Increasingly erratic flow in the upper reaches of the Mara River, has directed attention to land use change as the major cause of this problem. The semi-distributed hydrological model SWAT and Landsat imagery were utilized in order to 1) map existing land use practices, 2) determine the impacts of land use change on water flux; and 3) determine the impacts of climate change scenarios on the water flux of the upper Mara River. This study found that land use change scenarios resulted in more erratic discharge while climate change scenarios had a more predictable impact on the discharge and water balance components. The model results showed the flow was more sensitive to the rainfall changes than land use changes but land use changes reduce dry season flows which is a major problem in the basin. Deforestation increased the peak flows which translated to increased sediment loading in the Mara River.

Die vorliegende Arbeit 'Abflu&szlig;entwicklung in Teileinzugsgebieten des Rheins - Simulationen f&uuml;r den Ist-Zustand und f&uuml;r Klimaszenarien' untersucht Auswirkungen m&ouml;glicher zuk&uuml;nftiger Klima&auml;nderungen auf das Abflu&szlig;geschehen in ausgew&auml;hlten, durch Mittelgebirge gepr&auml;gten Teileinzugsgebieten des Rheins: Mosel (bis Pegel Cochem); Sieg (bis Pegel Menden 1) und Main (bis Pegel Kemmern).<br><br>In einem ersten Schritt werden unter Verwendung des hydrologischen Modells HBV-D wichtige Modellprozesse entsprechend der Einzugsgebietscharakteristik parametrisiert und ein Abbild der Gebietshydrologie erzeugt, das mit Zeitreihen gemessener Tageswerte (Temperatur, Niederschlag) eine Zeitreihe der Pegeldurchfl&uuml;sse simulieren kann. Die G&uuml;te der Simulation des Ist-Zustandes (Standard-Me&szlig;zeitraum 1.1.1961-31.12.1999) ist f&uuml;r die Kalibrierungs- und Validierungszeitr&auml;ume in allen Untersuchungsgebieten gut bis sehr gut.<br>Zur Erleichterung der umfangreichen, zeitaufwendigen einzugsgebietsbezogenen Datenaufbereitung f&uuml;r das hydrologische Modell HBV-D wurde eine Arbeitsumgebung auf Basis von Programmerweiterungen des Geoinformationssystems ArcView und zus&auml;tzlichen Hilfsprogrammen entwickelt. Die Arbeitsumgebung HBV-Params enth&auml;lt eine graphische Benutzeroberfl&auml;che und r&auml;umt sowohl erfahrenen Hydrologen als auch hydrologisch geschulten Anwendern, z.B. Studenten der Vertiefungsrichtung Hydrologie, Flexibilit&auml;t und vollst&auml;ndige Kontrolle bei der Ableitung von Parameterwerten und der Editierung von Parameter- und Steuerdateien ein. Somit ist HBV-D im Gegensatz zu Vorl&auml;uferversionen mit rudiment&auml;ren Arbeitsumgebungen auch au&szlig;erhalb der Forschung f&uuml;r Lehr- und &Uuml;bungszwecke einsetzbar.<br><br>In einem zweiten Schritt werden Gebietsniederschlagssummen, Gebietstemperaturen und simulierte Mittelwerte des Durchflusses (MQ) des Ist-Zustandes mit den Zust&auml;nden zweier Klimaszenarien f&uuml;r den Szenarienzeitraum 100 Jahre sp&auml;ter (2061-2099) verglichen. Die Klimaszenarien beruhen auf simulierten Zirkulationsmustern je eines Modellaufes zweier Globaler Zirkulationsmodelle (GCM), die mit einem statistischen Regionalisierungsverfahren in Tageswertszenarien (Temperatur, Niederschlag) an Me&szlig;stationen in den Untersuchungsgebieten &uuml;berf&uuml;hrt wurden und als Eingangsdaten des hydrologischen Modells verwendet werden.<br>F&uuml;r die zweite H&auml;lfte des 21. Jahrhunderts weisen beide regionalisierten Klimaszenarien eine Zunahme der Jahresmittel der Gebietstemperatur sowie eine Zunahme der Jahressummen der Gebietsniederschl&auml;ge auf, die mit einer hohen Variabilit&auml;t einhergeht. Eine Betrachtung der saisonalen (monatlichen) &Auml;nderungsbetr&auml;ge von Temperatur, Niederschlag und mittlerem Durchflu&szlig; zwischen Szenarienzeitraum (2061-2099) und Ist-Zustand ergibt in allen Untersuchungsgebieten eine Temperaturzunahme (h&ouml;her im Sommer als im Winter) und eine generelle Zunahme der Niederschlagssummen (mit starken Schwankungen zwischen den Einzelmonaten), die bei der hydrologischen Simulation zu deutlich h&ouml;heren mittleren Durchfl&uuml;ssen von November bis M&auml;rz und leicht erh&ouml;hten mittleren Durchfl&uuml;ssen in den restlichen Monaten f&uuml;hren. Die St&auml;rke der Durchflu&szlig;erh&ouml;hung ist nach den individuellen Klimaszenarien unterschiedlich und im Sommer- bzw. Winterhalbjahr gegenl&auml;ufig ausgepr&auml;gt. Hauptursache f&uuml;r die simulierte starke Zunahme der mittleren Durchfl&uuml;sse im Winterhalbjahr ist die trotz Temperaturerh&ouml;hung der Klimaszenarien winterlich niedrige Evapotranspiration, so da&szlig; erh&ouml;hte Niederschl&auml;ge direkt in erh&ouml;hten Durchflu&szlig; transformiert werden k&ouml;nnen.<br>Der Vergleich der Untersuchungsgebiete zeigt in Einzelmonaten von West nach Ost abnehmende &Auml;nderungsbetr&auml;ge der Niederschlagssummen, die als Hinweis auf die Bedeutung der Kontinentalit&auml;tseinfl&uuml;sse auch unter ge&auml;nderten klimatischen Bedingungen in S&uuml;dwestdeutschland aufgefa&szlig;t werden k&ouml;nnten.<br>Aus den regionalisierten Klimaszenarien werden &Auml;nderungsbetr&auml;ge f&uuml;r die Modulation gemessener Zeitreihen mittels synthetischer Szenarien abgeleitet, die mit einem geringen Rechenaufwand in hydrologische Modellantworten &uuml;berf&uuml;hrt werden k&ouml;nnen. Die direkte Ableitung synthetischer Szenarien aus GCM-Ergebniswerten (bodennahe Temperatur und Gesamtniederschlag) an einzelnen GCM-Gitterpunkten erbrachte unbefriedigende Ergebnisse.<br>Ob, in welcher H&ouml;he und zeitlichen Verteilung die in den (synthetischen) Szenarien verwendeten Niederschlags- und Temperatur&auml;nderungen eintreten werden, kann nur die Zukunft zeigen. Eine Absch&auml;tzung, wie sich die Abflu&szlig;verh&auml;ltnisse und insbesondere die mittleren Durchfl&uuml;sse der Untersuchungsgebiete bei m&ouml;glichen &Auml;nderungen entwickeln w&uuml;rden, kann jedoch heute schon vorgenommen werden. <br><br>Simulationen auf Szenariogrundlagen sind ein Weg, unbekannte zuk&uuml;nftige Randbedingungen sowie regionale Auswirkungen m&ouml;glicher &Auml;nderungen des Klimasystems ausschnittsweise abzusch&auml;tzen und entsprechende Risikominderungsstrategien zu entwickeln. Jegliche Modellierung und Simulation nat&uuml;rlicher Systeme ist jedoch mit betr&auml;chtlichen Unsicherheiten verkn&uuml;pft. Vergleichsweise gro&szlig;e Unsicherheiten sind mit der zuk&uuml;nftigen Entwicklung des sozio&ouml;konomischen Systems und der Komplexit&auml;t des Klimasystems verbunden. Weiterhin haben Unsicherheiten der einzelnen Modellbausteine der Modellkette Emissionsszenarien/Gaszyklusmodelle - Globale Zirkulationsmodelle/Regionalisierung - hydrologisches Modell, die eine Kaskade der Unsicherheiten ergeben, neben Datenunsicherheiten bei der Erfassung hydrometeorologischer Me&szlig;gr&ouml;&szlig;en einen erheblichen Einflu&szlig; auf die Vertrauensw&uuml;rdigkeit der Simulationsergebnisse, die als ein dargestellter Wert eines Ergebnisbandes zu interpretieren sind.<br><br>Der Einsatz <br>(1) robuster hydrologischer Modelle, die insbesondere temperaturbeeinflu&szlig;te Prozesse ad&auml;quat beschreiben,<br>(2) die Verwendung langer Zeitreihen (wenigsten 30 Jahre) von Me&szlig;werten und<br>(3) die gleichzeitige vergleichende Betrachtung von Klimaszenarien, die auf unterschiedlichen GCMs beruhen (und wenn m&ouml;glich, verschiedene Emissionsszenarien ber&uuml;cksichtigen),<br>sollte aus Gr&uuml;nden der wissenschaftlichen Sorgfalt, aber auch der besseren Vergleichbarkeit der Ergebnisse von Regionalstudien im noch jungen Forschungsfeld der Klimafolgenforschung beachtet werden. / This thesis 'Development of runoff in subcatchments of the River Rhine - simulations of the current state and for climate change scenarios' investigates the impacts of possible future climate changes on runoff and runoff regime in selected subcatchments of the River Rhine. The regional climate in the selected subcatchments Mosel (up to gauge Cochem), Sieg (gauge Menden 1) and Main (gauge Kemmern) is affected by the middle mountain ranges.<br><br>In a first step, important model processes are parameterized according to catchment characteristics. A representation of the regional hydrology is then produced by using the hydrological model HBV-D. Based on time series of daily measurements (temperature, precipitation) at stations within the catchment, this representation can be used to realistically simulate time series of runoff and discharge. <br>In all examined areas, the quality of simulations of the calibration and validation periods for the current state (standard period of measurements 01/01/1961-12/31/1999) can be regarded as good to excellent. <br>To aid the catchment-specific, extensive and time-consuming data processing, a working environment for the hydrological model HBV-D has been developed. It is based on program extensions of the geographical information system ArcView and further programs. The working environment HBV-Params contains a graphical interface that gives both experienced hydrologists and students full control and enables them to flexibly derive parameter values and edit parameter and control files. In contrast to previous versions with only rudimentary working environments, HBV-D can therefore be utilized for research as well as for educational purposes. <br><br>In a second step, the current states of areal precipitation, areal temperature and simulated mean discharge (MQ) are compared to the corresponding states for two scenarios of future climate changes (100 years later, 2061-2099). These scenarios are based on simulated global circulations of one model run for each of two global circulation models (GCM). These global circulations are regionalized (downscaled) using a statistical approach into scenario time series of daily values (temperature, precipitation - input for the hydrological model) at control stations within the individual catchments. <br>For the second half of the 21st century, both regionalized climate change scenarios indicate increases in the mean annual areal temperature and mean annual sum of precipitation, along with a high variability of the latter. The seasonal (monthly) changes in temperature, precipitation and mean discharge between scenario state (2061-2099) and current state indicate increases in temperature (higher in summer than in winter) as well as a general increase in precipitation sums (strong fluctuations between individual months). In the hydrological simulations for all investigated catchments, this results in considerably higher mean discharges from November to March and small increases in mean discharge for the other months. The magnitude of the increases in discharge depends on the individual climate change scenario, one showing higher increases than the other during the summer half-year and vice versa for the winter half-year. The main reason for the simulated strong increase in mean discharge during winter half-year is, in spite of higher temperatures, the still relatively low evapotranspiration which allows higher precipitation to be directly transformed into higher discharges. <br>The comparison of the investigated catchments shows decreasing amounts of changes in the sum of precipitation from West to East in individual months. This indicates the importance of continentality under changed climatic conditions in Southwest Germany. <br>For the modification of measured time series (temperature, precipitation), which can be easily converted as synthetic scenarios into simulated hydrological results, amounts of change are derived from regionalized (downscaled) climate change scenarios. The derivation of synthetic scenarios directly from GCM output at individual GCM gridpoints yielded unsatisfactory results. <br>Only the future itself can show whether the timing and amount of changes in temperature and precipitation used in (synthetic) climate change scenarios come close to reality. An assessment of possible developments in runoff regime and specifically mean discharge under possible changed climatic conditions in the investigated catchments is already feasible today. <br><br>Simulations based on scenarios are one way to establish unknown future boundary conditions for the estimation of regional impacts of possible changes of the climate system. Nevertheless, all types of modeling and simulation of natural systems are linked with uncertainties. Rather large uncertainties persist regarding the future development of the socio-economic system and the complexity of the climate system and earth system. Furthermore, besides data uncertainties associated with the measurement of hydro-meteorological values, uncertainties associated with individual components of the model chain emission scenarios/gas cycle model - GCM/regionalization - hydrological model, which form a cascade of uncertainty, have a great influence on the trustworthiness of the simulation results (which are understood as one shown value within a range of results). <br><br>In the young field of climate impact research the use of <br>(1) robust hydrological models that adequately describe temperature-dependent processes,<br>(2) long time series (at least 30 years long) of measurements, <br>(3) concurrent comparisons of climate change scenarios, based on different GCMs (and, if possible, different emission scenarios)<br>should be considered for reasons of scientific thoroughness and to improve comparability of regional impact studies.

O clima é um dos fatores ambientais que impõe os maiores riscos para a atividade agrícola, sendo responsável pelas oscilações e frustrações das safras no Brasil. Em cenário de mudanças climáticas, os atuais níveis de produtividade do milho de 1ª e de 2ª safra deverão ser alterados. Para se avaliar tais impactos, os modelos de simulação de culturas possibilitam estimar o crescimento, o desenvolvimento fenológico e a produtividade das culturas sob ampla gama de condições ambientais e de manejo, sendo, portanto, ferramentas eficientes para esse tipo de estudo. Considerando os possíveis impactos das mudanças climáticas na produtividade da cultura do milho, o presente estudo teve por objetivos: i) realizar preenchimento de falhas em séries de dados meteorológicos e, gerar séries sob projeções futuras do clima a curto (2010- 2039), médio (2040-2069) e longo (2070-2099) prazos, para os cenários de emissão intermediária (RCP4.5) e de alta emissão (RCP8.5); ii) calibrar e validar os modelos DSSAT/CERES-Maize e MONICA para simular a produtividade do milho de 1ª e de 2ª safra, e analisar a sensibilidade desses modelos, identificando os fatores de maior influência na produtividade do milho; iii) aplicar o modelo DSSAT/CERES-Maize, para determinar a produtividade do milho de 1ª e de 2ª safra, em condições de clima atual e futuro, e avaliar possíveis estratégias de manejo, de forma individual e combinada, como épocas de semeadura, ciclo da cultivar, irrigação e adubação nitrogenada, para minimização dos possíveis impactos. Para o preenchimento de falhas em séries de dados meteorológicos, o método de Bristow- Campbell (estimação da radiação solar), e a base em ponto de grade XAVIER foram as que apresentaram melhor desempenho. As projeções de clima futuro evidenciaram condições de clima mais quente, com redução no total acumulado de chuva nas regiões Norte-Nordeste e aumento no Sul do país, e as regiões Sudeste e Centro-Oeste configurando como áreas de transição. Os modelos DSSAT/CERES-Maize e MONICA apresentaram índice de desempenho (c) muito bom para ambas as safras, na estimação da produtividade do milho, com EAM inferior a 450 e 350 kg ha-1 na 1ª e na 2ª safra, respectivamente. Para as estimativas por conjunto, os valores de c foram avaliados como ótimos para as duas safras, com EAM caindo para 276 e 194 kg ha-1, na 1ª e na 2ª safra, respectivamente. Ambos os modelos mostraram sensibilidade às alterações climáticas e de adubação, porém com o modelo DSSAT/CERES-Maize se mostrando mais adequado para estudos de impactos de mudanças climáticas na cultura do milho. As simulações sob clima futuro com o modelo DSSAT/CERES-Maize, mostraram perdas de produtividade em relação aos atuais níveis, variando de 41 a 63% para milho da 1ª safra, e de 58 a 65% para o milho da 2ª safra, com as estratégias de manejo quanto a data de semeadura, ciclo da cultivar, irrigação e adubação nitrogenada mostrando redução das perdas e até mesmo ganhos de produtividade quando adotadas em condições de clima futuro. / Climate is one of the major environmental factors that impose the greatest risks for the agricultural activity, being responsible for the oscillations and frustrations of the crops in Brazil. In a scenario of climate change, the current yield levels of maize growing in-season and offseason should be impacted. In order to evaluate such impacts, crop simulation models allow estimating the growth, phenological development and yield under a wide range of environmental and crop management conditions, being efficient tools for applying to this kind of study. Considering the possible impacts of climate change on maize crop yield, the present study had as objectives: i) to fill gaps in meteorological data series and to generate series under future climate projections in the short (2010-2039), medium (2040-2069) and long (2070-2099) terms periods, for the intermediate emission (RCP4.5) and high emission (RCP8.5) scenarios; ii) to calibrate and validate the DSSAT/CERES-Maize and MONICA models to simulate inseason and off-season maize yield and to analyze the sensitivity of these models, identifying the factors that have the major influence on yield; (iii) to apply the DSSAT/CERES-Maize model to determine maize yield in the in-season and off-season, under current and future climate conditions, and evaluate possible crop management strategies, individually and in combination, such as sowing dates, crop cycle, irrigation and nitrogen fertilization, to minimize possible negative impacts. In order to fill the gaps in meteorological data series, the Bristow- Campbell method (for solar radiation estimation) and the XAVIER daily gridded database were the ones that presented the best performance. The projections of future climate showed warmer climate conditions, with a reduction in the rainfall amounts in the North-Northeast and an increase in the South of the country, with the Southeast and Center-West regions representing transition areas. Both DSSAT/CERES-Maize and MONICA models showed very good performance index (c) in the estimation of maize yield for both seasons, with MAE lower than 450 and 350 kg ha-1 during the in-season and off-season, respectively. For the ensemble estimation, the estimation improve, with optimal performance index, with MAE falling to 276 and 194 kg ha-1, for in-season and off-season maize growing, respectively. Both models showed sensitivity to climate change and fertilization, but with the DSSAT/CERES-Maize model being more suitable for studies of climate change impacts on maize crop. The simulations under future climate with DSSAT/CERES-Maize model showed a yield loss in relation to current levels, ranging from 41 to 63% for in-season, and from 58 to 65% for off-season, with management strategies regarding sowing date, cultivar cycle, irrigation and nitrogen fertilization, showing reduction of losses and even yield gains when adopted in the future climate conditions.

To characterise the habitat preferences of ring ouzel (Turdus torquatus) and blackbird (T. merula) in Switzerland, we adopt species distribution modelling and predict the species’ spatial distribution. We model on two different scales to analyse in how far downscaling leads to a different set of predictors to describe the realised habitat best. While the models on macroscale (grid of one square kilometre) cover the entire country, we select a set of smaller plots for modelling on territory scale. Whereas ring ouzels occur in altitudes above 1’000 m a.s.l. only, blackbirds occur from the lowlands up to the timber line. The altitudinal range overlap of the two species is up to 400 m. Despite both species coexist on macroscale, a direct niche overlap on territory scale is rare. Small-scale differences in vegetation cover and structure seem to play a dominant role for habitat selection. On macroscale however, we observe a high dependency on climatic variables mainly representing the altitudinal range and the related forest structure preferred by the two species. Applying the models for climate change scenarios, we predict a decline of suitable habitat for the ring ouzel with a simultaneous median altitudinal shift of +440 m until 2070. In contrast, the blackbird is predicted to benefit from higher temperatures and expand its range to higher elevations. / Unter Verwendung von Habitatmodellen beschreiben wir die Habitatpräferenz von Amsel (Turdus merula) und Ringdrossel (T. torquatus) in der Schweiz. Mit Hilfe verschiedener Klimaszenarien prognostizieren wir zudem die künftige potentielle Verbreitung beider Arten. Zur Beschreibung eines eventuell auftretenden Skalensprungs, d.h. einer Änderung in der Beschreibungskraft der Variablen auf verschiedenen räumlichen Ebenen, erstellten wir Modelle auf zwei unterschiedlichen Skalen. Während das Modell auf Makroskala mit einer Maschenweite von einem Quadratkilometer die gesamte Schweiz abdeckt, erstellten wir zudem eine Auswahl an Untersuchungsgebieten auf Revierebene. Ringdrosseln zeigen ihren Verbreitungsschwerpunkt in der subalpinen Lage, während Amseln vornehmlich das Tiefland und die Tallagen besiedeln und nur vereinzelt in hohe Lagen vordringen. In einem Gürtel von ungefähr 400 Höhenmetern siedeln beide Arten parallel.Trotz dieses auf der Makroskala erkennbaren Überschneidungsbereiches konnten wir in unserer Untersuchung auf Revierebene, von einer Ausnahme abgesehen, keine Koexistenz beobachten. Kleinräumige Unterschiede in der Habitatstruktur, insbesondere in der Vegetationsbedeckung scheinen demnach für die Habitatselektion von maßgeblicher Bedeutung zu sein. Auf Makroebene hingegen wurde der Einfluss klimatischer Variablen deutlich, die neben der Höhenlage auch dort typische Vegetationsstrukturen widerspiegeln. Wie die Klimaszenarien zeigen, nehmen geeignete Ringdrosselhabitate bei steigenden Temperaturen ab und die Art weicht im Mittel um 440 m in höhere Lagen zurück. Für Amseln scheint sich eine zunehmende Erwärmung jedoch positiv auszuwirken, während das Verbreitungsgebiet im Tiefland beibehalten wird, dringt sie von den Tälern ausgehend zunehmend in höhere Lagen vor.

O clima é um dos fatores ambientais que impõe os maiores riscos para a atividade agrícola, sendo responsável pelas oscilações e frustrações das safras no Brasil. Em cenário de mudanças climáticas, os atuais níveis de produtividade do milho de 1ª e de 2ª safra deverão ser alterados. Para se avaliar tais impactos, os modelos de simulação de culturas possibilitam estimar o crescimento, o desenvolvimento fenológico e a produtividade das culturas sob ampla gama de condições ambientais e de manejo, sendo, portanto, ferramentas eficientes para esse tipo de estudo. Considerando os possíveis impactos das mudanças climáticas na produtividade da cultura do milho, o presente estudo teve por objetivos: i) realizar preenchimento de falhas em séries de dados meteorológicos e, gerar séries sob projeções futuras do clima a curto (2010- 2039), médio (2040-2069) e longo (2070-2099) prazos, para os cenários de emissão intermediária (RCP4.5) e de alta emissão (RCP8.5); ii) calibrar e validar os modelos DSSAT/CERES-Maize e MONICA para simular a produtividade do milho de 1ª e de 2ª safra, e analisar a sensibilidade desses modelos, identificando os fatores de maior influência na produtividade do milho; iii) aplicar o modelo DSSAT/CERES-Maize, para determinar a produtividade do milho de 1ª e de 2ª safra, em condições de clima atual e futuro, e avaliar possíveis estratégias de manejo, de forma individual e combinada, como épocas de semeadura, ciclo da cultivar, irrigação e adubação nitrogenada, para minimização dos possíveis impactos. Para o preenchimento de falhas em séries de dados meteorológicos, o método de Bristow- Campbell (estimação da radiação solar), e a base em ponto de grade XAVIER foram as que apresentaram melhor desempenho. As projeções de clima futuro evidenciaram condições de clima mais quente, com redução no total acumulado de chuva nas regiões Norte-Nordeste e aumento no Sul do país, e as regiões Sudeste e Centro-Oeste configurando como áreas de transição. Os modelos DSSAT/CERES-Maize e MONICA apresentaram índice de desempenho (c) muito bom para ambas as safras, na estimação da produtividade do milho, com EAM inferior a 450 e 350 kg ha-1 na 1ª e na 2ª safra, respectivamente. Para as estimativas por conjunto, os valores de c foram avaliados como ótimos para as duas safras, com EAM caindo para 276 e 194 kg ha-1, na 1ª e na 2ª safra, respectivamente. Ambos os modelos mostraram sensibilidade às alterações climáticas e de adubação, porém com o modelo DSSAT/CERES-Maize se mostrando mais adequado para estudos de impactos de mudanças climáticas na cultura do milho. As simulações sob clima futuro com o modelo DSSAT/CERES-Maize, mostraram perdas de produtividade em relação aos atuais níveis, variando de 41 a 63% para milho da 1ª safra, e de 58 a 65% para o milho da 2ª safra, com as estratégias de manejo quanto a data de semeadura, ciclo da cultivar, irrigação e adubação nitrogenada mostrando redução das perdas e até mesmo ganhos de produtividade quando adotadas em condições de clima futuro. / Climate is one of the major environmental factors that impose the greatest risks for the agricultural activity, being responsible for the oscillations and frustrations of the crops in Brazil. In a scenario of climate change, the current yield levels of maize growing in-season and offseason should be impacted. In order to evaluate such impacts, crop simulation models allow estimating the growth, phenological development and yield under a wide range of environmental and crop management conditions, being efficient tools for applying to this kind of study. Considering the possible impacts of climate change on maize crop yield, the present study had as objectives: i) to fill gaps in meteorological data series and to generate series under future climate projections in the short (2010-2039), medium (2040-2069) and long (2070-2099) terms periods, for the intermediate emission (RCP4.5) and high emission (RCP8.5) scenarios; ii) to calibrate and validate the DSSAT/CERES-Maize and MONICA models to simulate inseason and off-season maize yield and to analyze the sensitivity of these models, identifying the factors that have the major influence on yield; (iii) to apply the DSSAT/CERES-Maize model to determine maize yield in the in-season and off-season, under current and future climate conditions, and evaluate possible crop management strategies, individually and in combination, such as sowing dates, crop cycle, irrigation and nitrogen fertilization, to minimize possible negative impacts. In order to fill the gaps in meteorological data series, the Bristow- Campbell method (for solar radiation estimation) and the XAVIER daily gridded database were the ones that presented the best performance. The projections of future climate showed warmer climate conditions, with a reduction in the rainfall amounts in the North-Northeast and an increase in the South of the country, with the Southeast and Center-West regions representing transition areas. Both DSSAT/CERES-Maize and MONICA models showed very good performance index (c) in the estimation of maize yield for both seasons, with MAE lower than 450 and 350 kg ha-1 during the in-season and off-season, respectively. For the ensemble estimation, the estimation improve, with optimal performance index, with MAE falling to 276 and 194 kg ha-1, for in-season and off-season maize growing, respectively. Both models showed sensitivity to climate change and fertilization, but with the DSSAT/CERES-Maize model being more suitable for studies of climate change impacts on maize crop. The simulations under future climate with DSSAT/CERES-Maize model showed a yield loss in relation to current levels, ranging from 41 to 63% for in-season, and from 58 to 65% for off-season, with management strategies regarding sowing date, cultivar cycle, irrigation and nitrogen fertilization, showing reduction of losses and even yield gains when adopted in the future climate conditions.

Dans un contexte de changement climatique actuel et futur, se traduisantnotamment par un assèchement déjà observé en Méditerranée, cette thèse seconcentre sur l’étude de la variabilité contemporaine et future (1957-2100)des événements de très longs épisodes secs (eTLES) hivernaux (septembreavril)dans le bassin méditerranéen. Une méthodologie originale a été développéeau cours de cette thèse pour appréhender les eTLES comme desévénements climatiques singuliers, caractérisés par des critères de localisation,de durée et d’extension spatiale.Sur la période contemporaine (1957-2013), 76 eTLES ont été détectésdans le bassin méditerranéen. Ces événements sont répartis en 4 principalesconfigurations géographiques : Nord-Est, Ouest, Dispersés & Restreints etSud-Est. Les configurations de types Nord-Est et Ouest sont associées àdes blocages anticycloniques localisés à environ 1 000 km au nord-ouestdes secteurs principalement affectés par les eTLES, favorisant un ciel dégagéet l’absence de précipitations. Les configurations de types Dispersés& Restreints et Sud-Est sont particulières, car la première est caractériséecomme étant une classe résiduelle regroupant des eTLES à faible extensionspatiale répartis dans l’ensemble du bassin, la seconde est caractérisée pardes eTLES saisonniers qui s’insèrent dans la continuité de l’été sec observéà l’est du bassin méditerranéen.Les grands régimes de circulation atmosphérique du domaine Euro-Atlantique montrent un certain contrôle sur les eTLES. Ainsi, la phasepositive de l’oscillation nord-atlantique (NAO+) est le seul régime à êtreclairement favorable au développement d’eTLES sur la quasi-totalité du bassin.Le régime est-atlantique (EA) ne montre pas de contrôle sur les eTLES,atlantic ridge (AR) et la phase négative de l’Oscillation Nord-Atlantique(NAO-) sont deux régimes généralement défavorables aux eTLES. Cependant,des eTLES sont pourtant associés aux régimes AR, EA et NAO-. Pourcela, il faut que ces trois régimes de circulation soient associés à des pressionsatmosphériques légèrement plus élevées au nord-ouest des secteursimpactés par les eTLES, comparativement à leur climatologie respective.Les longues séquences des régimes AR, EA et NAO+, représentatives d’unecertaine stabilité atmosphérique durable dans le temps, sont préférentiellementassociées aux eTLES, contrairement aux courtes séquences de ces troisrégimes. À l’inverse, les longues séquences du régime NAO-, renforçant les basses pressions atmosphériques sur l’Europe et le bassin méditerranéen,sont peu associées aux eTLES.Bien que les deux modèles ALADIN52 et LMDZ4-NEMOMED8 affichentdes résultats différents à plusieurs niveaux, ils s’accordent tout demême à montrer que les eTLES devraient être de plus en plus longs d’icià l’horizon 2100, de façon encore plus marquée pour la trajectoire RCP8.5que RCP4.5. Une analyse multi-modèles comprenant 12 simulations CMIP5montre qu’en moyenne par saison, la pression réduite au niveau de la meraurait tendance à augmenter sur l’océan Atlantique, au large des côtes françaiseset sur le centre du bassin méditerranéen, de façon robuste pour latrajectoire RCP8.5. À l’inverse, la fréquence et la durée des séquences des 4régimes de circulation atmosphérique ne semblent pas être amenées à évoluerd’ici à l’horizon 2100.Une étude est enfin menée pour constater les impacts des eTLES surla production agricole en Espagne. Le nombre de jours d’eTLES impactedavantage les rendements d’orge, de blé et d’avoine (espèces d’hiver et cultivéesau travers d’une agriculture pluviale) que les simples ratios de jourssecs et les cumuls de pluie en Espagne. Une étude de cas réalisée sur deuxsaisons ayant reçu des cumuls de pluie comparables montre qu’en plus desrendements, un eTLES provoque une baisse significative de l’humidité dessols et du débit du fleuve Èbre. / In the context of climate change, as reflected by a dryer Mediterraneanbasin, this thesis focused on the study of the contemporary and future variability(1957-2100) of the wintertime (September to April) very long dryspells events (called VLDSe) in the Mediterranean basin. An original methodologywas developed in this thesis in order to define VLDSe as singularclimatic events, characterized by location, duration and spatial extent criteria.76 VLDSe were detected in the Mediterranean basin on the contemporaryperiod (1957-2013). These events are divided into 4 main geographicalpatterns: North-East, West, Scattered Localized and South-East. North-East and West configurations are associated with anticyclonic conditionslocated approximately 1 000 km northwestern to the areas affected by theVLDSe, favoring a clear sky and no precipitations. The Scattered Localizedand South-East configurations are special: the first one is characterized asa residual class grouping VLDSe with small spatial extent and distributedthroughout the entire basin, and the second one is characterized by seasonalVLDSe which are the continuation of the dry summer observed in the eastof the Mediterranean basin.Euro-Atlantic weather regimes have some control on the VLDSe. Thepositive phase of the north-atlantic oscillation regime (NAO+) is the onlyone that is clearly favorable to the development of VLDSe on almost the entirebasin. The east-atlantic regime (EA) does not show any control on theVLDSe, and the atlantic ridge (AR) and the negative phase of the northatlanticoscillation (NAO-) regimes are generally detrimental to VLDSe. However,some VLDSe can sometimes be associated with AR, EA and NAOregimes.This requires these three weather regimes to be associated withslightly higher atmospheric pressure northwest of the areas impacted bythe VLDSe, compared to their respective climatology. Long duration of theAR, EA and NAO+ regimes, which are coupled with sustained atmosphericstability, are preferentially associated with VLDSe, in contrast to the shortduration. Conversely, the long duration of the NAO- regime, reinforcingthe low atmospheric pressure on Europe and the Mediterranean basin, areweakly associated with VLDSe.Although the two climate models ALADIN52 and LMDZ4-NEMOMED8 differ in several respects, they agree in that VLDSe should be longer by 2100,especially in the RCP8.5 trajectory. A multi-model analysis with 12 CMIP5simulations shows that wintertime sea-level pressure tends to increase in theAtlantic Ocean, off the French coast and in the central the Mediterraneanbasin for the RCP8.5 trajectory. Conversely, the frequency and duration ofthe 4 weather regimes do not show significant trends until the end of the21st century.Finally, a study is carried out to assess the impact of VLDSe on agriculturalproduction in Spain. The number of VLDSe days has a larger impactson the yields of barley, wheat and oats (winter species and cultivatedthrough rainfed agriculture) than the simple ratio of dry days or seasonalrainfall amounts in Spain. A two-season case study, based on seasons withcomparable rainfall amounts, shows that in addition to yields, a VLDSecauses a significant decrease in soil moisture and in the Ebro River flow.