Attributing climate extremes to anthropogenic climate change and land–atmosphere coupling

Earth’s Climate has always affected mankind. On the other hand, humans started to affect Earth’s climate. The consumption of fossil fuels leads to higher global temperatures and more climate extremes. Extreme events, such as heat waves or droughts, can have adverse effects on societies and ecosystems. An improved understanding of their drivers helps to mitigate their impacts. However, extreme events are rare and changes in their frequency are hard to detect as natural variability is superimposed on possible trends. The land surface is an integral part of the climate system and is coupled to the atmosphere via the exchange of water and energy. Thereby, soil moisture conditions can influence the partitioning of the available surface net radiation into the latent and sensible heat flux. When the soil is sufficiently wet, the latent heat flux is controlled and limited by radiation, whereas under drought conditions it is limited by moisture availability. Hence, dry soils tend to decrease the latent heat flux, thereby increasing the sensible heat flux, exacerbating hot extremes during heat waves. In this thesis, I investigate the influence of past climate change and of the land surface on climate extremes. The importance of climate change for extreme events can be quantified with probabilistic event attribution. This method compares the probability of an extreme event in the current climate to its probability in a counterfactual climate absent anthropogenic influence. The effect of the land conditions on the atmosphere is assessed with numerical sensitivity experiments prescribing soil moisture in global climate models. This decouples the land from the atmosphere and therefore allows to infer the effect of soil moisture on climate. In the first part of this thesis I investigate the influence of climate change on climate extremes. Using the 2015 meteorological summer drought in Europe as test case, I assess the importance of methodological choices for probabilistic event attribution. The summer of 2015 was characterized by record low rainfall and river levels. Using a wide range of observations and model simulations I show that the conclusion whether this event is related to anthropogenic climate change heavily depends on methodological decisions. Our results highlight that uncertainties of event attribution studies are larger than commonly assumed, implying a high risk of erroneous attribution statements. In the second part of the thesis I investigate the influence of the land surface on climate extremes. I contrast four different methods to prescribe soil moisture, and compare the ramifications of prescribing the mean versus the median soil moisture climatological seasonal cycle. I show that soil moisture prescription methods that alter the soil ice content lead to large, undesirable, ground heat flux anomalies in the employed land model. Therefore, I developed a novel approach to prescribe soil moisture that prevents the ground heat flux anomalies, while still muting the land–atmosphere coupling. Furthermore, I find that the median soil moisture climatology is lower than the mean for most regions and that prescribing the mean climatology leads to larger temperature anomalies. However, prescribing soil moisture does not conserve water, and I show how the atmospheric temperature response is directly related to the amount of artificially added and removed water, which implies that not the entire temperature signal can be attributed to land–atmosphere coupling. In the third part of the thesis I investigate the joint influence of climate change and the land surface on climate extremes. In 2010 western Russia experienced a severe heat wave, which led to record-high temperatures. It was hypothesized that climate change and land–atmosphere interactions were important for the build-up of the observed temperature extremes. To test this hypothesis, I combine the approaches introduced in the first two parts of this thesis and I conducted dedicated climate model simulations that allow to disentangle to effect of recent climate change from that of dry soils. I find that climate change alone has increased the probability of the event threefold since the 1960s. Prescribing the soil moisture conditions observed during the event indicates that dry soils made the event six times more likely. Finally, the joint effect of the soil moisture anomaly and climate change leads to a 13-fold increase in the event probability. In two co-authored articles presented in the appendix, I briefly showcase two further methods to investigate the influence of climate change and the land surface on extremes. The first uses trend detection and attribution to identify the drivers behind the trend in the growing season length, a climate index indicating conditions favorable for plant growth. We show that in extratropical regions where some of the world’s major crops are grown, the observed increase in the growing season length can be attributed to anthropogenic climate change. The second analysis illustrates how land–temperature coupling metrics can be used to assess the importance of soil moisture for heat waves. Using two different coupling metrics we illustrate how dry soils exacerbated the extreme temperatures during the 2015 heat wave in the Western United States. Overall, the presented findings enhance the understanding of the role of climate change and land–atmosphere interactions for climate extremes. They highlight the importance of climate change for extreme events, but also demonstrate large uncertainties in some of the assessments. The results further underline the relevance of land–atmosphere interactions for aggravating temperature extremes during heat waves. The presented scientific advances emphasize the usefulness of combining different methodological frameworks, and serve as an example of fruitful collaboration across research communities