Methods for event attribution

 

Event attribution is used to quantify how human-influenced climate change affects the occurrence of a particular type (or class) of extreme event. Its goals are similar to those of the detection and attribution process described in Chapter 2 (see Section 2.3.4), but it focuses on individual events. Event attribution analyses (NASEM, 2016) compare the likelihood of a particular class of events (e.g., all events as extreme, or more extreme, than the event defined in the study) between a factual world, which includes the human component, and a counter-factual world that comprises only natural factors — that is, the “climate that might have been” in the absence of the human component.

To demonstrate, Figure 4.21 shows distributions of possible values of a climate variable for the world without the human contribution in blue, and for a scenario like the one we have experienced with the human contribution in red. The shaded regions represent the probability that a particular extreme event (an outcome as extreme, or more so, than the one indicated by the vertical bar) will occur in each scenario. The probability of the event increases when the human contribution is included — from 1 in 60 to 1 in 5. The ratio of the probability with the human contribution to the probability without the human contribution is referred to as a “risk ratio.” Although this event could occur in the absence of human influence, it is 12 times as likely (risk ratio of 12) when the human component is included.

The conclusions of an event attribution analysis often depend on how the question is posed. This includes the choices made when defining events and designing the analysis approach. For example, the change in probability between the two scenarios in Figure 4.21 depends on the placement of the vertical bar, or the lower bound on the magnitude that defines the chosen event. Changes in the probabilities of temperature and precipitation extremes depend on the probability of the events in the current climate, with larger risk ratios corresponding to more extreme (rarer) events (Kharin et al., 2018). The uncertainty in the risk ratio (i.e., the event attribution result) becomes larger for rarer events, as it is more difficult to estimate the probabilities of these very rare events. The choice of the variable and/or region to determine the distributions also has an impact on the results.

Two types of questions have been asked in event attribution analyses: How has the probability of the extreme event (frequency) changed, and how has the intensity of the event (magnitude) changed? As an example, event attribution for a flood-producing heavy rainfall event may try to answer, “Has human-induced climate change made this type of heavy rainfall event occur more often?” (frequency) or “Has human-induced climate change increased the amount of rainfall in these types of storms?” (magnitude). The human influence could have a different impact on the frequency than on the magnitude of a particular event. It is thus important to understand the characteristics of the event being assessed and to interpret the results of an event attribution analysis in context.