Adaptation Actions for a Changing Arctic: Perspectives from the Barents Area

104

Adaptation Actions for a Changing Arctic: Perspectives from the Barents Area

Appendix 4 . 1 Knowledge, information and uncertainties Everybody deals with some degree of uncertainty every day, as nobody knows how the day will pan out when they get out of bed in the morning.The term“uncertainty” is also commonly used, e.g. in the assessment reports of the IPCC, however, its exact definition is not always clear. It may mean different things to different people. Within the scientific community, it may embody aspects such as model shortcomings, known unknowns, unknown unknowns, lack of information, lack of knowledge, lack of understanding, lack of precision, probabilities, error bars, and errors. Uusitalo et al., (2015) list a number of approaches for dealing with uncertainties: (i) expert assessment, (ii) model sensitivity analysis, (iii) model emulation, (iv) variability, (v) multiple models, and (vi) data based approaches. Several of these approaches are present in this chapter: The contribution from a number of different experiments brings in the first approach, and the use of different emission scenarios (global drivers, socio-economic futures), large ensembles, and different downscaling strategies (ESD and RCMs) introduce both a kind of sensitivity analysis (ii) and make use of multiple models (v).The comparison with past trends also serves to address uncertainty, e.g. in terms of variability (iv) and data based approaches (vi). A key issue concerning uncertainty is a proper validation of the models being used to address some question, and this is part of some of the projections presented in this chapter.

by also including a description of alternative outcomes and the likelihood associated with each. For a quantitative statement (e.g. temperature or precipitation), a small set of categories can cover all possibilities: a reduction, no change, or increase. The number of categories reflects the confidence and the amount of details that can be provided. Unlike quantitative forecasts, it is likely that a range of qualitative scenarios do not cover all possible futures (unknown unknowns). In this case, the stakeholder needs to ask ‘what can happen?’ and assess which conditions are important and which are not by applying a sensitivity test for impact models in a “bottom- up approach” (Pielke Sr. and Wilby, 2012). Such a sensitivity test needs to go hand-in-hand with an uncertainty analysis, and can be carried out in terms of downscaling applied to large ensembles of GCM runs (different emission scenarios, models, and initial conditions) in addition to testing impact models with a range of different inputs. The results of such a sensitivity test can be used to create a risk contingency table and guide contingency plans, but experience has shown that they may not always be complete (e.g. the tsunami that hit Fukishima in 2011). Uncertain outcome can also be stated as risk analysis based on statistical estimation of probabilities.The risk-based approach can include alternative conditions as well as a combination of events (a “perfect storm”), providing information about what can happen and what is likely.Most of this report has dealt with the respective drivers without much emphasis on the fact that many are likely to take place in combination with others.This is known to be true with respect to temperature, precipitation, permafrost, and ice, although ecological response and socio- economic pressures will also be part of this mix. Because of this complex nature, confidence statements as presented in the IPCC reports, which are also useful for describing the degree of belief in a statement, do not necessarily provide information about alternative types of outcomes in terms of more complex situations. Confidence intervals (statements) for each factoid can even be estimated objectively through statistical estimation of confidence intervals,however,they are not so easy to combine for a complexmix of conditions whichmay reinforce or weaken the effect of each other.There are also likely surprises which by definition are difficult to anticipate, partly due to non-linear and convoluted interactions between different drivers. They may also involve tipping points or unknown unknowns. There are different parallel sources of information available for assessing different questions. For instance, there are few known laws of nature when it comes to socio-economic questions, as opposed to scientific disciplines such as physics and chemistry. Nevertheless, demography is fairly straightforward in terms of births and deaths, although migration patterns is a major unknown. Both socio-economic and physics- related assessments can involve statistics and empirical evidence, however, an additional unknown is that there is no immediate and direct links between a changing climate and societal changes. Furthermore, there is an increasingly faster pace of change in the Arctic and in the rest of the world, in terms of politics, economy, climate change, and with many intertwined ties between these. This makes it tricky to rely on

The discussion of uncertainty often implies the presence of some information, e.g. about the limitation of our ability to predict or the characteristics of the probability distribution function. The character of the type of uncertainty connected to some question is furthermore contextual, and it is difficult to put all in one basket (Uusitalo et al., 2015). In many cases, much of the uncertainty concerns aspects which are not crucial, and we may arrive at an answer by starting with the information that is available, and apply known constraints to infer a range of possibilities (elimination). Furthermore, the picture of the situation about which we wish to study may become more complete by bringing together information from different and independent sources and applying proper statistics. This also includes studying data from different sites across regions, e.g. at different sides of national borders. Some of the analysis presented in this report builds on multiple lines of evidence from a set of independent station observations. The ESD presented here made use of a group of station within the Barents region and combined the information from all these to improve the quality and reduce the effect of errors (Benestad et al., 2015). The degree of confidence and the uncertainty often has to be explained in even simpler terms to stakeholders, while being based on the principles above, and some sketchy ideas could be based on a table of possible outcomes where the likelihood for each is assessed, based on scientific results and findings.This is essentially similar to a simple risk analysis or risk management that makes use of a contingency table.This type of presentation may provide a common format across a range of situations, both where the information is quantitative and qualitative. Such tables may provide a more intuitive description than the description provided in the IPCC reports