This dissertation examines the performance of an ensemble Kalman filter (EnKF) implemented in a mesoscale model in increasingly realistic contexts from under a perfect model assumption and in the presence of significant model error with synthetic observations to real-world data assimilation in comparison to the three-dimensional variational (3DVar) method via both case study and month-long experiments. The EnKF is shown to be promising for future application in operational data assimilation practice. The EnKF with synthetic observations, which is implemented in the mesoscale model MM5, is very effective in keeping the analysis close to the truth under the perfect model assumption. The EnKF is most effective in reducing larger-scale errors but less effective in reducing errors at smaller, marginally resolvable scales. In the presence of significant model errors from physical parameterization schemes, the EnKF performs reasonably well though sometimes it can be significantly degraded compared to its performance under the perfect model assumption. Using a combination of different physical parameterization schemes in the ensemble (the so-called “multi-scheme” ensemble) can significantly improve filter performance due to the resulting better background error covariance and a smaller ensemble bias. The EnKF performs differently for different flow regimes possibly due to scale- and flow-dependent error growth dynamics and predictability. Real-data (including soundings, profilers and surface observations) are assimilated by directly comparing the EnKF and 3DVar and both are implemented in the Weather Research and Forecasting model. A case study and month-long experiments show that the EnKF is efficient in tracking observations in terms of both prior forecast and posterior analysis. The EnKF performs consistently better than 3DVar for the time period of interest due to the benefit of the EnKF from both using ensemble mean for state estimation and using a flow-dependent background error covariance. Proper covariance inflation and using a multi-scheme ensemble can significantly improve the EnKF performance. Using a multi-scheme ensemble results in larger improvement in thermodynamic variables than in other variables. The 3DVar system can benefit substantially from using a short-term ensemble mean for state estimate. Noticeable improvement is also achieved in 3DVar by including some flow dependence in its background error covariance.

The goal of any data assimilation or history matching algorithm is to enable better reservoir management decisions through the construction of reliable reservoir performance models and the assessment of the underlying uncertainties. A considerable body of research work and enhanced computational capabilities have led to an increased application of robust and efficient history matching algorithms to condition reservoir models to dynamic data. Moreover, there has been a shift towards generating multiple plausible reservoir models in recognition of the significance of the associated uncertainties. This provides for uncertainty analysis in reservoir performance forecasts, enabling better management decisions for reservoir development. Additionally, the increased deployment of permanent well sensors and downhole monitors has led to an increasing interest in maintaining 'live' models that are current and consistent with historical observations. One such data assimilation approach that has gained popularity in the recent past is the Ensemble Kalman Filter (EnKF) (Evensen 2003). It is a Monte Carlo approach to generate a suite of plausible subsurface models conditioned to previously obtained measurements. One advantage of the EnKF is its ability to integrate different types of data at different scales thereby allowing for a framework where all available dynamic data is simultaneously or sequentially utilized to improve estimates of the reservoir model parameters. Of particular interest is the use of partitioning tracer data to infer the location and distribution of target un-swept oil. Due to the difficulty in differentiating the relative effects of spatial variations in fractional flow and fluid saturations and partitioning coefficients on the tracer response, interpretation of partitioning tracer responses is particularly challenging in the presence of mobile oil saturations. The purpose of this research is to improve the performance of the EnKF in parameter estimation for reservoir characterization studies without the use of a large ensemble size so as to keep the algorithm efficient and computationally inexpensive for large, field-scale models. To achieve this, we propose the use of streamline-derived information to mitigate problems associated with the use of the EnKF with small sample sizes and non-linear dynamics in non-Gaussian settings. Following this, we present the application of the EnKF for interpretation of partitioning tracer tests specifically to obtain improved estimates of the spatial distribution of target oil.

Submitted by Pedro Barros (pedro.silvabarros@ufpe.br) on 2018-09-04T21:17:21Z No. of bitstreams: 2 license_rdf: 811 bytes, checksum: e39d27027a6cc9cb039ad269a5db8e34 (MD5) DISSERTAÇÃO Roberto Navarro Pareja.pdf: 3454842 bytes, checksum: add35a3c61e16d12a7de8eec98d2b411 (MD5) / Approved for entry into archive by Alice Araujo (alice.caraujo@ufpe.br) on 2018-09-17T22:46:50Z (GMT) No. of bitstreams: 2 license_rdf: 811 bytes, checksum: e39d27027a6cc9cb039ad269a5db8e34 (MD5) DISSERTAÇÃO Roberto Navarro Pareja.pdf: 3454842 bytes, checksum: add35a3c61e16d12a7de8eec98d2b411 (MD5) / Made available in DSpace on 2018-09-17T22:46:50Z (GMT). No. of bitstreams: 2 license_rdf: 811 bytes, checksum: e39d27027a6cc9cb039ad269a5db8e34 (MD5) DISSERTAÇÃO Roberto Navarro Pareja.pdf: 3454842 bytes, checksum: add35a3c61e16d12a7de8eec98d2b411 (MD5) Previous issue date: 2014-08-26 / CAPES / A simulação de reservatórios é uma importante ferramenta usada pela indústria do petróleo para o gerenciamento de reservatórios. A fim de obter previsões da produção de óleo confiáveis, diferentes propriedades petrofísicas do reservatório, como porosidade e permeabilidade são usadas nos modelos de reservatórios. Porém, medições diretas dessas propriedades são possíveis apenas em alguns poucos poços. Uma forma de melhorar o conhecimento sobre essas propriedades é através do processo de ajuste ao histórico. O ajuste ao histórico consiste em melhorar estimativas de propriedades do reservatório usadas na construção de um modelo de reservatório de forma que as previsões do modelo se aproximem dos dados medidos em campo. Nesta dissertação apresentamos um estudo para o ajuste ao histórico automático baseado em um modelo areal, isto é, que considera o reservatório plano e horizontal, descrito por apenas duas dimensões geométricas, de um reservatório bifásico (óleo/água), onde desejamos estimar a distribuição de permeabilidades do reservatório. Devido à sua simplicidade e eficiência, o método do Filtro de Kalman com Ensembles (EnKF), é usado para assimilar as medições estáticas e dinâmicas, atualizando continuamente as propriedades do reservatório. O EnKF nos últimos anos tem ganhado muita popularidade, é um método de assimilação de dados para modelos dinâmicos não lineares de alta dimensão e portanto adequado para ser usado no ajuste ao histórico de modelos de simulação de reservatórios. O EnKF foi implementado em Matlab e acoplado ao Matlab Reservoir Simulation Toolbox (MRST), que foi desenvolvido pelo SINTEF para simulação de reservatórios, e foi aplicado a dois casos sintéticos simples. Os resultados mostraram que a rotina EnKF funcionou corretamente, mostrando-se que, para muitos dos parâmetros com incerteza inicial, esta foi reduzida a um nível aceitável, para a produção de petróleo e água. / Reservoir simulation is an important tool used by the oil industry for reservoir management. In order to obtain reliable predictions of oil production, different petrophysical properties such as porosity and permeability are used to build the reservoir models. However, direct measurements of these properties are only possible in a few wells. One way to improve the knowledge of these properties is through the history matching process. History matching improves the estimates of reservoir properties used in the construction of the reservoir model so that the model predictions are closer to the measured production of the field. In this paper we present a study for an automatic history matching based upon and two-dimensional model of two-phase (oil/water) reservoir, where we wish to improve the estimate of the distribution of the reservoir permeabilities. Due to its simplicity and efficiency, the method of the Ensemble Kalman Filter (EnKF) is used to assimilate the static and dynamic measurements, continuously updating the properties of the reservoir. The EnKF, in recent years has gained much popularity, as it is a method for dynamic data assimilation for nonlinear models of high dimension and therefore suitable for use in history matching models of reservoir simulations. The EnKF was implemented in Matlab and coupled to Matlab Reservoir Simulation Toolbox (MRST), which was developed by SINTEF for reservoir simulation, and was applied to two simple synthetic cases. The results showed that the EnKF routine works properly, showing that, for many of the parameters and initial uncertainty has been reduced to an acceptable level for the production of oil and water.

El sur de Chile, en particular la X región, cuenta con importantes actividades de salmonicultura y mitilicultura. Estas dos actividades, principalmente orientadas hacia la exportación, representan una parte importante de la economía de esta región, y de Chile entero. Por ser muy ligadas estas actividades al mar, es importante para los productores conocer la circulación oceanográfica en el área de los centros de cultivo. En efecto, esto permite conocer la circulación de los nutrientes necesarios al crecimiento de los salmones o mejillones, de las floraciones algales y de los viruses nefastos a los peces y mariscos (mareas rojas y virus ISA por ejemplo). También permite conocer con anticipación las floraciones a venir, con el fin de tomar medidas de precaución para desplazar salmones y mejillones a aguas sanas. La primera etapa necesaria a este pronóstico de floraciones es la modelación de la circulación oceanográfica. En este trabajo se busca primero realizar una modelación suficientemente precisa de la marea y de las corrientes de marea en la región del mar interior de Chiloé con el fin de proveer la modelación necesaria a los propósitos de explotación mencionados anteriormente. En segundo lugar, con el objetivo de afinar aún más la calidad de la modelación, se pretende asimilar datos de radares HF en esa modelación.\nPara la modelación, se ocupa el modelo hidrodinámico Télémac, que discretiza les ecuaciones de aguas someras (o de Saint-Venant) por elementos finitos. Se ocupa como condiciones de borde los armónicos de marea del atlas global TPXO7, y un tiempo de spin-up para que el modelo llegue al regimen estacionario. El dominio que cubre la modelación es la zona ubicada en latitud entre Puerto Montt y Puerto Chacabuco, y en longitud, entre la costa y el océano profundo afuera de Chiloé. La malla tiene resolución variable, que va de 10km mar afuera a 250m en el canal de Chacao. Se escogió estudiar solamente el proceso dominante que es la marea, así que el modelo no está forzado con datos meteorológicos o con los caudales de los ríos que desembocan en la zona.\nEn cuanto a asimilación, se utilizan datos de corriente radial superficial en el Canal de Chacao. Se usa una metodología de filtro de Kalman de conjuntos (EnKF) para asimilarlos. Para ello, se creó un conjunto haciendo variar el coeficiente de roce (también conocido como coeficiente de Chézy) sobre el fondo, y se buscó crear estructuras coherentes con las escalas de la batimetría. En el etapa de análisis, se ocupa el error (conocido) sobre las mediciones para generar un conjunto de mediciones. Esta etapa permite reducir la varianza del conjunto inicial. Se compararon los resultados entregados por la modelación en ausencia de asimilación con las observaciones de dos mareógrafos en Ancud y Puerto Montt, y se constató que se acercaban bien a las mediciones. Sin embargo, incertitudes sobre la batimetría entre otros impiden mejorar estos resultados aún más, sin recurrir a la asimilación de datos de corriente. Se pudo constatar que efectivemente, el EnKF permite al modelo acercarse más a las mediciones, lo cual es su objetivo principal. Pero otra conclusión interesante destaca. El coeficiente de roce medio después del análisis presenta una estructura coherente con la batimetría local: donde más irregular el fondo, y mayores los gradientes de batimetría, más roce, es decir, menor el coeficiente de roce. Sin embargo, quedan puntos en cuales la asimilación realizada se aleja de la teoría. En efecto, la teoría del EnKF prevé que la varianza de ensemble debe modelar el error entre el modelo y las observaciones, lo cual no es cierto en este trabajo. Además, una de las limitaciones principales de este método es el tiempo de cálculo que requiere: como hay que integrar cada miembro del ensemble, y que la malla que se ocupó tiene bastante puntos, esta etapa es muy larga. Esa limitación impidió hacer muchas pruebas con condiciones de marea distintas. Implementar exitósamente un modelo hidrodinámico y validarlo, como se hizo en este trabajo, es el primer paso para integrar después un modelo biológico, que es la etapa siguiente y crucial en el monitoreo y control de la producción de salmones y mejillones.

Reservoir history matching is becoming increasingly important with the growing demand for higher quality formation characterization and forecasting and the increased complexity and expenses for modern hydrocarbon exploration projects. History matching has long been dominated by adjusting reservoir parameters based solely on well data whose spatial sparse sampling has been a challenge for characterizing the flow properties in areas away from the wells. Geophysical data are widely collected nowadays for reservoir monitoring purposes, but has not yet been fully integrated into history matching and forecasting fluid flow. In this thesis, I present a pioneering approach towards incorporating different time-lapse geophysical data together for enhancing reservoir history matching and uncertainty quantification. The thesis provides several approaches to efficiently integrate multiple geophysical data, analyze the sensitivity of the history matches to observation noise, and examine the framework’s performance in several settings, such as the Norne field in Norway. The results demonstrate the significant improvements in reservoir forecasting and characterization and the synergy effects encountered between the different geophysical data. In particular, the joint use of electromagnetic and seismic data improves the accuracy of forecasting fluid properties, and the usage of electromagnetic data has led to considerably better estimates of hydrocarbon fluid components. For volatile oil and gas reservoirs the joint integration of gravimetric and InSAR data has shown to be beneficial in detecting the influx of water and thereby improving the recovery rate. Summarizing, this thesis makes an important contribution towards integrated reservoir management and multiphysics integration for reservoir history matching.

Les méthodes d'assimilation de données sont en constante évolution pour s'adapter aux problèmes à résoudre dans les multiples domaines d’application. En sciences de l'atmosphère, chaque nouvel algorithme a d'abord été implémenté sur des modèles de prévision numérique du temps avant d'être porté sur des modèles de chimie atmosphérique. Ce fut le cas des méthodes variationnelles 4D et des filtres de Kalman d'ensemble par exemple. La nouvelle génération d'algorithmes variationnels d'ensemble quadridimensionnels (EnVar 4D) ne fait pas exception. Elle a été développée pour tirer partie des deux approches variationnelle et ensembliste et commence à être appliquée au sein des centres opérationnels de prévision numérique du temps, mais n'a à ce jour pas été testée sur des modèles opérationnels de chimie atmosphérique.En effet, la complexité de ces modèles rend difficile la validation de nouvelles méthodes d’assimilation. Il est ainsi nécessaire d'avoir à disposition des modèles d’ordre réduit, qui doivent être en mesure de synthétiser les phénomènes physiques à l'{oe}uvre dans les modèles opérationnels tout en limitant certaines des difficultés liées à ces derniers. Un tel modèle, nommé L95-GRS, a donc été développé. Il associe la météorologie simpliste du modèle de Lorenz-95 à un module de chimie de l'ozone troposphérique avec 7 espèces chimiques. Bien que de faible dimension, il reproduit des phénomènes physiques et chimiques observables en situation réelle. Une méthode d'assimilation de donnée, le lisseur de Kalman d'ensemble itératif (IEnKS), a été appliquée sur ce modèle. Il s'agit d'une méthode EnVar 4D itérative qui résout le problème non-linéaire variationnel complet. Cette application a permis de valider les méthodes EnVar 4D dans un contexte de chimie atmosphérique non-linéaire, mais aussi de soulever les premières limites de telles méthodes.Fort de cette expérience, les résultats ont été étendus au cas d’un modèle réaliste de prévision de pollution atmosphérique. Les méthodes EnVar 4D, via l'IEnKS, ont montré leur potentiel pour tenir compte de la non-linéarité du modèle de chimie dans un contexte maîtrisé, avec des observations synthétiques. Cependant, le passage à des observations réelles d'ozone troposphérique mitige ces résultats et montre la difficulté que représente l'assimilation de données en chimie atmosphérique. En effet, une très forte erreur est associée à ces modèles, provenant de sources d'incertitudes variées. Deux démarches doivent alors être entreprises pour pallier ce problème.Tout d’abord, la méthode d’assimilation doit être en mesure de tenir compte efficacement de l’erreur modèle. Cependant, la majorité des méthodes sont développées en supposant au contraire un modèle parfait. Pour se passer de cette hypothèse, une nouvelle méthode a donc été développée. Nommée IEnKF-Q, elle étend l'IEnKS au cas avec erreur modèle. Elle a été validée sur un modèle jouet, démontrant sa supériorité par rapport à des méthodes d'assimilation adaptées naïvement pour tenir compte de l’erreur modèle.Toutefois, une telle méthode nécessite de connaître la nature et l'amplitude exacte de l'erreur modèle qu'elle doit prendre en compte. Aussi, la deuxième démarche consiste à recourir à des outils statistiques pour quantifier cette erreur modèle. Les algorithmes d'espérance-maximisation, de emph{randomize-then-optimize} naïf et sans biais, un échantillonnage préférentiel fondé sur l'approximation de Laplace, ainsi qu'un échantillonnage avec une méthode de Monte-Carlo par chaînes de Markov, y compris transdimensionnelle, ont ainsi été évalués, étendus et comparés pour estimer l'incertitude liée à la reconstruction du terme source des accidents des centrales nucléaires de Tchernobyl et Fukushima-Daiichi.Cette thèse a donc enrichi le domaine de l'assimilation de données EnVar 4D par ses apports méthodologiques et en ouvrant la voie à l’application de ces méthodes sur les modèles de chimie atmosphérique / Data assimilation methods are constantly evolving to adapt to the various application domains. In atmospheric sciences, each new algorithm has first been implemented on numerical weather prediction models before being ported to atmospheric chemistry models. It has been the case for 4D variational methods and ensemble Kalman filters for instance. The new 4D ensemble variational methods (4D EnVar) are no exception. They were developed to take advantage of both variational and ensemble approaches and they are starting to be used in operational weather prediction centers, but have yet to be tested on operational atmospheric chemistry models.The validation of new data assimilation methods on these models is indeed difficult because of the complexity of such models. It is hence necessary to have at our disposal low-order models capable of synthetically reproducing key physical phenomenons from operational models while limiting some of their hardships. Such a model, called L95-GRS, has therefore been developed. It combines the simple meteorology from the Lorenz-95 model to a tropospheric ozone chemistry module with 7 chemical species. Even though it is of low dimension, it reproduces some of the physical and chemical phenomenons observable in real situations. A data assimilation method, the iterative ensemble Kalman smoother (IEnKS), has been applied to this model. It is an iterative 4D EnVar method which solves the full non-linear variational problem. This application validates 4D EnVar methods in the context of non-linear atmospheric chemistry, but also raises the first limits of such methods.After this experiment, results have been extended to a realistic atmospheric pollution prediction model. 4D EnVar methods, via the IEnKS, have once again shown their potential to take into account the non-linearity of the chemistry model in a controlled environment, with synthetic observations. However, the assimilation of real tropospheric ozone concentrations mitigates these results and shows how hard atmospheric chemistry data assimilation is. A strong model error is indeed attached to these models, stemming from multiple uncertainty sources. Two steps must be taken to tackle this issue.First of all, the data assimilation method used must be able to efficiently take into account the model error. However, most methods are developed under the assumption of a perfect model. To avoid this hypothesis, a new method has then been developed. Called IEnKF-Q, it expands the IEnKS to the model error framework. It has been validated on a low-order model, proving its superiority over data assimilation methods naively adapted to take into account model error.Nevertheless, such methods need to know the exact nature and amplitude of the model error which needs to be accounted for. Therefore, the second step is to use statistical tools to quantify this model error. The expectation-maximization algorithm, the naive and unbiased randomize-then-optimize algorithms, an importance sampling based on a Laplace proposal, and a Markov chain Monte Carlo simulation, potentially transdimensional, have been assessed, expanded, and compared to estimate the uncertainty on the retrieval of the source term of the Chernobyl and Fukushima-Daiichi nuclear power plant accidents.This thesis therefore improves the domain of 4D EnVar data assimilation by its methodological input and by paving the way to applying these methods on atmospheric chemistry models

This thesis presents an investigation of ensemble Gaussianity, the effect of non- Gaussianity on covariance structures, storm-centered data assimilation techniques, and the relationship between commonly used data assimilation variables and the underlying dynamics for the case of Hurricane Humberto. Using an Ensemble Kalman Filter (EnKF), a comparison of data assimilation results in Storm-centered and Eulerian coordinate systems is made. In addition, the extent of the non-Gaussianity of the model ensemble is investigated and quantified. The effect of this non-Gaussianity on covariance structures, which play an integral role in the EnKF data assimilation scheme, is then explored. Finally, the correlation structures calculated from a Weather Research Forecast (WRF) ensemble forecast of several state variables are investigated in order to better understand the dynamics of this rapidly intensifying cyclone. Hurricane Humberto rapidly intensified in the northwestern Gulf of Mexico from a tropical disturbance to a strong category one hurricane with 90 mph winds in 24 hours. Numerical models did not capture the intensification of Humberto well. This could be due in large part to initial condition error, which can be addressed by data assimilation schemes. Because the EnKF scheme is a linear theory developed on the assumption of the normality of the ensemble distribution, non-Gaussianity in the ensemble distribution used could affect the EnKF update. It is shown that multiple state variables do indeed show significant non-Gaussianity through an inspection of statistical moments. In addition, storm-centered data assimilation schemes present an alternative to traditional Eulerian schemes by emphasizing the centrality of the cyclone to the assimilation window. This allows for an update that is most effective in the vicinity of the storm center, which is of most concern in mesoscale events such as Humberto. Finally, the effect of non-Gaussian distributions on covariance structures is examined through data transformations of normal distributions. Various standard transformations of two Gaussian distributions are made. Skewness, kurtosis, and correlation between the two distributions are taken before and after the transformations. It can be seen that there is a relationship between a change in skewness and kurtosis and the correlation between the distributions. These effects are then taken into consideration as the dynamics contributing to the rapid intensification of Humberto are explored through correlation structures.

Groundwater is one of the main sources for drinking water and agricultural activities. Various activities of both humans and nature may lead to groundwater pollution. Very often, pollution, or contamination, of groundwater goes undetected for long periods of time until it begins to affect human health and/or the environment. Cleanup technologies used to remediate pollution can be costly and remediation processes are often protracted. A more practical and feasible way to manage groundwater contamination is to monitor and predict contamination and act as soon as there is risk to the population and the environment. Predicting groundwater contamination requires advanced numerical models of groundwater flow and solute transport. Such numerical modeling is increasingly becoming a reference criterion for water resources assessment and environmental protection. Subsurface numerical models are, however, subject to many sources of uncertainties from unknown parameters and approximate dynamics. This dissertation considers the sequential data assimilation approach and tackles the groundwater contamination problem at the port of Rotterdam in the Netherlands. Industrial concentration data are used to monitor and predict the fate of organic contaminants using a three dimensional coupled flow and reactive transport model. We propose a number of 5 novel assimilation techniques that address different challenges, including prohibitive computational burden, the nonlinearity and coupling of the subsurface dynamics, and the structural and parametric uncertainties. We also investigate the problem of optimal observational designs to optimize the location and the number of wells. The proposed new methods are based on the ensemble Kalman Filter (EnKF), which provides an efficient numerical solution to the Bayesian filtering problem. The dissertation first investigates in depth the popular joint and dual filtering formulations of the state-parameters estimation problem. New methodologies, algorithmically similar, but more efficient numerically, are then proposed based on a more consistent derivation with the Bayesian filtering approach. To reduce computational cost, I further extend the formulation of the hybrid EnKF-variational approach to the state parameter estimation problem and propose an adaptive scheme for the specification of the weights of the flow-dependent and static background covariance matrices. The new adaptive hybrid scheme is shown to provide much better results than the EnKF while using a fraction of the ensemble size. The new methods are implemented and successfully tested with a realistic coupled subsurface and transport-reaction model of the port of Rotterdam by assimilating industrial data on biodegradable chlorinated hydrocarbons. The observational design problem for placing hydrologic wells is subsequently considered and a new efficient solution is proposed that combines concepts from both information theory and data assimilation

Le calage historique permet l'intégration de données acquises après la production dans la construction de modèles de réservoir. Le filtre de Kalman d'ensemble (EnKF) est une méthode d'assimilation (ou calage historique) séquentielle capable d'intégrer les données mesurées dès qu'ils sont obtenus. Ce travail est basé sur l'application de l' EnKF pour le calage historique et est divisé en deux sections principales. La première section traite l'application de la EnKF à plusieurs cas d'études afin de mieux comprendre les avantages et les inconvénients de la méthode. Ces cas d'étude incluent deux cas d'étude synthétiques (un simple et un plutôt complexe), un modèle de faciès et un modèle de réservoir réel. Dans la plupart des cas, la méthode a réussi à reproduire les données mesurées. Les problèmes rencontrés sont expliqués et des solutions possibles sont proposées. La seconde partie traite deux nouveaux algorithmes proposé en combinant l'EnKF avec deux méthodes de paramétrisation: méthode des points pilotes et méthode de déformation graduelle, permettant la préservation les propriétés statistiques de l'ordre de deux (moyenne et covariance). Les deux algorithmes développés sont appliqués au cas d'étude synthétique simple : la première méthode peut réussir avec un nombre suffisant et un bon positionnement des points pilotes. Pour la déformation graduelle, l'application peut réussir si l'ensemble de fond est assez grand.

No description available.