A Multivariate Conditional Probability Ratio Framework For The Detection And Attribution Of Compound Climate Extremes

Most attribution studies tend to focus on the impact of anthropogenic forcing on individual variables. However, studies have already established that many climate variables are interrelated, and therefore, multidimensional changes can occur in response to climate change. Here, we propose a multivariate method which uses copula theory to account for underlying climate conditions while attributing the impact of anthropogenic forcing on a given climate variable. This method can be applied to any relevant pair of climate variables; here we apply the methodology to study high temperature exceedances given specified precipitation conditions (e.g., hot droughts). With this method, we introduce a new conditional probability ratio indicator, which communicates the impact of anthropogenic forcing on the likelihood of conditional exceedances. Since changes in temperatures under droughts have already accelerated faster than average climate conditions in many regions, quantifying anthropogenic impacts on conditional climate behavior is important to better understand climate change.Plain Language Summary Most studies investigating human impacts on climate conditions focus on characterizing changes in individual variables such as precipitation or temperature. However, since many climate conditions are interconnected, these individual variables do not comprehensively represent the many changes that can occur in response to human activity. Here, we introduce a method that takes into account underlying climate conditions while quantifying the impact of human activity on a given climate variable. This method can be used to study pairs of climate variables and here we provide an example application to examine high temperature occurrences during dry precipitation conditions using climate models. For example, we show that regions such as the Amazon have a 4.1 times higher likelihood of experiencing high temperatures under dry climate conditions as a result of human activity. Given our knowledge of future climate change, we anticipate that the relationships between key climate variables may continue to change, which makes the study of human impacts on conditional climate behavior important for a more complete understanding of climate change.