L’objectif de mon travail de thèse est de contribuer à la compréhension des rôles joués par les interactions sol-atmosphère et par la circulation à grande-échelle dans la formation d’aanomalies de températures estivales en Europe. Cela constitue un double défi du fait de la rareté des observations d’une part et des incertitudes liées aux paramétrages des modèles atmosphériques d’autre part. Je me concentre sur 4 sujets principaux : 1) Comment les interactions sol-atmosphère influencent-elles les projections climatiques et leurs incertitudes ? 2) Comment les paramétrisations atmosphériques influencent les simulations des vagues de chaleur extrêmes ? 3) Quelle importance jouent l’humidité du sol et les circulations atmosphériques sur les anomalies de température estivales? Et 4) Quels sont les impacts des sécheresses et de la chaleur sur la végétation ?Concernant la première question j’ai montré que les différences de partitionnement des flux de chaleur conduisent à un réchauffement spatialement hétérogène et incertain en Europe dans l'avenir. En particulier en Europe centrale, les modèles prédisent une grande gamme de réchauffements en été, alors qu’un relatif accord est obtenu en Europe du Sud. J’ai montré que l'utilisation d’une contrainte sur l’ensemble de projections par des observations des flux de chaleur sensible permet de réduire cette incertitude future au niveau régional jusqu’à 40% dans cette région.En outre, en testant des physiques atmosphériques pour des conditions caniculaires, j’ai constaté une grande variabilité des températures entres toutes les configurations physiques. La température est principalement sous-estimée par rapport aux observations. Le rayonnement à ondes courtes et les précipitations sont généralement surestimées. J’ai sélectionné un sous-ensemble réduit de configurations pour simuler une hausse futures des températures estivales en Europe.Avec la meilleure configuration trouvée j’ai quantifié la contribution de l’humidité du sol au début d’été et les ‘drivers’ à grandes échelle sur les températures estivales. J’ai montré que la contribution de l’humidité du sol peut monter jusqu’à 1°C au maximum pendant les vagues de chaleurs de 2003 et de2010. La contribution des ‘drivers’ à grande échelle est la plus importante, jusqu’à 3°C en 2003 et jusqu’à 6°C en 2010. Toutefois, la contribution de l’humidité du sol en début d’été a augmenté au cours des dernières décennies en Europe centrale et en Russie, correspondant aux régions avec une tendance négative significative de l’humidité du sol. Les ‘drivers’ a grands échelles ont joué un plus grand rôle en Europe de l’Est.Enfin, j’ai étudié les effets des sécheresses et de la chaleur sur la végétation. J’ai trouvé une surestimation de la GPP simulée à l'échelle locale sur un site méditerranéen pendant l’été. Cela indique que le modèle de végétation ne peut pas actuellement reproduire les conséquences complexes du stress hydrique. Pour simuler l'avenir, avec des impacts possiblement plus importants des sécheresses, le modèle doit être adapté avec des processus spécifiques liés aux sécheresses et leurs effets différés. / Through this work I aimed to improve the understanding of the role of land-atmosphere feedbacks and large-scale circulation that lead to warm summer temperatures in Europe. This is challenging due to the scarcity of observations and the uncertainties of parameterized atmospheric processes. I focused on four main issues: 1) How do land-atmosphere feedbacks affect climate projections and their uncertainties? 2) How do different physical parameterizations affect the simulation of extreme heatwaves? 3) How large are the roles of soil moisture and atmospheric circulation in the development of European summer temperature anomalies? And 4) What are the impacts of heat and drought stress on vegetation?Regarding the first question I found that the different partitioning of land heat fluxes between models leads to spatially different warming over Europe in the future. The uncertainty of future climate change was especially high in central Europe, largely due to the uncertainty in heat flux partitioning, while in Southern Europe the models mostly agreed. The use of observation-based sensible heat fluxes allowed to reduce this climate change uncertainty regionally up to 40%.While studying different atmospheric parameterizations for the extreme heatwaves of 2003 and 2010, I found a large temperature spread between the simulations. Compared to observations, temperature was mostly underestimated. Shortwave radiation and precipitation were generally overestimated. I selected a reduced model ensemble of well performing configurations compared to observations, to perform future studies on warm summer temperatures over Europe.The best physics configuration was consequently used to quantify the role of early summer soil moisture and large-scale drivers on summer temperature anomalies. The contribution of soil moisture was up to maximum 1°C during the heatwaves of 2003 and 2010. The contribution of large-scale drivers was larger, reaching up to 3°C in 2003 and up to 6°C in 2010. However, the contribution of early summer soil moisture to the temperature anomalies has been increasing over the last decades over parts of central Europe and Russia, corresponding to the regions with a significant negative trend of soil moisture. Large-scale drivers showed an increasing importance in the Eastern European region.Lastly, I studied the impacts of drought and heat stress on several European forest tree species. I found an overestimation of modeled GPP at a local scale in the Mediterranean region during summer with ORCHIDEE. This indicates that the vegetation model does not well reproduce the complicated consequences of drought stress. To model future, possibly more severe impacts of drought, the model may need to be adapted with drought-specific processes and lagged effects.

Due to climate change, extreme weather and heatwaves will become more frequent inSweden. Heatwaves have been proven all around the world to cause increased sickness and death rates in populations. Even in the Nordic countries heatwaves can cause negative health effects and sickness already in the current climate. Because of this, it´s important to know what building companies are doing to prevent overheating in their buildings and how they adapt their buildings for a changed future climate. The purpose of this study was to examine how Swedish building companies adapt their buildings to prevent future overheating and what method they are using to do so. The result showed that most of the interviewed companies prioritized adapting their buildings to prevent overheating. Companies which did not prioritize this issue argued that they follow customer demands or that they have not perceived overheating as a problem. There were no significant differences between companies in southern and northern Sweden in which methods the companies decide to use to lower temperatures inside buildings. To avoid risks related to overheated apartments in the future every company in the building sector needs to prioritize these problems and government agencies needs to construct stricter laws regarding indoor temperatures. Thus, the companies that are working based on customer request would also need to adapt to a changed future climate.

Die Vorhersage von extremen Wetterereignissen wie Hitzewellen ist aufgrund ihrer Auswirkungen auf die Gesellschaft von großer Bedeutung. Subsaisonale Wettervorhersage (von 2 Wochen bis 2 Monaten) kann Frühwarnungen liefern, die für Pläne zur Risikoeinschränkung unerlässlich sind. Das fachübergreifende Ziel dieser Arbeit ist daher die Verbesserung der Wettervorhersage von Hitzewellen über Europoa auf subsaisonalen Zeitskala. Um dieses Ziel zu erreichen, werden zunächst die Quellen der Vorhersagbarkeit auf der subsaisonalen Skala herausarbeitet und analysiert und anschließend die Vorteile quantifiziert, die diese einzelne Prädiktoren bieten können. Im ersten Teil werden die Haupttypen von Hitzewellen über Europa anhand ihrer atmosphärischen Zirkulation definiert. Die europäischen Hitzewellen werden dazu in fünf Hitzewellentypen mit spezifischen Grosswetterlagen eingeteilt. Diese ermöglichen es wiederum, die vorhersagbare Komponente der Hitzewellenereignisse zu bestimmen: allen gemein sind starke, anhaltende antizyklonale Anomalien über der Region der Höchsttemperaturen. Anhand dieser Klassifizierung lässt sich zudem die relative Bedeutung anderer subsaisonaler Faktoren wie Bodenfeuchtigkeit und verstärkte tropische Konvektion bestimmen. Dabei hat es sich erwiesen, dass eine geringere Bodenfeuchtigkeit vor Hitzewellen nur für Hitzewellen über Südeuropa und für sehr extreme Hitzewellen von Bedeutung ist, da sie die Temperaturanomalien weiter verstärkt. Die Boreale Sommer Intrasaisonale Oszillation (BSISO) ist durch verschiedene Phasen schwacher und starker tropischer Konvektion gekennzeichnet. Die Beobachtung der Entwicklung der BSISO zeigt einen klaren Zusammenhang zwischen bestimmten aktiven Phasen der BSISO und dem Beginn russischer Hitzewellen, dass deren Überwachung zur einer besseren Vorhersagbarkeit beitragen könnte. Im zweiten Teil dieser Arbeit wird der durch die Verwendung der identifizierten Prädiktoren entstehende Vorteil quantifiziert. Dazu wird eine musterbasierte Methode entwickelt, bei der die Großwetterlagen als Prädiktoren verwendet werden, mit dem Ziel die Wahrscheinlichkeit extrem warmer Temperaturen abzuleiten. Anhand von Hindcast-Daten des Europäischen Zentrums für mittelfristige Wettervorhersage (EZMW) wird diese Methode mit der direkten gitterpunktbasierten Wettervorhersage verglichen. Die musterbasierte Vorhersage erzielt für kurze bis mittelfristige Vorhersagen zwar keine gute Vorhersagequalität. Die Vorhersagequalität dieser Ergebnisse wird jedoch auch nach mehr als 10 Tagen beibehalten, was die Vorhersagequalität-Horizont über den Hitzewellenregionen erheblich verbessert. Die Verwendung von anhaltenden Großwetterlagen zur Vorhersage anhaltenden Hitzewellen führt zu ähnlichen Ergebnissen, allerdings mit lokal begrenzteren und bescheideneren Verbesserungen. Außerdem verbessern aktive BSISO-Phasen die Vorhersage zwar nicht systematisch, aber sie verbessern die Vorhersagbarkeit des Ausbruchs von russischen Hitzewellen, indem sie die Vorhersagespanne deutlich verringern und deren Genauigkeit, wenn auch in geringerem Maße erhöhen. Die Einbeziehung dieser subsaisonalen Einflussfaktoren in einen Entscheidungsprozess zur Aktivierung von Risikominderungsplänen könnte wesentliche Informationen für Frühwarnungen liefern.:1. Introduction ................................................................................................................................. 13 1.1 Impact of extremes............................................................................................................... 13 1.2 Mitigation and early warnings .............................................................................................. 14 1.3 Content ................................................................................................................................. 14 2. Background knowledge ................................................................................................................ 16 2.1 Generalities on heatwaves ................................................................................................... 16 2.1.1 Heatwave definitions ................................................................................................ 16 2.1.2 Processes responsible for heatwaves ....................................................................... 17 2.1.3 Climatic trends for heatwaves .................................................................................. 18 2.2 Generalities on predictions................................................................................................... 20 2.2.1 Numerical Prediction ................................................................................................ 21 2.2.1.1 Governing equations............................................................................................. 21 2.2.1.2 Discretisation and parametrisation....................................................................... 23 2.2.1.3 Data assimilation .................................................................................................. 24 2.2.2 Predictability and forecast verification ..................................................................... 24 2.2.2.1 Predictability ......................................................................................................... 25 2.2.2.1.1 Predictability of the first kind ............................................................................. 25 2.2.2.1.2 Predictability of the second kind ........................................................................ 27 2.2.2.2 Forecast verification ............................................................................................. 30 2.2.2.2.1 Observations and reanalysis ............................................................................... 30 2.2.2.2.2 Forecast skill metrics .......................................................................................... 31 3. European heatwaves and their link to large-scale circulation patterns and subseasonal drivers 34 3.1 Data and methods ................................................................................................................ 35 3.1.1 Land surface feedback .............................................................................................. 35 3.1.2 Enhanced tropical convection ................................................................................... 36 3.1.3 Heatwave detection.................................................................................................. 37 3.1.4 Classification of heatwave patterns .......................................................................... 38 3.2 Heatwave types and their relation to circulation patterns ................................................... 42 3.2.1 Heatwave types description ..................................................................................... 42 3.2.2 Heatwave circulation patterns .................................................................................. 47 3.3 Potential sources of predictability at the subseasonal time scale ........................................ 49 3.3.1 Land surface feedback .............................................................................................. 50 3.3.2 The Boreal Summer IntraSeasonal Oscillation .......................................................... 55 3.3.2.1 BSISO phases facouring the occurrence of heatwaves ......................................... 57 3.4 Summary .............................................................................................................................. 64 12 4. Subseasonal prediction of heatwaves enhanced using a pattern-based forecasting system ....... 66 4.1 Data and methods ................................................................................................................ 67 4.1.1 Connection between extreme high temperatures and circulation patterns ............. 68 4.1.2 Skill evaluation of ECMWF forecasts ......................................................................... 70 4.2 Prediction of extreme temperatures at the subseasonal range ........................................... 72 4.2.1 Direct forecasting ..................................................................................................... 72 4.2.2 Pattern-based/conditional forecast .......................................................................... 77 4.2.3 Skill sensitivity to tropical convections ..................................................................... 83 4.3 Summary .............................................................................................................................. 88 5. Conclusion .................................................................................................................................... 91 5.1 Summary and key findings .................................................................................................... 91 5.2 Outlook ................................................................................................................................. 92 References ............................................................................................................................... / The prediction of extreme events such as heatwaves is of high importance due to their impact on society. Subseasonal prediction (from 2 weeks to 2 months) can provide early warnings which are essential for setting up mitigation plans. Therefore, the overarching goal of this work is to improve the forecast of heatwaves at the subseasonal time scale over Europe. The approach used to tackle this goal, is to first identify and analyse the sources of predictability at the subseasonal scale, and to then quantify the benefits of each of these predictors. In the first phase, the main heatwave types over Europe are defined based on their atmospheric circulation. European heatwaves are therefore classified into five heatwave types with specific circulation patterns, allowing to determine the predictable component of heatwave events. They all have strong persistent anti-cyclonic anomalies over the region of maximum temperatures. The classification further allows to determine the relative importance of other subseasonal drivers such as soil moisture and tropical enhanced convection. Reduced soil moisture content prior to heatwaves is shown to be relevant only to heatwaves over southern Europe and for very extreme heatwaves, by further amplifying the temperature anomalies. The Boreal Summer Intraseasonal Oscillation (BSISO) is characterised by different phases of weak and strong tropical convection. Monitoring the evolution of the BSISO shows a clear link between certain active phases of the BSISO and the onset of Russian heatwaves in particular, suggesting that they could provide enhanced predictability for the onset of Russian heatwaves. In the second phase, the advantage of using the identified predictors is quantified. A pattern-based method is constructed, using the circulation patterns as predictors to infer the probability of extreme warm temperatures. Using reforecast data from the European Centre for Medium-Range Weather Forecasts (ECMWF), this method is compared to the direct grid-point based prediction. The pattern-based prediction shows low skill at short to medium range, however it maintains this skill beyond 10 days and significantly improves the forecast range over the regions of heatwaves. Using persistent circulation patterns to forecast persistent heatwaves shows similar results, but with more localised and modest improvements. In addition, while active BSISO phases do not systematically improve the prediction, they do enhance the predictability of the onset of Russian heatwaves by reducing significantly the forecast spread and to a lesser extent increase accuracy. Incorporating these subseasonal drivers into a decision-making process for mitigation plans could provide essential information for early warnings.:1. Introduction ................................................................................................................................. 13 1.1 Impact of extremes............................................................................................................... 13 1.2 Mitigation and early warnings .............................................................................................. 14 1.3 Content ................................................................................................................................. 14 2. Background knowledge ................................................................................................................ 16 2.1 Generalities on heatwaves ................................................................................................... 16 2.1.1 Heatwave definitions ................................................................................................ 16 2.1.2 Processes responsible for heatwaves ....................................................................... 17 2.1.3 Climatic trends for heatwaves .................................................................................. 18 2.2 Generalities on predictions................................................................................................... 20 2.2.1 Numerical Prediction ................................................................................................ 21 2.2.1.1 Governing equations............................................................................................. 21 2.2.1.2 Discretisation and parametrisation....................................................................... 23 2.2.1.3 Data assimilation .................................................................................................. 24 2.2.2 Predictability and forecast verification ..................................................................... 24 2.2.2.1 Predictability ......................................................................................................... 25 2.2.2.1.1 Predictability of the first kind ............................................................................. 25 2.2.2.1.2 Predictability of the second kind ........................................................................ 27 2.2.2.2 Forecast verification ............................................................................................. 30 2.2.2.2.1 Observations and reanalysis ............................................................................... 30 2.2.2.2.2 Forecast skill metrics .......................................................................................... 31 3. European heatwaves and their link to large-scale circulation patterns and subseasonal drivers 34 3.1 Data and methods ................................................................................................................ 35 3.1.1 Land surface feedback .............................................................................................. 35 3.1.2 Enhanced tropical convection ................................................................................... 36 3.1.3 Heatwave detection.................................................................................................. 37 3.1.4 Classification of heatwave patterns .......................................................................... 38 3.2 Heatwave types and their relation to circulation patterns ................................................... 42 3.2.1 Heatwave types description ..................................................................................... 42 3.2.2 Heatwave circulation patterns .................................................................................. 47 3.3 Potential sources of predictability at the subseasonal time scale ........................................ 49 3.3.1 Land surface feedback .............................................................................................. 50 3.3.2 The Boreal Summer IntraSeasonal Oscillation .......................................................... 55 3.3.2.1 BSISO phases facouring the occurrence of heatwaves ......................................... 57 3.4 Summary .............................................................................................................................. 64 12 4. Subseasonal prediction of heatwaves enhanced using a pattern-based forecasting system ....... 66 4.1 Data and methods ................................................................................................................ 67 4.1.1 Connection between extreme high temperatures and circulation patterns ............. 68 4.1.2 Skill evaluation of ECMWF forecasts ......................................................................... 70 4.2 Prediction of extreme temperatures at the subseasonal range ........................................... 72 4.2.1 Direct forecasting ..................................................................................................... 72 4.2.2 Pattern-based/conditional forecast .......................................................................... 77 4.2.3 Skill sensitivity to tropical convections ..................................................................... 83 4.3 Summary .............................................................................................................................. 88 5. Conclusion .................................................................................................................................... 91 5.1 Summary and key findings .................................................................................................... 91 5.2 Outlook ................................................................................................................................. 92 References ...............................................................................................................................

Philosophiae Doctor - PhD (Biodiversity and Conservation Biology) / Seawater temperature from regional to global scale is central to many measures of biodi- versity and continues to aid our understanding of the evolution and ecology of biolog- ical assemblages. Therefore, a clear understanding of the relationship between marine biodiversity and thermal structures is critical for effective conservation planning. In the an- thropocene, an epoch characterised by anthropogenic forcing on the climate system, future patterns in biodiversity and ecological functioning may be estimated from projected climate scenarios however; absent from many of these scenarios is the inclusion of extreme thermal events, known as marine heatwaves (MHWs). There is also a conspicuous absence in knowl- edge of the drivers for all but the most notorious of these events. Before the drivers of MHWs along the coast of South Africa could be determined, it was first necessary to validate the 129 in situ coastal seawater temperature time series that could be used to this end. In doing so it was found that time series created with older (longer), lower precision (0.5 Degrees Celsius) instruments were more useful than newer (shorter) time series produced with high precision (0.001 Degrees Celsius) instruments. With the in situ data validated, a history of the occurrence of MHWs along the coastline (nearshore) was created and compared against MHWs detected by remotely sensed data (offshore). This comparison showed that the forcing of offshore temperatures onto the nearshore was much lower than anticipated, with the rates of co-occurrence for events between the datasets along the coast ranging from 0.2 to 0.5. To accommodate this lack of consistency between datasets, a much larger mesoscale area was then taken around southern Africa when attempting to determine potential mesoscale drivers of MHWs along the coast. Using a self organising-map (SOM), it was possible to organise the synoptic scale oceanographic and atmospheric states during coastal MHWs into discernible groupings. It was found that the most common synoptic oceanographic pattern during coastal MHWs was Agulhas Leakage, and the most common atmospheric pattern was anomalously warmoverland air temperatures.With these patterns known it is now necessary to calculate how often they occur when no MHW has been detected. This work may then allow for the development of predictive capabilities that could help mitigate the damage caused by MHWs.

Background: Vietnam suffers consequences of global warming. There is limited data of the relationship between weather, extreme heat and potential mental health problems. It is therefore crucial to study heat-related mental illnesses and to establish good solutions with relevant adaptations to global warming. The adaptation measures should give attention to people that live in areas facing annual extreme weather, and protecting health in general and more specifically mental health of citizens. The study aimed to examine relationships between weather patterns, extreme heat or heatwaves, and mental disorders, and to investigate factors contributing to increased vulnerability and susceptibility. Methods: The thesis includes a systematic review and a hospital-based study using data from the Hanoi Mental Hospital for five years (2008 – 2012), with mental disorders diagnosed by ICD10 (F00-99) to estimate the effects of weather variation, seasonality, increased temperatures, and heatwaves on hospital admissions for depression and other mental disorders. A negative binomial regression model accounting for yearly study period, time trends, and day of the week was used to analyze the relationship between seasonality, heatwaves, and monthly and daily mental disorder hospitalizations. Results: Our findings showed (i) a general tendency for more admissions between May and December, with a seasonal bi-annual high between May-June and November-December, and elevated ambient temperature was significantly related to increasing admissions for depressive disorders; (ii) the number of hospital cases for mental disorders increased in the summer seasone specially in June, and two percent of cases emerged during elevated temperature of one degree Celsius; and (iii) when compared with non-heatwave periods, heatwaves amounted to increasing risks for admission for the whole group of mental disorders (F00-79), and admissions for mental disorders among residents in rural communities and in the elderly population increased significantly during heatwaves. Conclusion: There were associations between hospital admissions for depression and other mental disorders and seasonality, weather patterns, elevated temperatures, and heatwaves. The associations grew stronger with the length of the heatwaves and particularly the elderly appeared more sensitive to seasonality, hot weather and heatwaves.

Pays en développement avec une demande d’électricité croissante, le Vietnam a instauré la tarification progressive de l’électricité résidentielle. La fixation du tarif de l’énergie est toujours une question délicate, entre gestion de la demande, lutte contre la pauvreté, effets sur l’inflation, besoins d’investissement pour assurer la sécurité énergétique et le développement des technologies vertes. Cette action nécessite une maîtrise très profonde du comportement des consommateurs ainsi que la demande des ménages. La thèse a pour but d’explorer les facteurs qui impactent la demande d’électricité Vietnamienne au niveau résidentielle en se basant sur : les prix, les revenus, la démographie (comprenant la taille et la composition des foyers) et les vagues de chaleur. Les données de « pool et panel » sont collectées à partir des trois micro enquêtes sur le niveau de vie des foyers vietnamiens en 2012, 2014, 2016.Cette thèse estime économétriquement la demande d’électricité des ménages. Elle innove sur deux points de méthode.Premièrement, elle utilise les données individuelles issues des enquêtes nationales, avec le détail de la structure des tarifs et des factures d’électricité des ménages répondants. Cela dépasse donc les limites de beaucoup de recherches passées qui étaient basées soit sur données nationales agrégées, soit sur données individuelles mais avec une quantité et un prix imputés, soit sur données individuelles avec le détail de la structure des tarifs et des factures d’électricité mais au niveau régional seulement. Cette innovation est possible car le marché de l’électricité au Vietnamien est monopolistique, avec un seul vendeur – Electricité de Vietnam (EVN), à qui le gouvernement commande d’utiliser une grille tarifaire homogène pour tout le pays.Deuxièmement, la thèse propose une nouvelle façon d’explorer l’impact des hautes températures sur la demande d’électricité. La méthode propose d’ajouter une variable muette qui représente l’occurrence d’une vague de chaleur. Cette variable est complémentaire de la notion « Degrés-jours de refroidissement » qui représente la température dans la plupart des études précédentes.Les conclusions principales sont que: (i) Les ménages réagissent aux prix marginaux, la demande est élastique par rapport au prix. (ii) Il existe un seuil de revenu à partir duquel la consommation d'électricité des foyers augmente quand le revenu augmente : la consommation d'électricité des foyers ayant ce revenu peut être considérée comme le niveau de besoin fondamental, un seuil de pauvreté pour l’électricité. (iii) La progressivité de la tarification ne pénalise pas les familles nombreuses : le tarif progresse moins vite que les d’économies d'échelle des dépenses d'électricité. (iv) Nous n’observons pas d’effet de la composition du foyer en termes enfants / adultes / personnes âgés sur la dépense d'électricité. (v) Les vagues de chaleur - un phénomène lié au changement climatique - impactent la demande d’électricité et devraient être mieux prises en compte dans l’estimation de la demande. / As a developing country with surging demand in electricity, Vietnam has implemented demand-side management in the residential electricity market, such as increasing block tariffs to balance the tension between energy security and the development of clean technology. The implementation of demand-side management requires a deep understanding of customer behaviors and household demand. The thesis aims to explore the factors impacting on Vietnamese residential electricity demand in the period of 2012–16. The exploration focuses on four main factors: prices, income, demographics (including household size and composition), and heatwaves. The data are a pool data set and a panel data set which have been constructed from the three rounds of the micro survey Vietnam Household Living Standard Survey (VHLSS) in 2012, 2014 and 2016.The thesis has two novel points in estimating household electricity demand function.First, it uses micro survey data at national level, with detailed tariff structures and private electricity billing. In the past, researches have often used national aggregate data or national micro survey data with imputed quantity or price. Researches that use micro survey data with detail tariff schedules and electricity bills are often at a regional level rather than at a national level due to the absence of national data on tariff structures. The residential electricity market in Vietnam is a monopoly with a single seller, Vietnam Electricity (EVN). Electricity tariff schedules are proposed by EVN and set by the Government and are thus uniform in national scale. This provides a chance to estimate demand function from national micro survey data, with full detail of electricity prices and billings.Second, the thesis proposes a new way to capture the impact of high temperature on electricity demand. That is, to include an additional dummy variable to represent the extreme distribution of temperature. The additional dummy variable is a complement to the concept of cooling degree days which is a popular representation of temperature in previous researches.The estimate results lead to five main conclusions. (i) Households do respond to marginal prices and demand is elastic to price. (ii) There exists an income threshold from which household electricity consumption increases as income increases. The electricity consumption of households in the income group is the reference level of electricity poverty threshold. (iii) The increasing block tariff does not cancel out economies of scale in electricity expenditure of households. (iv) There is no difference in electricity expenditure across children, adults and elders. (v) Heatwaves – a climate change related phenomenon – do have impacts on electricity demand and need to be addressed carefully in estimating electricity demand in the future.

En Europe, l'augmentation des températures moyennes de surface de l'air projetée au cours du 21ème siècle s'accompagne d'une augmentation des extrêmes chauds et d'une diminution des extrêmes froids. Dans les dernières décennies, des indices témoignent déjà de ces changements, comme l'établissement récurrent de nouveaux records de chaleur ou l'augmentation des canicules. Nous étudions l'évolution des extrêmes journaliers de température au cours du 20ème et du 21ème siècle en France et en Europe, et ce en termes d'occurrence et d'intensité. Un intérêt particulier est aussi porté aux mécanismes responsables de ces futurs extrêmes climatiques, ainsi qu'aux futures températures maximales. Nous nous intéressons tout d'abord à l'évolution des records journaliers de température à partir d'observations et de modèles de climat. Entre 1950 et 1980, l'évolution théorique des records dans le cadre d'un climat stationnaire représente correctement l'évolution observée des records chauds et froids. Depuis les années 1980, un écart à ce climat stationnaire est observé, avec respectivement une augmentation et une diminution de l'occurrence des records chauds et froids. Les modèles climatiques suggèrent une accentuation de ces changements au cours du siècle. L'occurrence moyenne des records chauds à la fin du siècle présente une forte augmentation par rapport aux premières décennies de la période observée. L'augmentation la plus importante des records chauds est projetée en été, en particulier dans la région méditerranéenne. Quant aux records froids, les modèles indiquent une diminution très importante de leur occurrence, avec une occurrence quasi-nulle dans les dernières décennies. Les variations observées d'occurrence de records sont, au début du 21ème siècle, toujours dans l'éventail des fluctuations de la variabilité interne du climat. Au cours du siècle, l'émergence de l'influence anthropique de ces fluctuations est détectable dans l'évolution des records chauds et froids en été, et ce respectivement autour des décennies 2030 et 2020. À l'horizon de la fin du siècle, les changements moyens d'occurrence de records ne peuvent pas être uniquement expliqués par des fluctuations naturelles. Nous nous sommes ensuite intéressés aux futures températures estivales extrêmes, ainsi qu'aux canicules intenses qui peuvent être à l'origine de ces extrêmes. Pour cela, l'utilisation de modèles climatiques globaux est associée à la modélisation climatique régionale et à des stations d'observations en France. Tout d'abord, l'augmentation maximale des valeurs maximales des records journaliers de température en été en France est estimée à partir d'une simulation régionale à haute résolution spatiale. À l'horizon 2100, les projections indiquent une augmentation maximale de ces valeurs extrêmes en été comprise entre de 6.6°C et 9.9°C selon les régions de la France. La comparaison de ces projections avec un ensemble de modèles climatiques indique que ces augmentations maximales pourraient être plus importantes. La médiane de la distribution des modèles indique en effet une augmentation maximale de ces valeurs maximales des records journaliers de température de 11.8°C en été et en France. Puis, des expériences de modélisation de canicules intenses du climat européen de la fin du 21ème siècle ont été réalisées à partir d'événements particuliers d'un modèle de climat. Ces expériences ont mis en évidence le rôle des interactions entre le sol et l'atmosphère dans l'amplification des températures extrêmes lors de futurs évènements caniculaire intenses. L'occurrence de telles canicules est d'abord dépendante de la circulation atmosphérique, mais l'intensité des températures peut ensuite être fortement amplifiée en fonction du contenu en humidité des sols avant la canicule, et donc des conditions climatiques des semaines et des mois précédents. / Over the 21st century, the mean increase in surface air temperatures is projected to be associated with an increase in warm temperature extremes and a decrease in the cold ones. Over the last decades, evidence already suggests these changes, as for example recurrent warm record-breaking temperatures or the increase in heatwave occurrence. We investigate the evolution of daily temperature extremes over the 20th and the 21st centuries in France and in Europe, their possible changes in frequency and intensity. We also focus on the mechanisms responsible for these projected climate extremes, as well as the maximum values of temperature extremes at the end of the century. First, we investigate the evolution of daily record-breaking temperatures in Europe based on the observations and an ensemble of climate models. From the 1950s to the 1980s, the theoretical evolution of the records in a stationary climate correctly reproduce the observed one, for both cold and warm records. From 1980, a shift from that theoretical evolution is observed, with an increase in the occurrence of warm records and a decrease in the occurrence of the cold ones. Climate models suggest an amplification of these changes over the century. At the end of the 21st century, the mean number of warm records shows a strong increase compared to the first decades of the observed period. The strongest increase in warm record-breaking temperatures is found in summer, and particularly over the Mediterranean edge. On the contrary, the occurrence of cold record-breaking temperatures is projected to strongly decrease, with almost no new records in the last decades of the century, for all seasons and over the entire European domain. Observed variations of daily record-breaking temperatures are still, at the beginning of the 21st century, consistent with internal climate variability only. Over the century, the anthropogenic influence emerge from these fluctuations in the summer record evolutions, around the 2030 and the 2020 for the warm and cold records respectively. By 2100, the mean changes in record occurrences cannot be explained by the internal climate variability solely, for all seasons and over the entire European domain. Then, we investigate future extreme temperatures at the end of the 21st century, as well as severe heatwaves leading to these extremes. Climate models analyses are associated with regional climate modeling and a French station-based dataset of observations. The summer 21st century evolution of the maximum values of daily warm record-breaking temperatures is first examined in the observations and the high resolution simulation of the regional model. By 2100, an increase of these values is projected, with maximum changes between +6.6°C and +9.9°C in summer among the French regions. These projections assessed from a regional model may underestimate the changes. The multi-model mean estimate of the maximum increase of these values is indeed around +11.8°C in summer over France. Finally, regional modeling experiments of severe heatwaves in the climate of the end of the 21st century in Europe are performed. These severe heatwaves are selected cases from a global climate model trajectory. The experiments results show the role of the soil-atmosphere interactions in the amplification of the extreme temperatures during such future severe warm events. The occurrence of the heatwave is first caused by the atmospheric circulation, but the temperature anomaly can then be amplified according to the soil moisture content before the event, and thus the climatic conditions of the preceding weeks and months.

Europa erlebte in den letzten 20 Jahren einige sehr heiße und trockene Sommer mit regionalen Rekordwerten heißer Temperaturen oder geringer Niederschlagssummen. In anderen Jahren führten Starkregen zu Überflutungen unterschiedlichen räumlichen Ausmaßes. Da solche Extremereignisse mit vielfältigen negativen Auswirkungen auf die menschliche Gesellschaft, natürliche Ökosysteme und verschiedene Wirtschaftssektoren verbunden sind, ist die langzeitliche Veränderung in ihrem Auftreten im Rahmen der globalen Erwärmung von großer Bedeutung. Konzepte: Maßgeblich für die Qualität von Klimawandel(folgen)studien ist die Verfügbarkeit und Qualität von Daten. Daher werden Konzepte für die Sicherstellung einer zuverlässigen und vergleichbaren Datenbasis entwickelt. Für die Beschreibung der Eigenschaften eines bestimmten Ereignisses existiert eine Vielzahl an Definitionen und Indizes, was zu unterschiedlichen Ergebnissen von Studien führen kann, welche die zeitlichen Veränderungen der Charakteristik solcher Ereignisse analysieren. Die Integration einer Reihe von Indizes in einen aggregierten Index ermöglicht eine robustere Bewertung der Klimabedingungen und Trends. Die Vergleichbarkeit von Klimafolgenbewertungen verlangt zudem die Verwendung eines gemeinsamen Analyserahmens sowie abgestimmter Datensätze (Beobachtungsdaten, Klimaprojektionen) und Methoden (z.B. Untersuchungszeiträume, Ensemble-Ansatz, Qualitätsbewertung, Korrekturalgorithmen, Impactmodelle und -indizes, Elemente der Klimafolgen- oder Risikoanalyse). Trends: Sommerliche Trockenheit hat über weiten Teilen Europas – mit Ausnahme des Nordens – zugenommen. Besonders stark zugenommen haben die Dürrebedingungen im Sommerhalbjahr für Indizes, welche die Evapotranspiration einbinden. Der reine Fokus auf den Niederschlag zur Bewertung von Dürre in verschiedenen Speichern des Wasserkreislaufs ist unzureichend. Neben dieser beobachteten Zunahme in der Sommertrockenheit, ist auch für die Intensität von Starkniederschlagsereignissen und ihrem Anteil am Gesamtniederschlag ein Anstieg über Europa zu beobachten. Verschiedene Stationen in Mitteleuropa zeigen für das Sommerhalbjahr gleichzeitige Anstiege in den Dürrebedingungen und Starkniederschlägen, was die mit solchen Niederschlagsextremen verbundenen Folgen und Risiken erhöht. Folgen: Viele Sektoren sind durch die Folgen des Klimawandels und extreme Wettereignisse negativ betroffen, so auch das Verkehrssystem. Dessen Verfügbarkeit und Leistungsfähigkeit ist von hoher Bedeutung für die Gesellschaft (Mobilität) und Wirtschaft (Waren, Transportketten). Extreme Wettereignisse wie Hitzewellen, Überschwemmungen, Dürren, Stürme und Sturmfluten können Unfälle und Staus verursachen, die Infrastrukturen beschädigen und damit Transportketten unterbrechen sowie zu Verspätungen und Ausfällen führen. Die Verkehrsträger sind dabei in unterschiedlicher Weise und Intensität betroffen. Um die Klimawandelfolgen für das Bundesverkehrssystem zu bewerten und Anpassungsbedarfe zu priorisieren wurde ein methodischer Rahmen für die Durchführung von Klimawirkungsanalysen entwickelt. Ergänzt werden diese nationalen Analysen durch Klimafolgenstudien für die UNECE-Region (UNECE: Wirtschaftskommission für Europa der Vereinten Nationen). Zielgerichtete Klimadienstleistungen, welche die Bedarfe der Anwendenden integrieren, sind eine Grundvoraussetzung für die Entwicklung praktikabler Anpassungsoptionen.:Abstract 1 Zusammenfassung 2 1. Research topic and questions addressed 3 2. Outline and structure of this thesis 6 3. Concepts – How to evaluate changes in heat, drought and wetness? 11 3.1 How to define drought? 11 3.2 How to measure changes in (extreme) temperature and precipitation? 11 3.2.1 Applying established climate indices 11 3.2.2 Developing new indices to measure drought and wetness 12 3.2.3 Assessing extreme events and their impacts 14 3.3 How to ensure good quality climate data sets? 15 3.3.1 Separating climate variability from changes in non-climatic parameters 15 3.3.2 Regionalizing climate information 15 3.3.3 Adjusting biases in climate projections 16 3.4 How to ensure comparable results of climate impact assessments? 17 3.4.1 Agreeing on common assumptions and scenarios 17 3.4.2 Applying an ensemble analysis approach 17 3.4.3 Implementing a common analysis framework for impact assessment 18 4. Changes – Which variations are seen in the regional climate? 20 4.1 Variations and changes in the average climate – temperature and precipitation 20 4.1.1 Changes in wet and dry periods over Europe 20 4.1.2 Observed and projected temperature and precipitation trends over Germany 21 4.1.3 Observed climatic changes in North-eastern Brazil (NEB) 21 4.1.4 Observed precipitation variations in the Palestinian territories and surrounding areas 22 4.2 Extreme weather and climate events: spatio-temporal variations and trends 22 4.2.1 Increases in temperature extremes and heatwaves 22 4.2.2 Characteristics of and changes in heavy precipitation 23 4.2.3 Temporal variations in meteorological drought conditions 26 4.2.4 Drought and heavy precipitation 28 4.3 Characterising selected record hot and dry summers 30 4.3.1 The five record drought summers in Europe – 1947, 2018, 2003, 1921 and 1911 30 4.3.2 The summer of 2018 31 4.3.3 The summer of 2015 32 4.3.4 Recent hot and dry summers in Germany in comparison to climate projections 33 5. Consequences – Which climate impacts do we have to expect and how to adapt to them? The case of the transport system 35 5.1 Relevance of climate change considerations for the transport system 35 5.2 Networks supporting the development of climate resilient transport systems 35 5.2.1 BMDV Network of Experts on Climate Change Impacts and Adaptation 36 5.2.2 DAS core service “climate and water” 37 5.2.3 UNECE Group of Experts on Assessment of Climate Change Impacts and Adaptation for Inland Transport 38 5.3 Climate change impact analysis for the transportation sector 39 5.3.1 Methodology of the integrated climate impact assessment 39 5.3.2 Exemplary results of the exposure analysis 40 5.3.3 Integrated climate impact assessment 40 5.4 Stress testing the transport system 41 5.4.1 The stress test methodology 41 5.4.2 Exemplary results of the traffic simulations 41 5.5 Developing an adaptation framework and specific measures 42 5.5.1 Background and classification of adaptation measures 42 5.5.2 Information and consultation services 42 5.5.3 Reviewing and revising technical guidelines and standards 43 5.5.4 Structural adaptation measures 43 5.5.5 Adapting management practices of transportation infrastructure 43 5.5.6 Adapting the operative management of traffic flows 44 5.5.7 Survey results on suitable adaptation measures 44 6. Conclusions 45 6.1 Concepts 45 6.2 Changes 45 6.3 Consequences 46 7. References 48 / Over the last 20 years, some very hot and dry summers affected Europe, regionally resulting in record breaking high temperature or low precipitation values. In other years, torrential rains led to flood events at different spatial scales. Long-term changes of such extreme events within a warming world are of great relevance, as they are connected with manifold negative impacts on human society, natural ecosystems and diverse economic sectors. Concepts: The quality of climate change (impact) studies is often hampered by availability and quality of datasets. Thus, concepts for securing reliable and comparable data are developed and applied. For the description of the characteristics of a specific event a vast number of definitions and indices exists. Therefore, results on the temporal changes of event characteristics may differ between studies. By integrating a number of indices into an aggregated index, a more robust evaluation of the climate conditions and trends is facilitated. Furthermore, comparable climate impact assessments demand a common analysis framework with agreements on the data bases (observational data and climate projections) and methodologies (e.g., study periods, ensemble approach, quality assessment, correction algorithms, climate impact models and indices, elements considered in the impact or risk analysis). Changes: Summer drought conditions increased over most of Europe, except for some stations in northern Europe. Thereby, the observed increase in drought conditions during the warm part of the year is particularly pronounced for indices integrating evapotranspiration in their definition. Purely focussing on precipitation to evaluate drought conditions in the different water reservoirs does not suffice any longer. While observing increases in summer drought, the intensity of heavy precipitation events as well as their contribution to total precipitation show a positive trend over Europe, too. Several stations in Central Europe show increasing drought conditions and increasing heavy precipitation events during the summer half year at the same time, which increases the risks connected with precipitation extremes. Consequences: Climate change impacts on the transport system are studied exemplarily for the many sectors that are affected negatively by the projected changes in climate and extreme weather events. The availability and performance of the transport system are of high importance for the society (mobility) and economy (goods, transport chains). Extreme weather events such as heatwaves, flooding, droughts, and storm surges might 1) cause accidents and congestion, 2) severely damage to infrastructures and disrupt transport chains, and 3) result in delays and cancellations. Different modes of transport are affected by climate change in different ways and with different intensity. A climate impact assessment framework was defined and tested for the German Federal transport system to support the prioritization of adaptation options. Climate change impact studies for the UNECE-region (United Nations Economic Commission for Europe) complement these Federal analyses. It is shown that tar-geted climate services that integrate user requirements are key in developing feasible adaptation options.:Abstract 1 Zusammenfassung 2 1. Research topic and questions addressed 3 2. Outline and structure of this thesis 6 3. Concepts – How to evaluate changes in heat, drought and wetness? 11 3.1 How to define drought? 11 3.2 How to measure changes in (extreme) temperature and precipitation? 11 3.2.1 Applying established climate indices 11 3.2.2 Developing new indices to measure drought and wetness 12 3.2.3 Assessing extreme events and their impacts 14 3.3 How to ensure good quality climate data sets? 15 3.3.1 Separating climate variability from changes in non-climatic parameters 15 3.3.2 Regionalizing climate information 15 3.3.3 Adjusting biases in climate projections 16 3.4 How to ensure comparable results of climate impact assessments? 17 3.4.1 Agreeing on common assumptions and scenarios 17 3.4.2 Applying an ensemble analysis approach 17 3.4.3 Implementing a common analysis framework for impact assessment 18 4. Changes – Which variations are seen in the regional climate? 20 4.1 Variations and changes in the average climate – temperature and precipitation 20 4.1.1 Changes in wet and dry periods over Europe 20 4.1.2 Observed and projected temperature and precipitation trends over Germany 21 4.1.3 Observed climatic changes in North-eastern Brazil (NEB) 21 4.1.4 Observed precipitation variations in the Palestinian territories and surrounding areas 22 4.2 Extreme weather and climate events: spatio-temporal variations and trends 22 4.2.1 Increases in temperature extremes and heatwaves 22 4.2.2 Characteristics of and changes in heavy precipitation 23 4.2.3 Temporal variations in meteorological drought conditions 26 4.2.4 Drought and heavy precipitation 28 4.3 Characterising selected record hot and dry summers 30 4.3.1 The five record drought summers in Europe – 1947, 2018, 2003, 1921 and 1911 30 4.3.2 The summer of 2018 31 4.3.3 The summer of 2015 32 4.3.4 Recent hot and dry summers in Germany in comparison to climate projections 33 5. Consequences – Which climate impacts do we have to expect and how to adapt to them? The case of the transport system 35 5.1 Relevance of climate change considerations for the transport system 35 5.2 Networks supporting the development of climate resilient transport systems 35 5.2.1 BMDV Network of Experts on Climate Change Impacts and Adaptation 36 5.2.2 DAS core service “climate and water” 37 5.2.3 UNECE Group of Experts on Assessment of Climate Change Impacts and Adaptation for Inland Transport 38 5.3 Climate change impact analysis for the transportation sector 39 5.3.1 Methodology of the integrated climate impact assessment 39 5.3.2 Exemplary results of the exposure analysis 40 5.3.3 Integrated climate impact assessment 40 5.4 Stress testing the transport system 41 5.4.1 The stress test methodology 41 5.4.2 Exemplary results of the traffic simulations 41 5.5 Developing an adaptation framework and specific measures 42 5.5.1 Background and classification of adaptation measures 42 5.5.2 Information and consultation services 42 5.5.3 Reviewing and revising technical guidelines and standards 43 5.5.4 Structural adaptation measures 43 5.5.5 Adapting management practices of transportation infrastructure 43 5.5.6 Adapting the operative management of traffic flows 44 5.5.7 Survey results on suitable adaptation measures 44 6. Conclusions 45 6.1 Concepts 45 6.2 Changes 45 6.3 Consequences 46 7. References 48