ALBERT

Description: Partitioning uncertainty in projections of future climate change into contributions from internal variability, model response uncertainty and emissions scenarios has historically relied on making assumptions about forced changes in the mean and variability. With the advent of multiple single-model initial-condition large ensembles (SMILEs), these assumptions can be scrutinized, as they allow a more robust separation between sources of uncertainty. Here, the framework from Hawkins and Sutton (2009) for uncertainty partitioning is revisited for temperature and precipitation projections using seven SMILEs and the Coupled Model Intercomparison Project CMIP5 and CMIP6 archives. The original approach is shown to work well at global scales (potential method bias 

Description: Robust appraisals of climate impacts at different levels of global-mean temperature increase are vital to guide assessments of dangerous anthropogenic interference with the climate system. The 2015 Paris Agreement includes a two-headed temperature goal: "holding the increase in the global average temperature to well below 2 °C above pre-industrial levels and pursuing efforts to limit the temperature increase to 1.5 °C". Despite the prominence of these two temperature limits, a comprehensive overview of the differences in climate impacts at these levels is still missing. Here we provide an assessment of key impacts of climate change at warming levels of 1.5 °C and 2 °C, including extreme weather events, water availability, agricultural yields, sea-level rise and risk of coral reef loss. Our results reveal substantial differences in impacts between a 1.5 °C and 2 °C warming that are highly relevant for the assessment of dangerous anthropogenic interference with the climate system. For heat-related extremes, the additional 0.5 °C increase in global-mean temperature marks the difference between events at the upper limit of present-day natural variability and a new climate regime, particularly in tropical regions. Similarly, this warming difference is likely to be decisive for the future of tropical coral reefs. In a scenario with an end-of-century warming of 2 °C, virtually all tropical coral reefs are projected to be at risk of severe degradation due to temperature-induced bleaching from 2050 onwards. This fraction is reduced to about 90 % in 2050 and projected to decline to 70 % by 2100 for a 1.5 °C scenario. Analyses of precipitation-related impacts reveal distinct regional differences and hot-spots of change emerge. Regional reduction in median water availability for the Mediterranean is found to nearly double from 9 % to 17 % between 1.5 °C and 2 °C, and the projected lengthening of regional dry spells increases from 7 to 11 %. Projections for agricultural yields differ between crop types as well as world regions. While some (in particular high-latitude) regions may benefit, tropical regions like West Africa, South-East Asia, as well as Central and northern South America are projected to face substantial local yield reductions, particularly for wheat and maize. Best estimate sea-level rise projections based on two illustrative scenarios indicate a 50 cm rise by 2100 relative to year 2000-levels for a 2 °C scenario, and about 10 cm lower levels for a 1.5 °C scenario. In a 1.5 °C scenario, the rate of sea-level rise in 2100 would be reduced by about 30 % compared to a 2 °C scenario. Our findings highlight the importance of regional differentiation to assess both future climate risks and different vulnerabilities to incremental increases in global-mean temperature. The article provides a consistent and comprehensive assessment of existing projections and a good basis for future work on refining our understanding of the difference between impacts at 1.5 °C and 2 °C warming.

Description: The intensity of precipitation events is expected to increase in the future. The rate of increase depends on the strength or rarity of the events; very strong and rare events tend to follow the Clausius–Clapeyron relation, whereas weaker events or precipitation averages increase at a smaller rate than expected from the Clausius–Clapeyron relation. An often overlooked aspect is seasonal occurrence of such events, which might change in the future. To address the impact of seasonality, we use a large ensemble of regional and global climate model simulations, comprising tens of thousands of model years of daily temperature and precipitation for the past, present, and future. In order to make the data comparable, they are quantile mapped to observation-based time series representative of the Aare catchment in Switzerland. Model simulations show no increase in annual maximum 1-day precipitation events (Rx1day) over the last 400 years and an increase of 10 %–20 % until the end of the century for a strong (RCP8.5) forcing scenario. This fits with a Clausius–Clapeyron scaling of temperature at the event day, which increases less than annual mean temperature. An important reason for this is a shift in seasonality. Rx1day events become less frequent in late summer and more frequent in early summer and early autumn, when it is cooler. The seasonality shift is shown to be related to summer drying. Models with decreasing annual mean or summer mean precipitation show this behaviour more strongly. The highest Rx1day per decade, in contrast, shows no change in seasonality in the future. This discrepancy implies that decadal-scale extremes are thermodynamically limited; conditions conducive to strong events still occur during the hottest time of the year on a decadal scale. In contrast, Rx1day events are also limited by other factors. Conducive conditions are not reached every summer in the present, and even less so in the future. Results suggest that changes in the seasonal cycle need to be accounted for when preparing for moderately extreme precipitation events and assessing their socio-economic impacts.

Description: Bias adjustment is often a necessity in estimating climate impacts because impact models usually rely on unbiased climate information, a requirement that climate model output rarely fulfills. Most currently used statistical bias adjustment methods adjust each climate variable separately, even though impacts usually depend on multiple, potentially dependent variables. Human heat stress, for instance, depends on temperature and relative humidity, two variables that are often strongly correlated. Whether univariate bias adjustment methods effectively improve estimates of impacts that depend on multiple drivers is largely unknown, and the lack of long-term impact data prevents a direct comparison between model output and observations for many climate related impacts. Here we use two hazard indicators, heat stress and a simple fire risk indicator, as proxies for more sophisticated impact models. We show that univariate bias adjustment methods such as univariate quantile mapping often cannot effectively reduce biases in multivariate hazard estimates. In some cases, it even increases biases. These cases typically occur (i) when hazards depend equally strongly on more than one climatic driver, (ii) when models exhibit biases in the dependence structure of drivers, and (iii) when univariate biases are relatively small. Using a perfect model approach, we further quantify the uncertainty of bias-adjusted hazard indicators due to internal variability and show how inadequate bias adjustment can amplify this uncertainty. Both issues can be addressed successfully with a statistical bias adjustment that corrects the multivariate dependence structure in addition to the marginal distributions of the climate drivers. Our results suggest that currently many modelled climate impacts are associated with substantial uncertainties associated with the choice of bias adjustment. We conclude that, as long as bias adjustment is unavoidable for climate impact assessments, the use of statistical bias adjustment approaches that correct the multivariate dependence structure of drivers is preferable.

Description: Bias adjustment is often a necessity in estimating climate impacts because impact models usually rely on unbiased climate information, a requirement that climate model outputs rarely fulfil. Most currently used statistical bias-adjustment methods adjust each climate variable separately, even though impacts usually depend on multiple potentially dependent variables. Human heat stress, for instance, depends on temperature and relative humidity, two variables that are often strongly correlated. Whether univariate bias-adjustment methods effectively improve estimates of impacts that depend on multiple drivers is largely unknown, and the lack of long-term impact data prevents a direct comparison between model outputs and observations for many climate-related impacts. Here we use two hazard indicators, heat stress and a simple fire risk indicator, as proxies for more sophisticated impact models. We show that univariate bias-adjustment methods such as univariate quantile mapping often cannot effectively reduce biases in multivariate hazard estimates. In some cases, it even increases biases. These cases typically occur (i) when hazards depend equally strongly on more than one climatic driver, (ii) when models exhibit biases in the dependence structure of drivers and (iii) when univariate biases are relatively small. Using a perfect model approach, we further quantify the uncertainty in bias-adjusted hazard indicators due to internal variability and show how imperfect bias adjustment can amplify this uncertainty. Both issues can be addressed successfully with a statistical bias adjustment that corrects the multivariate dependence structure in addition to the marginal distributions of the climate drivers. Our results suggest that currently many modeled climate impacts are associated with uncertainties related to the choice of bias adjustment. We conclude that in cases where impacts depend on multiple dependent climate variables these uncertainties can be reduced using statistical bias-adjustment approaches that correct the variables' multivariate dependence structure.

Description: For sectors like agriculture, hydrology and ecology, increasing interannual variability (IAV) can have larger impacts than changes in the mean state, whereas decreasing IAV in winter implies that the coldest seasons warm more than the mean. IAV is difficult to reliably quantify in single realizations of climate (observations and single-model realizations) as they are too short, and represent a combination of external forcing and IAV. Single-model initial-condition large ensembles (SMILEs) are powerful tools to overcome this problem, as they provide many realizations of past and future climate and thus a larger sample size to robustly evaluate and quantify changes in IAV. We use three SMILE-based regional climate models (CanESM-CRCM, ECEARTH-RACMO and CESM-CCLM) to investigate downscaled changes in IAV of summer and winter temperature and precipitation, the number of heat waves, and the maximum length of dry periods over Europe. An evaluation against the observational data set E-OBS reveals that all models reproduce observational IAV reasonably well, although both under- and overestimation of observational IAV occur in all models in a few cases. We further demonstrate that SMILEs are essential to robustly quantify changes in IAV since some individual realizations show significant IAV changes, whereas others do not. Thus, a large sample size, i.e., information from all members of SMILEs, is needed to robustly quantify the significance of IAV changes. Projected IAV changes in temperature over Europe are in line with existing literature: increasing variability in summer and stable to decreasing variability in winter. Here, we further show that summer and winter precipitation, as well as the two summer extreme indicators mostly also show these seasonal changes.