No description available.

Sich ändernde Klima- und Wetterbedingungen in Verbindung mit einer begrenzt ausdehnbaren Ackerfläche werden den Druck auf Nahrungsmittelproduktionssysteme weiter erhöhen. Um dieser Herausforderung gerecht zu werden, ist eine Erhöhung und Stabilisierung der Ernteerträge unverzichtbar. Dies erfordert aber ein tieferes Verständnis der Einflussfaktoren, die auf die Ertragsvariabilität wirken. Diese Dissertation leistet einen Forschungsbeitrag zu Ertragsmodellen in Deutschland, Tansania und auf globaler Ebene. Dazu analysiere und kombiniere ich statistische und prozessbasierte Ertragsmodelle in fünf Schritten: (i) Zunächst entwickele ich einen statistischen Modellansatz, um den Einfluss von Wetter und agronomischem Management auf Winterweizenerträge in Deutschland zu separieren. (ii) Auf der Grundlage dieses Modells erweitere ich die statistischen Methoden und wende sie für Winterweizen und Silomais auf regionale Ebene an. (iii) Diesen erweiterten Modellansatz verwende ich daraufhin zum Testen einer Kreuz-Validierung um zukünftige Ertragsänderungen unter Klimawandel zu projizieren. (iv) Anschließend wird in einer globalen statistischen Anwendung dieses Modell für kurzfristige Ertragsprognosen getestet. (v) Schließlich kombiniere ich für das Fallbeispiel Mais in Tansania statistische und prozessbasierte Ertragsmodelle, um wetterbedingte Ertragsverluste von nicht-wetterbedingten Ertragsverlusten zu separieren. Als Ergebnis lässt sich zusammenfassen, dass der Anteil der wetterbedingten Ertragsvariabilität in Deutschland höher ist als in Tansania. Dementsprechend sind die Ertragsschwankungen in Tansania eher auf das agronomische Management und sozioökonomische Einflüsse zurückzuführen. Für beide Länder stelle ich fest, dass der Anteil der wetterbedingte Ertragsvariabilität auf aggregierter Ebene höher ist als auf regionaler Ebene. Der kombinierte statistisch-prozessbasierte Ansatz zur Bewertung von wetterbedingten Ertragsverlusten kann für Versicherungszwecke genutzt werden. / Due to changing climate and weather patterns in combination with limitations to extend global arable land area, the pressure on food production systems will increase. To cope with this challenge, it will be indispensable to increase and stabilize crop yields. This requires, however, a deeper understanding of the factors influencing crop yield variability. This dissertation contributes to that research need as I further develop and apply crop models to assess regional wheat and maize yield variability in Germany, Tanzania and on a global scale. For this, I analyze and combine statistical and process-based crop models within five steps: (i) First, I develop a statistical crop modeling approach to decompose the influence of weather and agronomic management on winter wheat yields in Germany. (ii) Based on the first step, I expand the statistical methods and apply augmented models for winter wheat and silage maize on a disaggregated level. (iii) Then this model approach is used to investigate an out-of-sample cross validation to demonstrate the models’ capability to project future yield changes under climate change. (iv) In a global statistical application, this models’ capability of projecting yields is tested for short-term yield forecasts. (v) Finally, I combine statistical and process-based crop modeling to decompose weather-related maize yield losses from losses caused by non-weather factors for the case of Tanzania. Across these five steps, I find that the share of weather-related yield variability is higher in Germany than in Tanzania. Accordingly, crop yield variability in Tanzania is to a higher share attributable to agronomic management and socio-economic influences. For both countries, I find that the share of explained weather-related yield variability is higher on an aggregated level than on the regional level. Finally, this combined statistical-process-based approach can be used for assessing weather-related crop yield losses for insurance purposes.

Die Ernährungssicherheit ist bedroht, unter anderem durch den Klimawandel. Eine Zunahme der Wetterextreme kann zu deutlichen Ertragseinbußen führen. Deshalb ist eine Quantifizierung des Klimaeinflußes auf die Landwirtschaft nötig um eine rechtzeitige Anpassung zu ermöglichen. Die vorliegende Dissertation schlägt daher Verbesserungen für Ertragsmodelle in Bezug auf Klimaextreme vor. Der erste Teil ist eine Metastudie zur Strukturierung von Wissen. Eine neue Methode wird zum Aufbau eines enzyklopädischen Netzwerks verwendet. Dieses erlaubt Vorschläge zur Modellverbesserung abzuleiten. Zwei davon, Ozonschäden und extreme Temperaturen, werden folgend behandelt. Der zweite Teil behandelt Ertragseinbußen durch Ozon. Das Ertragsmodell LPJmL wird um Ozonstress erweitert und damit globale Ertragseinbußen bei Weizen und Soja abgeschätzt. Wasserdargebot, Temperatur und CO2-Konzentration werden berücksichtigt, im Gegensatz zu früheren Abschätzungen. Laut Analyse kann Ozon zu Ernteeinbußen von bis zu 50% führen. Der dritte Teil behandelt Schäden durch hohe Temperaturen. Es wird untersucht, inwieweit neun Modelle die Effekte von Hitze auf Mais, Soja und Weizen in den USA abbilden. Das Modellkollektiv kann beobachtete Verluste quantitativ reproduzieren und legt Wasserstress als Ursache dafür nahe. Erhöhte CO2-Konzentrationen können laut Modellen die Ernteeinbußen nicht verringern, im Gegensatz zu gegenwärtigen Überzeugungen. Der vierte Teil enthält ein statistisches Modell, mit dem der Anteil des Wetters an globalen Ertragsschwankungen berechnet wird – unter Berücksichtigung von Hitze- und Froststress. Dieser Anteil wird bei Mais, Soja und Weizen global auf 15-42% beziffert. Weitere Ergebnisse zeigen über 50% Vorhersagekraft des Modells bereits zwei Monate vor der Ernte. Die vorliegende Arbeit stellt die negativen Einflüsse von Ozon und Hitze für die Landwirtschaft und damit die Ernährungssicherheit heraus. Die Vorteile der Anwendung mehrerer Modelltypen werden hervorgehoben. / Agricultural production and thus food security are under pressure, in particular by climate change. Climate extremes are likely to increase and may diminish harvests. Hence it is decisive to quantify such impacts. Consequently, this thesis aims at improving crop models with respect to climate extremes. The first part is a meta-study for structuring knowledge on crop physiology. A novel method is used to build a network-based encyclopedia. This allows for deducing improvement suggestions for crop models. Two of these suggestions (ozone and extreme temperatures) are treated in the following. The second part analyses crop losses from ozone damage. The crop model LPJmL is enhanced by ozone stress and used to simulate global historical wheat and soybean yield losses. Crop water status, temperature and CO2 are considered as modulators of ozone damage – an improvement over previous global assessments. The analysis indicates that ozone can cause yield losses up to occasional 50%. The third part treats effects of high temperatures on yields. It is assessed to what extent nine crop models can reproduce effects of heat on maize, soybean and wheat yields in the US. The model ensemble simulates observed yield losses in the correct quantities and suggests that they stem from water stress. It is hypothesized that future US yields could suffer from heat losses even under elevated CO2, contrary to current convictions. The fourth part describes a statistical model to assess the global share of weather-driven yield variability, considering heat and frost stress. The influence of weather on yield variability of maize, wheat and soybeans is quantified as 15-42% globally. Results also suggest a yield forecasting capacity of more than 50% two months before harvest in several countries. This thesis underlines the negative influence of ozone and high temperature stress on agricultural production and, consequently, food security. The benefits of using diverse types of models are highlighted.

Ein computergestütztes Simulationsmodell für die Schnittblumen-Steinwollkultur der Gerbera jamesonii (H. Bolus ex Hook.) wurde am Beispiel der Sorte 'Moana' erstellt. Aufbauend auf Messdaten morphologischer und physiologischer Parameter an Versuchspflanzen ließ sich das Wachstum von Blättern, Wurzeln und Blumen modellieren. Unter Berücksichtigung äußerer Einflussgrößen konnte mit dem Modell die Entwicklung eines Gerberabestandes hinsichtlich des Wachstums der Blätter, Blumen und Wurzeln einschließlich der wichtigsten Ertragskomponente, dem kumulativen Blumenertrag, in Abhängigkeit von physiologischen Leistungskriterien wie Atmung, Kohlendioxid- und Stickstoffaufnahme simuliert werden. Die Auswertungen des zeitlichen und räumlichen Wachstums der Gerbera der Sorte 'Moana' zeigten, dass die Blattverteilung unter Berücksichtigung des Alters Einfluss auf die Photosyntheseleistung des Bestandes nahm. Ebenso wurde eine Abhängigkeit der Blumenentwicklung von der Blattbildung nachgewiesen. Diese Ergebnisse waren die Basis für die Entwicklung der Modellstruktur. Teile von in der Literatur beschriebenen Modellen konnten erfolgreich in eine übergeordnete Modellstruktur eingebunden werden. Für den Wechsel zwischen vegetativem und generativem Wachstum unter Berücksichtigung der Blattalterung wurden für die Gerbera spezifische Modellfunktionen entwickelt. Bislang nicht verfügbare Gleichungssysteme zur Definition des Einflusses der Umweltparameter auf die Kohlendioxid- und Stickstoffaufnahme waren das Ergebnis von Gaswechsel- und Stickstoffaufnahmeanalysen. Die mit dem Modell berechneten Wachstumswerte für die Blattentwicklung und den Blumenertrag stimmten bei Verwendung gleicher Umweltbedingungen mit denen einer Gewächshauskultur überein. Des weiteren erwies sich das Modell für die Analyse der Auswirkungen eines simulierten Blattbrechvorgangs mit seinen Prognosen als geeignet. Die Verwendung des Modells zur Bestimmung optimaler Klima- und Wachstumsparameter für die Kultur der Sorte 'Moana' führte zu Werten, die bei konstanter Einhaltung eine 60 prozentige Ertragssteigerung gegenüber anfänglich eingesetzten Parametern bedingten. Für den Praktiker stellt das Modell ein Hilfsmittel dar, in einer Gerberakultur mit der Sorte 'Moana' die Sollwerte der Klimaführung bei gegebenen nicht beeinflussbaren Kulturbedingungen zu optimieren, um größtmögliche Erträge zu erzielen. Dem Versuchsansteller in Forschungseinrichtungen wird damit ein Rüstzeug gegeben, klima- und kulturtechnische Einflussgrössen und deren Kombinationen vorab mit dem Modell zu überprüfen, Extremwerte zu identifizieren und somit sinnvolle Versuchsparameter in Abhängigkeit des Versuchszieles zu definieren. / A computer based crop simulation model for a cut flower rockwool cultivation of Gerbera jamesonii (H. Bolus ex Hook.) has been developed for the cultivar 'Moana'. Basing on measurements of morphological and physiological parameters at experimental plants growth of leaves, roots and flowers has been modelled. In consideration of different physical environmental conditions the model was able to simulate the development of a Gerbera crop concerning growth of leaves, flowers and roots including the most important yield component the cumulative flower yield in dependence on physiological efficiency parameters like respiration, carbon dioxide and nitrogen assimilation. The evaluation of time and space dependent growth of Gerberas of the cultivar 'Moana' showed the influence of leaf distribution with regard to leaf age on the photosynthetic efficiency of the crop. Furthermore there was evidence of a dependency of flower growth on leaf development. On the the basis of these results the model structure has been worked out. Elements of already in the relevant literature descripted models were successfully incorporated into the superior model structure. To switch between vegetative and generative growth with regard to leaf aging a special submodel has been developed for the Gerbera crop. Till now not available mathematical functions to define the influence of the environmental conditions on carbon dioxide and nitrogen assimilation resulted form gas exchange and nitrogen uptake analysis. Growth data for leaf development and flower yield calculated by the model corresponded to the observed data when the same values for the environmental parameters have been used. Furthermore the model was able by means of its predictions to analyse the effects of simulated leaf picking. The use of the model to calculate optimized climatic and growth parameters for a cultivation of the cultivar 'Moana' led to values that produced an increase in yield of about 60 percent in comparison to initially applied parameters. For the practice the model represents a tool for seeking optimal combinations of environmental control and crop management strategies for a Gerbera crop using the cultivar 'Moana' to maximise yield. For the research engineer in experimental stations an equipment is provided to test environmental control and crop management strategies in advance to identify extremes and senseful experimental parameters in dependence on the objective of the trial.

Good estimates of soil hydraulic parameters and their distribution in a catchment is essential for crop and hydrological models. Measurements of soil properties by experimental methods are expensive and often time consuming, and in order to account for spatial variability of these parameters in the catchment, it becomes necessary to conduct large number of measurements. Estimation of soil parameters by inverse modelling using observations on either surface soil moisture or crop variables has been successfully attempted in many studies, but difficulties to estimate root zone properties arise for heterogeneous layered soils. Although extensive soil data is becoming more and more available at various scales in the form of digital soil maps there is still a large gap between this available information and the input parameters needed for hydrological models. Inverse modeling has been extensively used but the spatial variability of the parameters and insufficient data sets restrict its applicability at the catchment scale. Use of remote sensed soil moisture data to estimate soil properties using the inverse modeling approach received attention in recent years but yielded only an estimate of the surface soil properties. However, in multilayered and heterogeneous soil systems the estimation of soil properties of different layers yielded poor results due to uncertainties in simulating root zone soil moisture from remote sensed surface soil moisture. Surface soil properties can be estimated by inverse approach using surface soil moisture data retrieved from remote sensing data. Since soil moisture retrieved from remote sensing is representative of the top 5 cm only, inversion of models using surface soil moisture cannot give good estimates of soil properties of deeper layers. Crop variables like biomass and leaf area index are sensitive to the deeper layer soil properties. The main focus of this study is to develop a methodology of estimation of root zone soil hydraulic properties in heterogeneous soils by crop model based inversion techniques. Further the usefulness of the radar soil moisture and leaf area index in retrieving soil hydraulic properties using the develop approach is be tested in different soil and crop combinations. A brief introduction about the soil hydraulic properties and their importance in agro-hydrological model is discussed in Chapter 1. Soil water retention parameters are explained in detail in this chapter. A detailed review of the literature is presented in chapter 2 to establish the state of art on the following: (i) estimation of soil hydraulic properties, (ii) role of crop models in estimating soil hydraulic properties, (iii) retrieval of surface soil moisture using water cloud model from SAR data, (iv) retrieval of leaf area index from SAR (synthetic aperture radar) data and (v) modeling of root zone soil moisture and potential recharge. The thesis proposes a methodology for estimating the root zone soil hydraulic properties viz. field capacity, wilting point and soil thickness. To test the methodology developed in this thesis for estimating the soil hydraulic properties and their uncertainty, three synthetic experiments were conducted by inversion of STICS (Simulateur mulTIdiscplinaire pour les Cultures Standard) model for maize crop using the GLUE (Generalized Likelihood Uncertainty Estimation) approach. The estimability of soil hydraulic properties in a layer-wise heterogeneous soil was examined with several sets of likelihood combinations, using leaf area index, surface soil moisture and above ground biomass. The robustness of the approach is tested with parameter estimation (model inversion) in two different meteorological conditions. The details of the numerical experiments and the several likelihood and meteorological cases examined are given in Chapter 3. The likelihood combination of leaf area index and surface soil moisture provided consistently good estimates of soil hydraulic properties for all soil types and different meteorological cases. Relatively wet year provided better estimates of soil hydraulic properties as compared with a dry year. To validate the approach of estimating root zone soil properties and to test the applicability of the approach in several crops and soil types, field measurements were carried out in the Berambadi experimental watershed located in the Kabini river basin in south India. The profile soil measurements were made for every 10 cm upto 1 m depth. Maize, Marigold, Sunflower, Sorghum and Turmeric crops were monitored during the four year period from 2010 to 2013. Crop growth parameters viz. leaf area index, above ground biomass, yield, phenological stages and crop management activities were measured/monitored at 10 day frequency for all the five crops in the study area. The details of the field experiments performed, the data collected and the results of the model inversion using the ground measured data are given in Chapter 4. The likelihood combination of leaf area index and surface soil moisture provided consistently lower root mean square error (1.45 to 2.63 g/g) and uncertainty in the estimation of soil hydraulic properties for all soil crop and meteorological cases. The uncertainty in the estimation of soil hydraulic properties was lower in the likelihood combination of leaf area index and soil moisture. Estimability of depth of root zone showed sensitivity to the rooting depth. Estimating root zone soil properties at field plot scale using SAR data (incidence angle 24o, wave length 5.3 GHz) of RADARSAT-2 is presented in the Chapter 5. In the first step, an approach of estimating leaf area index from radar vegetation index using the parametric growth curve of leaf area index and the retrieval of soil moisture using water cloud model are given in Chapter 5. The parameters of the growth curve and the leaf area index are generated using a time series of RADARSAT-2 for two years 2010-2011 and 2011-12 for the crops (maize, marigold, sunflower, sorghum and turmeric) considered in this study. The surface soil moisture is retrieved using the water cloud model, which is calibrated using the ground measured values of leaf area index and surface soil moisture for different soils and crops in the study area. The calibration and validation of LAI and water cloud models are discussed in this Chapter. Eventually, the retrieved leaf area index and surface soil moisture from RADARSAT-2 data were used to estimate the soil hydraulic properties and their uncertainty in a similar manner as discussed in Chapter 4 for various crop and soil plots and the results are presented in Chapter 5. The mean and uncertainty in the estimation of soil hydraulic properties using inversion of remote sensing data provided results similar to the estimates from inversion of ground data. The estimates of soil hydraulic properties compared well (R2 of 0.7 to 0.80 and RMSE of 2.1 to 3.16 g/g) with the physically measured vales of the parameters. In Chapter 6, root zone soil moisture and potential recharge are modelled using the STICS model and the soil hydraulic parameters estimated using the RADARSAT-2 data. The potential recharge is highly sensitive to the water holding capacity of rooting zone. Variability in the root zone soil moisture for wet and dry years for different soil types on irrigated and non-irrigated crops were investigated. Potential recharge from different crop and soil types were compared. The uncertainty in the estimation of potential recharge due to uncertainty in the estimation of field capacity is quantified. The root zone soil moisture modeled by STICS showed good agreement with the measured root zone soil moisture in all crop and soil cases. This was tested for both dry and wet year and provides similar results. The temporal variability of root zone soil moisture was also modeled well by the STICS model; the model also predicted well the intra-soil variability of soil moisture of root zone. The results of the modeling of root zone soil moisture and potential recharge are presented in Chapter 6. At the end, in Chapter 7, the major conclusions drawn from the various chapters are summarized.

Un des principaux enjeux de l’amélioration des plantes consiste aujourd’hui à faire face au changement climatique, en assurant un rendement élevé et plus stable dans des systèmes agricoles économes en intrants (eau, fertilisants) et respectueux de l’environnement. Les nouvelles variétés de blé devront non seulement être tolérantes aux stress hydriques et aux fortes températures, mais aussi continuer à être productives avec des apports limités en fertilisation, tout en maintenant une qualité du grain adaptés aux différents usages. De nouvelles méthodes de prédiction des réponses des blés à ces stress sont indispensables pour avancer dans cette direction. Dans ce travail, nous avons tout d’abord identifié les stress qui régissaient les interactions entre génotypes et les environnements (GxE) dans les essais considérés, puis développé un modèle génomique de l’adaptation à un stress environnemental (Factorial Regression genomic Best Linear Unbiased Prediction ou FR-gBLUP), en particulier pour le stress hydrique. En émettant l’hypothèse que plus des variétés de blés sont génétiquement proches, plus elles répondront de façon similaire à un stress environnemental donné, nous avons mesuré par validation croisée des gains de précision de prédiction par rapport à un modèle additif variant entre 3.5% et 15.4%. Des simulations complètent l’étude en démontrant que plus la part de variance expliquée par les réponses au stress considéré est importante, plus le modèle FR-gBLUP apporte un gain de précision. Pour prédire les réponses variétales à un stress particulier, les environnements doivent être finement caractérisés pour les stress limitant le développement des plantes. En nous intéressant plus particulièrement au stress azoté en France, nous avons établi des indicateurs de stress à partir d’un modèle de culture, et les avons comparés à des indicateurs classiques, tels que le type de conduite azotée ou l’azote disponible. Nous avons ainsi mis en évidence l’intérêt des modèles de culture pour caractériser les interactions GxE et pour prédire la réponse génomique au stress azoté, à condition que le signal d’interaction soit assez fort. Au-delà de l’application potentielle de ces méthodes pour la sélection ou la recommandation de variétés de blés plus adaptées ou plus résistantes au changement climatique, les résultats de ce travail démontrent aussi l’intérêt de la complémentarité des approches éco-physiologiques et génétiques. / In a climate change context, assuring high and stable yield in more sustainable agricultural systems is a major challenge for plant breeding. We are aiming for future wheat varieties which will be heat and drought tolerant, and also productive in limited fertilization input environments. New prediction methods of the response to these stresses are needed to move forward. In this study, we first identified stresses that generated interactions between genotypes and environments (GxE) in our experimental trials and then developed a genomic model for adaptation to a particular environmental stress (Factorial Regression genomic Best Linear Unbiased Prediction ou FR-gBLUP), in our case drought. This model hypothesizes that the more individuals are genetically close, the more their response to a stress will resemble. We used cross-validations to measure prediction accuracy gains compared to an additive model and observed gains between 3.5% and 15.4%. Besides, simulation studies showed that the more the variance explained by the responses to the stress is important, the more the FR-gBLUP model will improve the additive model. Furthermore, fine characterization of the stresses limiting the plants’ growth is required to predict varietal responses to a particular stress. We focused on the particular case of nitrogen stress in France. By establishing crop model based stress indicators and comparing them to classical indicators, such as the management system or the available nitrogen, we pointed out the interest of crop model to characterize GxE interactions and to predict the genomic response to nitrogen stress, as long as the GxE interaction signal is strong enough. Beyond the potential applications of these methods for breeding or recommendation for varieties more adapted or tolerant to environmental stresses, this study also raises the interest of coupling eco-physiological and genetics approaches.

Les modèles de culture constituent des outils indispensables pour comprendre l’influence des conditions agropédoclimatiques sur le système sol-plante à différentes échelles spatiales et temporelles. A l’échelle locale de la parcelle agricole, le modèle peut être utilisé dans le cadre de l’agriculture de précision pour optimiser les pratiques de fertilisation azotée de façon à maximiser le rendement ou le revenu tout en minimisant le lessivage des nitrates vers la nappe. Cependant, la pertinence de l’utilisation du modèle repose sur la qualité des prédictions réalisées, basée entre autres sur une bonne détermination des paramètres d’entrée du modèle. Dans le cadre de l’agriculture de précision, les paramètres concernant les propriétés des sols sont les plus délicates à connaître en tout point de la parcelle et il existe très peu de cartes de sols permettant de les déterminer de manière précise. Néanmoins, dans ce contexte, on peut disposer d’observations acquises automatiquement sur l’état du système sol-plante, telles que des images de télédétection, les cartes de rendement ou les mesures de résistivité électrique du sol. Il existe alors une alternative intéressante pour estimer les propriétés des sols à l’échelle de la parcelle qui consiste à inverser le modèle de culture à partir de ces observations pour retrouver les valeurs des propriétés des sols. L’objectif de cette thèse consiste (i) dans un premier temps à analyser les performances d’estimation des propriétés des sols par inversion du modèle STICS à partir de différents jeux d’observations sur des cultures de blé et de betterave sucrière, en mettant en oeuvre une méthode bayésienne de type Importance Sampling, (ii) dans un second temps à mesurer l’amélioration des prédictions de variables agro-environnementales réalisées par le modèle à partir des valeurs estimées des paramètres. Nous montrons que l’analyse de sensibilité globale permet de quantifier la quantité d’information contenue dans les jeux d’observations et les performances réalisées en matière d’estimation des paramètres. Ce sont les propriétés liées au fonctionnement hydrique du sol (humidité à la capacité au champ, profondeur de sol, conditions initiales) qui bénéficient globalement de la meilleure performance d’estimation par inversion. La performance d’estimation, évaluée par comparaison avec l’estimation fournie par l’information a priori, dépend fortement du jeu d’observation et est significativement améliorée lorsque les observations sont faites sur une culture de betterave, les conditions climatiques sont sèches ou la profondeur de sol est faible. Les prédictions agro-environnementales, notamment la quantité et la qualité du rendement, peuvent être grandement améliorées lorsque les propriétés du sol sont estimées par inversion, car les variables prédites par le modèle sont également sensibles aux propriétés liées à l’état hydrique du sol. Pour finir, nous montrons dans un travail exploratoire que la prise en compte d’une information sur la structure spatiale des propriétés du sol fournie par les mesures de résistivité électrique, peut permettre d’améliorer l’estimation spatialisée des propriétés du sol. Les observations acquises automatiquement sur le couvert végétal et la résistivité électrique du sol se révèlent être pertinentes pour estimer les propriétés du sol par inversion du modèle et améliorer les prédictions des variables agro-environnementales sur lesquelles reposent les règles de choix des pratiques agricoles / Dynamic crop models are very useful to predict the behavior of crops in their environment and are widely used in a lot of agro-environmental work. These models have many parameters and their spatial application require a good knowledge of these parameters,especially of the soil parameters. These parameters can be estimated from soil analysis at different points but this is very costly and requires a lot of experimental work. Nevertheless,observations on crops provided by new techniques like remote sensing or yield monitoring, is a possibility for estimating soil parameters through the inversion of crop models. In my work, the STICS crop model is studied for the wheat and the sugar beet and it includes more than 200 parameters. After a previous work based on a large experimental database for calibrate parameters related to the characteristics of the crop, I started my study with a global sensitivity analysis of the observed variables (leaf area index LAI and absorbed nitrogen QN provided by remote sensing data, and yield at harvest provided by yield monitoring) to the soil parameters, in order to determine which of them have to be estimated. This study was made in different climatic and agronomic conditions and it reveals that 7 soil parameters (4 related to the water and 3 related to the nitrogen) have a clearly influence on the variance of the observed variables and have to be therefore estimated. For estimating these 7 soil parameters, I chose a Bayesian data assimilation method (because I have prior information on these parameters) named Importance Sampling by using observations, on wheat and sugar beet crop, of LAI and QN at various dates and yield at harvest acquired on different climatic and agronomic conditions. The quality of parameter estimation is then determined by comparing the result of parameter estimation with only prio rinformation and the result with the posterior information provided by the Bayesian data assimilation method. The result of the parameter estimation show that the whole set of parameter has a better quality of estimation when observations on sugar beet are assimilated. At the same time, global sensitivity analysis of the observed variables to the 7 soil parameters have been performed, allowing me to build a criterion based on sensitivity indices (provided by the global sensitivity analysis) able to rank the parameters with respect to their quality of estimate. This criterion constitutes an interesting tool for determining which parameters it is possible to estimate to reduce probably the uncertainties on the predictions. The prediction of the crop behaviour when estimating the soil parameters is then studied. Indeed, the quality of prediction of agro-environmental variables of the STICS crop model (yield, protein of the grain and nitrogen balance at harvest) is determined by comparing the result of the prediction using the prior information on the parameters and the result using the posterior information. As for the estimation of soil parameters, the prediction of the variable is made on different climatic and agronomic conditions. According to the result of parameter estimation, assimilating observations on sugar beet lead to a better quality ofprediction of the variables than observations on wheat. It was also shown that the number ofcrop seasons observed and the number of observations improve the quality of the prediction

L'architecture des plantes est le résultat combiné des développements des structures topologique et géométrique qui interviennent dans l'acquisition de la biomasse et sa répartition sous l'influence des processus physiologiques. Pourtant cet aspect a été longtemps négligé dans la communauté des modèles dynamiques. Récemment les modèles structures fonction se sont montrés pertinents pour prendre en compte des questions comme les interactions plantes environnement (l'interception de la lumière), les interactions entre croissance et développement (répartition de la biomasse) en se plaçant au niveau de l'organe. Cependant les couts en calcul de la simulation numérique de ces processus rendent les applications impraticables en agriculture. Cette thèse vise a combiner le modèle structure fonction Greenlab avec d'une part un modèle de culture et d'autre part un modèle forestier basés sur le peuplement afin d'y introduire le concept d'architecture des plantes. Le modèle de culture Pilote fournit des prédictions de récoltes basés sur les paramètres de l'environnement (radiation, précipitations) et l'indice foliaire et l'indice de récolte. Une étude sur Maïs conjointe entre Pilote et GreenLab a permis d'expliciter en détail les paramètres de la production. Les indices foliaires et de récolte dépendent directement des paramètres sources puits, et la variabilité individuelle entre plantes est explicitée directement par les variations des retards a la germination et celles des surfaces disponibles par plantes (compétition spatiale). Tous ces paramétrés peuvent être calibré par méthodes inverses. Ainsi la jonction des deux types de modèles est réalisée au niveau du passage de la plante au peuplement.Une autre étude conjointe a été effectuée avec le modèle forestier empirique PNN qui modélise la croissance des peuplements forestiers de Pins noirs. A partir des données statistiques classiques sur les mesures de troncs et de houppiers, combinées avec les connaissances architecturales du Pin issues d'AMAP, GreenLab peut restituer l'architecture de l'arbre et visualiser des scenarios de sylviculture incorporant des élagages. Le procédé va jusqu'à l'obtention d'images de synthèse réalistes des peuplements. En conséquence il semble efficace de coupler les modèles de cultures et les modèles forestiers qui intègrent les connaissances écophysiologiques au niveau peuplement avec les modèles structures fonctions qui intègrent ces connaissances au niveau de l'architecture de la plante. Le modèle GreenLab par ses affinités avec ces deux types de modèles et ses performances en calcul, permet d'apporter un complément d'information essentiel sur la description du fonctionnement d'un peuplement tant du point de vue développement, que du point de vue des relations sources puits dans la plante. Enfin le modèle couplé a une plateforme comme Xplo (AMAP) permet en plus une simulation réaliste 3D du peuplement végétal aux divers stades de la croissance. / Plant architecture implies the development of both topological and geometrical structure over time, which determines resource acquisition, in the meantime interacts with physiological processes. However it has long been overlooked in traditional community dynamics models. Based on plant architecture, functional-structural plant models (FSPM) have showed their particular capability in addressing questions like interactions between plant and environment (e.g. light interception), between structure development and growth (e.g. carbon allocation), as they take into account morphogenesis with organ-level explicit descriptions. Anyway, high demand of time and memory for simulation and inverse calculation prevents FSPM from further agricultural or sylvicultural practice. This thesis attempts the combination of a mathematic FSPM GreenLab and a crop model or an empirical forest model (EFM) to introduce individual-based architectural support for community growth study. In the case of maize, disagreement from stand level (by crop model PILOTE) and individual level (by GreenLab) growth simulations implies different emergence time of individuals, which is used to quantify the distribution. By supposing that theoretical projective area (Sp) is determined by the growth situation and the final size of individual architecture, the variance of Sp is reversely computed with the variance of organ compartment measurements to characterize individual variability. In the case of Black pine, architecture dynamics built in GreenLab according to Rauh's model (architecture model for pine tree) are adapted to the simulation of an EFM PNN. As a consequence, thinning scenarios are well incorporated in the final stand visualization. From these preliminary applications, following conclusions can be drawn: (i) FSPM is able to provide individual performances (i.e. organ development and expansion) inside an area of crop field for crop models. (ii) The crop model may regulate the combined form of individuals from integral level. Both aspects are significant to deepen understanding of stand growth. (iii) Architecture conceptions integrated in FSPM may be adapted to EFM simulations for a data-driven visualization. (iv) EFM can guarantee ecological/sylvicultural function for 3D stand visualization. To take into consideration biomass processes, additional observations are needed. As models are independent in combinations, the same methods can be extended to linkage with other stand models.

Die Anzahl der unterernährten Menschen in der Welt steigt seit 2017 wieder an. Der Klimawandel wird den Druck auf die Landwirtschaft und die Ernährungssicherheit weiter erhöhen, insbesondere für kleinbäuerliche und von Subsistenzwirtschaft geprägte Agrarsysteme in den Tropen. Um die Widerstandsfähigkeit der Ernährungssysteme und die Ernährungssicherheit zu stärken, bedarf es eines Klimarisikomanagements und Klimaanpassung. Dies kann sowohl die Antizipation als auch die Reaktion auf die Auswirkungen der globalen Erwärmung ermöglichen. Eine zentrale Rolle spielen in dieser Hinsicht landwirtschaftliche Modelle. Sie können die Reaktionen von Pflanzen auf Veränderungen in den Klimabedingungen quantifizieren und damit Risiken identifizieren. Diese Dissertation demonstriert anhand dreier in Peru, in Tansania und in Burkina Faso durchgeführten Fallstudien, wie statistische Ertragsmodelle das Klimarisikomanagement und die Anpassung in der tropischen Landwirtschaft unterstützen können. Während die erste Studie zeigt, wie Klimaanpassungsbestrebungen unterstützt werden können, werden in Studie zwei und drei statistische Modelle genutzt, um Ertrags- und Produktionsvorhersagen zu erstellen. Die Ergebnisse können dazu beitragen, Frühwarnsysteme für Ernährungsunsicherheit zu unterstützen. In den drei Veröffentlichungen werden neue Ansätze statistischer Ertragsmodellierung auf verschiedenen räumlichen Ebenen vorgestellt. Ein besonderer Fokus liegt hierbei auf der Weiterentwicklung von bisherigen Ertragsvorhersagen, insbesondere in Bezug auf unabhängige Modellvalidierungen, eine stärkere Berücksichtigung von Wetterextremen und die Übertragbarkeit der Modelle auf andere Regionen. / The number of undernourished people in the world has been increasing since 2017. Climate change will further exacerbate pressure on agriculture and food security, particularly for smallholder and subsistence-based farming systems in the tropics. Anticipating and responding to global warming through climate risk management is needed to increase the resilience of food systems and food security. Crop models play an indispensable role in this regard. They allow quantifying crop responses to changes in climatic conditions and thus identify risks. This dissertation demonstrates how statistical crop modelling can inform climate risk management and adaptation in tropical agriculture in the case studies of Peru, Tanzania and Burkina Faso. While the first study shows how statistical crop models can support climate adaptation, studies two and three provide yield and production forecasts. The results can contribute to supporting early warning systems on food insecurity. The three publications present novel approaches of statistical yield modelling at different spatial scales. A particular focus is on further developing existing yield forecasts, especially with regard to independent rigorous model validations, improved consideration of weather extremes, and the transferability of the models to other regions.