As the decline in Arctic sea ice extent makes this region more accessible, the need is increasing for effective seasonal sea ice forecasting to facilitate operational planning. Recently, coupled global climate models (CGCMs) have been used to address the need for effective sea ice forecasting on seasonal time scales. This thesis assesses the operational utility of the Canadian Seasonal to Interannual Prediction System (CanSIPS) for seasonal sea ice forecasting. This assessment consists of two separate studies. The first uses hindcasting to analyze the skill of two versions of CanSIPS, as well as an intermediate version, on the pan-Arctic as well as regional scales. This approach allows for an overall assessment of the system's skill in addition to providing insight with regards to the features in each version which improved that skill. This study finds that the use of a new initialization procedure for sea ice concentration and thickness improved forecast skill on the pan-Arctic scale as well as in the Central Arctic, Barents Sea, Laptev Sea, and Sea of Okhotsk. This study also shows that the substitution of one of the constituent models in the system improved forecast skill on the pan-Arctic scale as well as in the GIN, Barents, Kara, East Siberian, Chukchi, Bering, and Beaufort Seas. Overall, the new version of CanSIPS was found to be generally more skillful than previous versions. The second study conducts a potential predictability experiment on CanCM4, the constituent CGCM common to all versions of CanSIPS considered in this study. This study follows the methodology introduced by \cite{Bushuk2018} which allows for a more complete assessment of the dependency of potential predictability on initialization month than previous studies and for comparisons to be made between potential predictability and operational skill. This analysis is again done on both the pan-Arctic and regional scale. The findings of this experiment show that CanCM4 has relatively low potential predictability relative to other models and explains results previously presented in a multi-model study by \cite{Day2016}. Further, the characteristics of CanCM4's potential predictability share similarities with other models including greater predictability at longer lead times for winter target months than summer target months, greater predictability in the Atlantic sector than the Pacific sector, and the presence of the spring predictability barrier on the pan-Arctic scale as well as in several regions. The comparison of operational skill to potential predictability provides a general overview of the ``skill gap" which may be closed with improvements in initialization procedures and model physics. This comparison does, however, come with some caveats due to differences in the statistical characteristics of the perfect model and the climate system it represents. Together, the operational skill assessment of different versions of CanSIPS and the potential predictability experiment conducted on one of its constituent models, CanCM4, demonstrate that while room for improvement exists, the recent development of this forecast system has clearly increased its operational utility as a seasonal sea ice forecasting tool. / Graduate

Dans les pays du Sud ruraux et à faibles revenus, la vulnérabilité des zones agricoles pluviales, face à la variabilité pluviométrique, nécessite de trouver des solutions efficaces pour limiter les effets des aléas climatiques sur les récoltes. La prévision des caractéristiques des saisons des pluies quelque temps avant leur démarrage devrait aider à l’établissement de stratégies agricoles d’adaptation aux aléas pluviométriques. C’est à cet objectif que s’attache ce travail, appliqué à l’Afrique de l’Est (Kenya et nord de la Tanzanie), et articulé en 3 parties :- Définir et comprendre le comportement des descripteurs intra saisonniers (DIS) qui feront l’objet de l’étude de prévisibilité. Un travail spécifique a permis le développement d’une nouvelle approche méthodologique dans la manière de définir les démarrages (DSP) et fins (FSP) de saisons des pluies à l’échelle régionale. Cette approche basée sur une analyse multivariée, permet de s’affranchir des choix subjectifs de seuils pluviométriques imposés par les définitions communément utilisée en agro-climatologie. Une analyse de cohérence spatiale à l’échelle inter annuelle montre que, pour les deux saisons des pluies (long rains et short rains), le cumul saisonnier et le nombre de jours de pluie présentent une forte cohérence spatiale, tandis qu’elle est plus modérée pour le démarrage et fin des saisons et faible pour l’intensité quotidienne moyenne.- Analyser la prévisibilité des DIS aux 2 échelles spatiales régionale et locale en s’appuyant sur les simulations numériques du modèle climatique global ECHAM 4.5. Les précipitations quotidiennes simulées par le modèle, même après correction des biais, ne permettent pas d’appréhender correctement la variabilité interannuelle des DIS. Une spécification de la variabilité des DSP et FSP menée par le biais de modèles statistiques construits à partir d’indices climatiques observés, présuppose une prévisibilité modérée des deux descripteurs à l’échelle locale (régionale), et cela quelle que soit la saison. Le développement de modèles statistico-dynamiques à partir des champs de vents simulés par ECHAM 4.5, en mode forcé par les températures marines observées d’une part et prévues d’autre part, montre également des performances faibles localement et régionalement. - Explorer la manière dont la variabilité spatio-temporelle des paramètres climatiques et environnementaux module la variabilité des rendements de sorgho. Ces rendements sont simulés par le modèle agronomique SARRA-H à partir de données climatiques observées (1973-2001) dans 3 stations localisées à différentes altitudes le long des pentes orientales du Mt Kenya. Le cumul précipité et la durée de la saison expliquent une part importante de la variabilité des rendements. D’autres variables apparaissent comme jouant un rôle non négligeable ; le nombre de jours de pluies, l’intensité quotidienne moyenne ou encore certains DIS relatifs à l’organisation temporelle des pluies au sein d’une saison en font partie. L’influence des autres variables météorologiques est seulement visible pour les ‘long rains’ avec une covariation négative entre les rendements et les températures maximales ou, le rayonnement global. La date de semis semble jouer un rôle dans la modulation des rendements pour les stations de haute et moyenne altitudes, mais avec des différences notables entre les deux saisons des pluies. / In Southern countries with rural low income populations, the vulnerability of rainfed agriculture to rainfall variability requires effective solutions to mitigate the effects of climatic hazards on crops. Predicting the characteristics of rainy seasons some time before they start should help the establishment of agricultural adaptation strategies to rainfall hazards. This is the objective of the present study, focused on East Africa (Kenya and northern Tanzania), and divided in three parts:- Define and document intra-seasonal descriptors (ISD) that will be considered in the predictability study. A new methodological approach has been developed in order to define the onset date (ORS) and the cessation date (CRS) of the rainy seasons at the regional level. Based on a multivariate analysis, it eliminates the subjective choice of rainfall thresholds imposed by the definitions commonly used in agroclimatology. An analysis of spatial coherence at interannual time-scale shows that for the two rainy seasons ("long rains" and "short rains"), the seasonal amount and the number of rainy days have a high spatial coherence, while it is medium for the onset and cessation dates and low for the average daily rainfall intensity.- Analyze the predictability of the ISD at both regional and local scales based on numerical simulations from the global climate model ECHAM 4.5. Daily precipitation simulated by the model, even after bias correction, do not correctly capture the IDS interannual variability. A specification of the ORS and CRS variability using statistical models applied to observed climate indices, suggests quite a low predictability of the descriptors at the local (regional) scale, regardless of the season. The development of statistical-dynamical models from wind fields simulated by ECHAM 4.5, in experiments forced by either observed or predicted sea temperatures, also shows quite poor skills locally and regionally.- Explore how the space-time variability of climatic and environmental factors modulate the variations of sorghum yields. Crop yields are simulated by the agronomic model SARRA-H using observed climate data (1973-2001) at three stations located at different elevations along the eastern slopes of Mt Kenya. The seasonal rainfall accumulation and the duration of the season account for a large part of the yields variability. Other rainfall variables also play a significant role, among which the number of rainy days, the average daily intensity and some ISD related to the temporal organization of rainfall within the season. The influence of other meteorological variables is only found during the long rains, in the form of a negative correlation between yields and both maximum temperature and global radiation. Sowing dates seem to play a role in modulating yields for high and medium altitude stations, but with notable differences between the two rainy seasons.