difficult to relate the characterisation of modelled change (a modelled mean or range) with what might happen in the real world. Visualisations and explanations may be interpreted differently by scientists and decision-makers [6]. Climate services experts have developed approaches to foster shared understanding, including “climate risk narratives” [7] and co-production of information [8]. These methods have led to progress through long engagements with specific decision-makers, but it is difficult to identify lessons for wider audiences.

Without innovation to better characterise and communicate future change – to address the “climate futures information gap” – there is a danger that climate services will fail to account for climate change risks. There are increasing efforts to use climate information to inform adaptation, but if decisions are based on flawed model projections or misguided interpretations, the resulting development may not be resilient to future change. For example, in East Africa, many climate models project future increases in rainfall [9]. Characterisation and communication of future climate based on these models could lead to planning which relies on more rainfall and does not account for the real risk of a drier conditions, for example expanding reliance on hydropower.

Countries such as the UK, the Netherlands, and Switzerland have made substantial investments in national climate scenarios [10,11,12], allowing for focused attention on characterising and communicating future change. However, in general there is insufficient resource for this “step” in the climate services process, particularly in the Global South. Climate scientists tend to focus on challenges in climate model analysis, and may feel uncomfortable determining how the results are used. Meanwhile climate services practitioners need to investigate decision contexts [13], where there are so many urgent priorities that climate change can seem a minor aspect of the problem.

The “climate futures information gap” is therefore under-resourced and under-appreciated, with potential implications for the resilience of adaptation planning, as well as the reputation of climate science and climate services. This challenge demands innovation and a dedicated area of research: a new climate futures science.

References

[1] Weaver et al. (2013) WIRES Climate Change, https://doi.org/10.1002/wcc.202

[2] Dessai & Hulme (2004) Climate Policy, Journal of Climate Change, https://doi.org/10.1080/14693062.2004.9685515

[3] Knutti et al. (2010) Journal of Climate, https://doi.org/10.1175/2009JCLI3361.1

[4] Hall et al. (2019) Nature Climate, https://doi.org/10.1038/s41558-019-0436-6

[5] Stainforth et al. (2007) Phil Trans A, https://doi.org/10.1098/rsta.2007.2073

[6] Harold et al. (2016) Nature Climate Change, https://doi.org/10.1038/nclimate3162

[7] Jack et al. (2020) Climate Risk Management, https://doi.org/10.1016/j.crm.2020.100239

[8] Bremer & Meisch (2017) WIRES Climate Change, https://doi.org/10.1002/wcc.482

[9] Rowell et al. (2015) Journal of Climate, https://doi.org/10.1175/JCLI-D-15-0140.1

[10] UK Climate Projections https://www.metoffice.gov.uk/research/approach/collaboration/ukcp/index

[11] Climate Scenarios for the Netherlands http://www.climatescenarios.nl/

[12] Swiss Climate Change Scenarios http://ch2011.ch/en/index.html

[13] Guido et al. (2019) Climate and Development, https://doi.org/10.1080/17565529.2019.1630352