Evaluating the dependence structure of compound precipitation and wind speed extremes

Abstract

Abstract. Estimating the likelihood of compound climate extremes such as concurrent drought and heatwaves or compound precipitation and wind speed extremes is important for assessing climate risks. Typically, simulations from climate models are used to assess future risks, but it is largely unknown how well the current generation of models represents compound extremes. Here, we introduce a new metric that measures whether the tails of bivariate distributions show a similar dependence structure across different datasets. We analyse compound precipitation and wind extremes in reanalysis data and different high-resolution simulations for central Europe. A state-of-the-art reanalysis dataset (ERA5) is compared to simulations with a weather model (Weather Research and Forecasting – WRF) either driven by observation-based boundary conditions or a global circulation model (Community Earth System Model – CESM) under present-day and future conditions with strong greenhouse gas forcing (Representative Concentration Pathway 8.5 – RCP8.5). Over the historical period, the high-resolution WRF simulations capture precipitation and wind extremes as well as their response to orographic effects more realistically than ERA5. Thus, WRF simulations driven by observation-based boundary conditions are used as a benchmark for evaluating the dependence structure of wind and precipitation extremes. Overall, boundary conditions in WRF appear to be the key factor in explaining differences in the dependence behaviour between strong wind and heavy precipitation between simulations. In comparison, external forcings (RCP8.5) are of second order. Our approach offers new methodological tools to evaluate climate model simulations with respect to compound extremes.

Assessing the dependence structure between oceanographic, fluvial, and pluvial flooding drivers along the United States coastline

“…In comparing dependence structures derived from model and observational data we followed the methodology in Zscheischler et al (2021). Results showed that for pairs S-Q and S-P the tail dependence derived from models is very similar to that derived from observations in the Gulf of Mexico.…”

“…We apply the Kullback-Leibler (KL) divergence to assess significance in the difference in tail dependence derived from the two types of data. We provide a brief description of the methodology (see Zscheischler et al (2021), Vignotto et al (2021), and references therein for more details). For two bivariate distributions 2021) we choose the 'minimum' corresponding to r( ) = min ( 1 , 2 ) , with = ( 1 , 2 ) as it covers both asymptotically dependent and independent data.…”

“…This is because two bivariate distributions with equal (or very similar) tail dependence coefficients may still vary in their dependence structure and this cannot be assessed by just calculating the difference in χ. The reason is that χ only focuses on the diagonal (Zscheischler et al, 2021), whereas KL divergence partitions the extremal space defined by the risk function (above the selected threshold) into a number of sets and thus better captures the dependence structure. We repeat the analysis again for the tropical and extra-tropical seasons to assess whether models perform better in one of them when tail dependence is compared to observations.…”

“…Our third and final objective is to compare the dependence structures of different combinations of flooding drivers derived from observations to those derived from model hindcast/reanalysis data. Comparing dependence structures across different datasets is something only very few studies have addressed to date (Paprotny et al, 2020;Ganguli et al, 2020;Zscheischler et al, 2021). This last analysis step will show how well models capture dependence structures between flooding drivers and identify the pairs of drivers and locations where model results overestimate or underestimate the dependence.…”

“…Typical flooding driver combinations that were previously assessed include: surge and discharge (e.g., Moftakhari et al, 2017); surge and precipitation (e.g., Wahl et al, 2015); surge and waves (e.g., Marcos et al, 2019); surge, discharge, and precipitation (e.g., Svensson and Jones, 2002;2004); surge, waves, and discharge (e.g., Petroliagkis et al, 2016); and surge, waves, discharge, and precipitation (e.g., Hawkes and Svensson, 2006;Camus et al, 2021). Many studies were performed using observational data (e.g., Wahl et al, 2015;Ward et al, 2018), while some included model hindcast data (Marcos et al, 2019;Couasnon et al, 2020;Camus et al, 2021), and very few included both or compared different datasets (e.g., Paprotny et al, 2020;Ganguli et al, 2020;Zscheischler et al, 2021). For the CONUS coastline, two previous studies assessed compound flooding potential at the continental scale (while the CONUS was also included in global scale assessments): Wahl et al (2015) analysed storm surge and precipitation and Moftakhari et al (2017) analysed storm surge and discharge.…”