Diffuse reflectance spectroscopy (DRS) is a rapid proximal-sensing method that is being used more and more in laboratory settings to measure soil properties. Diffuse reflectance spectroscopy research that has been completed in laboratories shows promising results, but very little has been reported on how DRS will work in a field setting on soils scanned in situ. Seventy-two soil cores were obtained from six fields in Erath and Comanche County, Texas. Each soil core was scanned with a visible near-infrared (VNIR) spectrometer with a spectral range of 350-2500 nm in four different combinations of moisture content and pre-treatment: field-moist in situ, air-dried in situ, field-moist smeared in situ, and air-dried ground. Water potential was measured for the field-moist in situ scans. The VNIR spectra were used to predict total and fine clay content, water potential, organic C, and inorganic C of the soil using partial least squares (PLS) regression. The PLS model was validated with data 30% of the original soil cores that were randomly selected and not used in the calibration model. The root mean squared deviation (RMSD) of the air-dry ground samples were within the in situ RMSD and comparable to literature values for each soil property. The validation data set had a total clay content root mean squared deviation (RMSD) of 61 g kg-1 and 41 g kg-1 for the field-moist and air-dried in situ cores, respectively. The organic C validation data set had a RMSD of 5.8 g kg-1 and 4.6 g kg-1 for the field-moist and air-dried in situ cores, respectively. The RMSD values for inorganic C were 10.1 g kg-1 and 8.3 g kg-1 for the field moist and air-dried in situ scans, respectively. Smearing the samples increased the uncertainty of the predictions for clay content, organic C, and inorganic C. Water potential did not improve model predictions, nor did it correlate with the VNIR spectra; r2-values were below 0.31. These results show that DRS is an acceptable technique to measure selected soil properties in-situ at varying water contents and from different parent materials.

Diffuse reflectance spectroscopy (DRS) is a rapid proximal-sensing method that is being used more and more in laboratory settings to measure soil properties. Diffuse reflectance spectroscopy research that has been completed in laboratories shows promising results, but very little has been reported on how DRS will work in a field setting on soils scanned in situ. Seventy-two soil cores were obtained from six fields in Erath and Comanche County, Texas. Each soil core was scanned with a visible near-infrared (VNIR) spectrometer with a spectral range of 350-2500 nm in four different combinations of moisture content and pre-treatment: field-moist in situ, air-dried in situ, field-moist smeared in situ, and air-dried ground. Water potential was measured for the field-moist in situ scans. The VNIR spectra were used to predict total and fine clay content, water potential, organic C, and inorganic C of the soil using partial least squares (PLS) regression. The PLS model was validated with data 30% of the original soil cores that were randomly selected and not used in the calibration model. The root mean squared deviation (RMSD) of the air-dry ground samples were within the in situ RMSD and comparable to literature values for each soil property. The validation data set had a total clay content root mean squared deviation (RMSD) of 61 g kg-1 and 41 g kg-1 for the field-moist and air-dried in situ cores, respectively. The organic C validation data set had a RMSD of 5.8 g kg-1 and 4.6 g kg-1 for the field-moist and air-dried in situ cores, respectively. The RMSD values for inorganic C were 10.1 g kg-1 and 8.3 g kg-1 for the field moist and air-dried in situ scans, respectively. Smearing the samples increased the uncertainty of the predictions for clay content, organic C, and inorganic C. Water potential did not improve model predictions, nor did it correlate with the VNIR spectra; r2-values were below 0.31. These results show that DRS is an acceptable technique to measure selected soil properties in-situ at varying water contents and from different parent materials.

Made available in DSpace on 2014-06-11T19:23:24Z (GMT). No. of bitstreams: 0 Previous issue date: 2010-01-22Bitstream added on 2014-06-13T20:10:46Z : No. of bitstreams: 1 rodrigues_ms_me_jabo.pdf: 1277888 bytes, checksum: fca43b2044e8d0aa95f9d93c7040361b (MD5) / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) / Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP) / Diversos fatores são responsáveis pela variabilidade espacial na produtividade das culturas e dentre estes estão os atributos do solo. Contudo, o que se tem observado, geralmente, é uma baixa correlação entre os mapas de produtividade e os mapas de atributos do solo. Uma hipótese para essa baixa correlação é a diferença na amostragem de cada uma dessas variáveis. O trabalho objetivou estudar o padrão de distribuição espacial dos atributos do solo e da produtividade do milho e as relações de causas e efeitos utilizando-se diferentes intensidades de amostragem. Os dados foram coletados seguindo uma grade amostral contendo 100 pontos georreferenciados. Os pontos foram dispostos em intervalos equidistantes, sendo de 10 m no sentido das linhas da cultura, ao longo de quatro transeções paralelas com 250 m de comprimento, distanciadas entre si 4,5 m, formando um retângulo com quatro colunas e 25 linhas, constituindo os 100 pontos de amostragem. Cada ponto amostral foi composto por cinco linhas de 10 m da cultura, constituindo uma célula amostral de 45 m2. Em cada ponto amostral retiraram-se amostras de solo nas profundidades de 0-0,1 e 0,1-0,2 m. Foram avaliados os atributos do solo textura, pH, teores de matéria orgânica, P, K+, Ca2+, Mg2+, H+Al, soma de bases, capacidade de troca de cátions, saturação por bases, densidade, porosidade total, macroporosidade e microporosidade. Foram estabelecidas intensidades de amostragens por meio de eliminação de pontos intermediários partindo da amostragem inicial, que continha 100 pontos, obtendo-se conjunto de dados com 100, 75, 50 e 25 pontos. Realizou-se a análise estatística descritiva dos dados e geoestatística. Baseado na análise de correlação simples, regressão múltipla ‘stepwise’ e semelhança dos mapas de isolinhas, os atributos que mais se correlacionaram... / Soil attributes play an important role in spatial variability of crop yield. However, what has been observed, generally, is a low correlation between the yield maps and maps of soil attributes. One hypothesis for this low correlation is the difference in the sampling of each of these variables. This research had as objective to study spatial distribution pattern of soil attributes affecting corn yield and their causeeffect relationship, for different sampling intensities. Spatial variability was determined for 100 georeferenced sample points from a sampling grid. Sample points were arranged in regular 10-m intervals along the rows and in four parallel 250-m long and 4.5-m spaced transects, resulting in a 4-column and 25-row rectangle (100 sample points). Each sample point consisted of five 10-m rows, totaling 45 m2. Samples were taken from the depths 0-0,1 and 0,1-0,2 m. Soil texture attributes, pH, organic matter content, P, K+, Ca2+, Mg2+ and H+Al levels, sum of bases, cation exchange capacity, base saturation, bulk density, total porosity, macroporosity and microporosity were evaluated. Sampling intensities were established by eliminating intermediary points from the initial 100-point sampling, resulting data sets of 100, 75, 50 and 25 points. Statistical analysis consisted of data description and geostatistics. Based on the simple correlation analysis, stepwise multiple regression and similarity among isoline maps, the attributes that showed stronger spatial correlation to corn yield were base saturation and clay content. For these attributes, data sets of 75 and 100 points showed higher similarity in the spatial distribution pattern for corn yield. The analysis of cross-semivariograms showed that the best correlation between corn yield and soil attributes was obtained with a sampling intensity of 100 points. It was possible to confirm the hypothesis that, when performing the soil... (Complete abstract click electronic access below)

High total organic carbon (TOC) and low clay content are two criteria to identify the "sweet spots" in shale gas plays. Recently, machine learning has been proved to be effective to estimate TOC and clay from well loggings. The remaining questions are what algorithm we should choose in the first place and whether we can improve the already built models. Automatic machine learning (AutoML) appears as a promising tool to solve those realistic questions by training multiple models and compares them automatically. Two wells with conventional well loggings and elemental capture spectroscopy are selected from a shale gas play to test the AutoML's ability in TOC and clay estimation. TOC and clay content are extracted from the Schlumberger's ELAN interpretation and calibrated to cores. Generalizability is proved in the blind test well and the mean absolute test errors for TOC and clay estimation are 0.23% and 3.77%. 829 data points are used to generate the final models with the train-test ratio of 75:25. The mean absolute test errors are 0.26% and 2.68% for TOC and clay, respectively, which are very low for TOC ranging from 0-6% and clay from 35-65%. The results show the AutoML's success and efficiency in the estimation. The trained models are interpreted to understand the variables effects in predictions. 235 wells are selected through data quality checking and feed into the models to create TOC and clay distribution maps. The maps provide guidance on where to drill a new well for higher shale gas production. / Master of Science / Locating "sweet spots", where the shale gas production is much higher than the average areas, is critical for a shale reservoir's successful commercial exploitation. Among the properties of shale, total organic carbon (TOC) and clay content are often selected to evaluate the gas production potential. For TOC and clay estimation, multiple machine learning models have been tested in recent studies and are proved successful. The questions are what algorithm to choose for a specific task and whether the already built models can be improved. Automatic machine learning (AutoML) has the potential to solve the problems by automatically training multiple models and comparing them to achieve the best performance. In our study, AutoML is tested to estimate TOC and clay using data from two gas wells in a shale gas field. First, one well is treated as blind test well and the other is used as trained well to examine the generalizability. The mean absolute errors for TOC and clay content are 0.23% and 3.77%, indicating reliable generalization. Final models are built using 829 data points which are split into train-test sets with the ratio of 75:25. The mean absolute test errors are 0.26% and 2.68% for TOC and clay, respectively, which are very low for TOC ranging from 0-6% and clay from 35-65%. Moreover, AutoML requires very limited human efforts and liberate researchers or engineers from tedious parameter-tuning process that is the critical part of machine learning. Trained models are interpreted to understand the mechanism behind the models. Distribution maps of TOC and clay are created by selecting 235 gas wells that pass the data quality checking, feeding them into trained models, and interpolating. The maps provide guidance on where to drill a new well for higher shale gas production.

Yes / This study aims to provide a thorough evaluation for the changes in the microstructure and evolution of strength of highly reactive clays that were treated with 7 % lime over a period of curing time as a function of the mixing moisture content. Three series of testing were carried out on specimens with 100 %, 85 % and 75 % of bentonite content and prepared with different moisture content of 10, 20, 30 and 40 % above the corresponding optimum moisture content. Specimens of 100 % bentonite were treated with 7 % of lime, compacted to achieve a predetermined dry unit weight and cured at temperatures of 20 OC and 40 OC for up to 28 days whereas the specimens with 85 % and 75 % of bentonite content were prepared by the addition of sand and were cured at 20 oC for up to 7 days. Unconfined Compressive Strength tests and Scanning Electron Microscopy were conducted to observe the strength and the microstructural changes resulting from increasing mixing moisture content. California Bearing Ratio and Resilient Modulus were correspondingly determined based on correlations with the Unconfined Compressive Strength. The failure pattern was also studied to better understand the ultimate behaviour of lime stabilised clays. The results revealed that the strength of treated bentonite increased with the increase in the moisture content up to 30 % above the corresponding optimum moisture content and with increasing the curing time and temperature. Nevertheless, substituting bentonite with sand on the specimen resulted in a significant reduction on the attained strength. Furthermore, the results of California Bearing Ratio and Resilient Modulus showed that values for both parameters are significantly enhanced with lime treatment. The microstructural analysis provided visual evidence to the improved strength in which the pozzolanic reaction was found to be significantly affected by the amount of moisture in the mixture. The results suggested that compacting lime treated expansive clays with moisture content moderately higher than the optimum moisture content would result in a significant enhancement to the attained strength over the period of curing.

Prediction of rock and fluid properties such as porosity, clay content, and water saturation is essential for exploration and development of hydrocarbon reservoirs. Rock and fluid property maps obtained from such predictions can be used for optimal selection of well locations for reservoir development and production enhancement. Seismic data are usually the only source of information available throughout a field that can be used to predict the 3D distribution of properties with appropriate spatial resolution. The main challenge in inferring properties from seismic data is the ambiguous nature of geophysical information. Therefore, any estimate of rock and fluid property maps derived from seismic data must also represent its associated uncertainty. In this study we develop a computationally efficient mathematical technique based on neural networks to integrate measured data and a priori information in order to reduce the uncertainty in rock and fluid properties in a reservoir. The post inversion (a posteriori) information about rock and fluid properties are represented by the joint probability density function (PDF) of porosity, clay content, and water saturation. In this technique the a posteriori PDF is modeled by a weighted sum of Gaussian PDF’s. A so-called mixture density network (MDN) estimates the weights, mean vector, and covariance matrix of the Gaussians given any measured data set. We solve several inverse problems with the MDN and compare results with Monte Carlo (MC) sampling solution and show that the MDN inversion technique provides good estimate of the MC sampling solution. However, the computational cost of training and using the neural network is much lower than solution found by MC sampling (more than a factor of 104 in some cases). We also discuss the design, implementation, and training procedure of the MDN, and its limitations in estimating the solution of an inverse problem. In this thesis we focus on data from a deep offshore field in Africa. Our goal is to apply the MDN inversion technique to obtain maps of petrophysical properties (i.e., porosity, clay content, water saturation), and petrophysical facies from 3D seismic data. Petrophysical facies (i.e., non-reservoir, oil- and brine-saturated reservoir facies) are defined probabilistically based on geological information and values of the petrophysical parameters. First, we investigate the relationship (i.e., petrophysical forward function) between compressional- and shear-wave velocity and petrophysical parameters. The petrophysical forward function depends on different properties of rocks and varies from one rock type to another. Therefore, after acquisition of well logs or seismic data from a geological setting the petrophysical forward function must be calibrated with data and observations. The uncertainty of the petrophysical forward function comes from uncertainty in measurements and uncertainty about the type of facies. We present a method to construct the petrophysical forward function with its associated uncertainty from the both sources above. The results show that introducing uncertainty in facies improves the accuracy of the petrophysical forward function predictions. Then, we apply the MDN inversion method to solve four different petrophysical inverse problems. In particular, we invert P- and S-wave impedance logs for the joint PDF of porosity, clay content, and water saturation using a calibrated petrophysical forward function. Results show that posterior PDF of the model parameters provides reasonable estimates of measured well logs. Errors in the posterior PDF are mainly due to errors in the petrophysical forward function. Finally, we apply the MDN inversion method to predict 3D petrophysical properties from attributes of seismic data. In this application, the inversion objective is to estimate the joint PDF of porosity, clay content, and water saturation at each point in the reservoir, from the compressional- and shear-wave-impedance obtained from the inversion of AVO seismic data. Uncertainty in the a posteriori PDF of the model parameters are due to different sources such as variations in effective pressure, bulk modulus and density of hydrocarbon, uncertainty of the petrophysical forward function, and random noise in recorded data. Results show that the standard deviations of all model parameters are reduced after inversion, which shows that the inversion process provides information about all parameters. We also applied the result of the petrophysical inversion to estimate the 3D probability maps of non-reservoir facies, brine- and oil-saturated reservoir facies. The accuracy of the predicted oil-saturated facies at the well location is good, but due to errors in the petrophysical inversion the predicted non-reservoir and brine-saturated facies are ambiguous. Although the accuracy of results may vary due to different sources of error in different applications, the fast, probabilistic method of solving non-linear inverse problems developed in this study can be applied to invert well logs and large seismic data sets for petrophysical parameters in different applications.

Coordenação de Aperfeiçoamento de Pessoal de Nível Superior / The objective of this work was to modeling runoff in soils with different physical characteristics, with three simulated rainfall intensities (30, 60 and 120 mm h-1) in Rio Grande do Sul and Uruguay. For simulated rainfall was used a portable rainfall simulator of multiple nozzles. Six plots were delimited by metal sheet borders (0,5 m2), with a runoff collector in the lower part. The runoff was determinate each five minutes. On each soil was determinate initial time and rate of runoff, rainfall (total, time and intensities), direction of slope, crop residue and cover percentage, soil densities (bulk and particle), soil porosity (bulk, macro and micro), textural fractions (clay, silt and sand), initial and saturated soil moisture. The runoff was estimated with Smith s modified model. The model parameters were adjusted by multivariate equations. The runoff losses accumulated in Uruguay was 64, 32, 30 e 15% from total rain, for Vertissolo 1, Chernossolo, Argissolo 2 e Vertissolo 2, respectively. The runoff losses accumulated in Rio Grande do Sul was 67, 45 and 27% from total rain, for Argissolo 1, Neossolo e Latossolo, respectively. In most of the cases, the initial runoff time decreased with increasing soil moisture and rain intensity, independently of the soil surface conditions. Smith s modified model estimated better the runoff with high soil moisture. The model adjustment was satisfactory for Rio Grande do Sul. / Com o objetivo principal de modelar o escoamento superficial para solos com diferentes características físicas a partir de diferentes intensidades de precipitação e duração das chuvas, realizaram-se chuvas simuladas em diferentes solos no Rio Grande do Sul e Uruguai, com intensidade de 30, 60 e 120 mm h-1. As chuvas artificiais foram aplicadas utilizando-se um simulador estacionário de bicos múltiplos e oscilantes. Seis parcelas foram delimitadas por chapas metálicas galvanizadas cravadas no solo (0,5m2), contendo na parte inferior uma calha coletora, para coletar a água do escoamento superficial (mensurado em intervalos de cinco minutos). Em cada solo foi determinado o tempo de início e a taxa de escoamento superficial, além da chuva (quantidade, duração e intensidade), declividade do terreno, massa seca na superfície e cobertura do solo, densidade (do solo e de partícula), porosidade do solo (macro, micro e total), textura (argila, silte e areia), umidade inicial e de saturação do solo. Utilizou-se o modelo modificado de Smith para estimativa do escoamento superficial. Os parâmetros do modelo foram ajustados através de equações multivariadas. No Uruguai, as perdas acumuladas por escoamento superficial foram de 64, 32, 30 e 15% do total aplicado, para o Vertissolo 1, Chernossolo, Argissolo 2 e Vertissolo 2, respectivamente. No Rio Grande do Sul, as perdas acumuladas por escoamento superficial foram de 67, 45 e 27% do total aplicado, para o Argissolo 1, Neossolo e Latossolo, respectivamente. Na maioria dos casos houve uma redução no tempo de início de escoamento superficial, com o aumento da umidade inicial do solo e da intensidade da chuva, independentemente das condições da superfície do solo. O modelo modificado de Smith estima melhor o escoamento superficial em condições de elevada umidade do solo e o ajuste foi satisfatório para o Rio Grande do Sul.

Resumo: Diversos fatores são responsáveis pela variabilidade espacial na produtividade das culturas e dentre estes estão os atributos do solo. Contudo, o que se tem observado, geralmente, é uma baixa correlação entre os mapas de produtividade e os mapas de atributos do solo. Uma hipótese para essa baixa correlação é a diferença na amostragem de cada uma dessas variáveis. O trabalho objetivou estudar o padrão de distribuição espacial dos atributos do solo e da produtividade do milho e as relações de causas e efeitos utilizando-se diferentes intensidades de amostragem. Os dados foram coletados seguindo uma grade amostral contendo 100 pontos georreferenciados. Os pontos foram dispostos em intervalos equidistantes, sendo de 10 m no sentido das linhas da cultura, ao longo de quatro transeções paralelas com 250 m de comprimento, distanciadas entre si 4,5 m, formando um retângulo com quatro colunas e 25 linhas, constituindo os 100 pontos de amostragem. Cada ponto amostral foi composto por cinco linhas de 10 m da cultura, constituindo uma célula amostral de 45 m2. Em cada ponto amostral retiraram-se amostras de solo nas profundidades de 0-0,1 e 0,1-0,2 m. Foram avaliados os atributos do solo textura, pH, teores de matéria orgânica, P, K+, Ca2+, Mg2+, H+Al, soma de bases, capacidade de troca de cátions, saturação por bases, densidade, porosidade total, macroporosidade e microporosidade. Foram estabelecidas intensidades de amostragens por meio de eliminação de pontos intermediários partindo da amostragem inicial, que continha 100 pontos, obtendo-se conjunto de dados com 100, 75, 50 e 25 pontos. Realizou-se a análise estatística descritiva dos dados e geoestatística. Baseado na análise de correlação simples, regressão múltipla 'stepwise' e semelhança dos mapas de isolinhas, os atributos que mais se correlacionaram... (Resumo completo, clicar acesso eletrônico abaixo) / Abstract: Soil attributes play an important role in spatial variability of crop yield. However, what has been observed, generally, is a low correlation between the yield maps and maps of soil attributes. One hypothesis for this low correlation is the difference in the sampling of each of these variables. This research had as objective to study spatial distribution pattern of soil attributes affecting corn yield and their causeeffect relationship, for different sampling intensities. Spatial variability was determined for 100 georeferenced sample points from a sampling grid. Sample points were arranged in regular 10-m intervals along the rows and in four parallel 250-m long and 4.5-m spaced transects, resulting in a 4-column and 25-row rectangle (100 sample points). Each sample point consisted of five 10-m rows, totaling 45 m2. Samples were taken from the depths 0-0,1 and 0,1-0,2 m. Soil texture attributes, pH, organic matter content, P, K+, Ca2+, Mg2+ and H+Al levels, sum of bases, cation exchange capacity, base saturation, bulk density, total porosity, macroporosity and microporosity were evaluated. Sampling intensities were established by eliminating intermediary points from the initial 100-point sampling, resulting data sets of 100, 75, 50 and 25 points. Statistical analysis consisted of data description and geostatistics. Based on the simple correlation analysis, stepwise multiple regression and similarity among isoline maps, the attributes that showed stronger spatial correlation to corn yield were base saturation and clay content. For these attributes, data sets of 75 and 100 points showed higher similarity in the spatial distribution pattern for corn yield. The analysis of cross-semivariograms showed that the best correlation between corn yield and soil attributes was obtained with a sampling intensity of 100 points. It was possible to confirm the hypothesis that, when performing the soil... (Complete abstract click electronic access below) / Orientador: José Eduardo Corá / Coorientador: Carolina Fernandes / Banca: Marcílio Vieira Martins Filho / Banca: Zigomar Menezes de Souza / Mestre

När geotekniska laborationsanalyser utförs undersöks ofta kolvprover från olika djup som får representera marken vid det djupet provet är taget. I ett homogent lerskikt kan denna punkt antas vara representativ. I en icke-homogen lerjord, till exempel i varvig lera, skulle denna punkt kunna infalla i en icke-representativ variation. För att ta reda på om dessa strukturella variationer påverkar en leras mekaniska egenskaper har i denna studie lerprover från olika platser i Stockholms län analyserats. Genom laborationsundersökningar har lerornas vattenkvot, konflytgräns och lerhalt bestämts och jämförts med varandra. Konflytgränsen definieras som vattenkvoten då en lera går från plastisk konsistens till flytande konsistens och bestäms i denna studie med fallkonmetoden (Axelsson & Mattsson, 2016). Vattenkvoten anger förhållandet mellan jordens fasta massa och vattnets massa och bestäms genom vägning och torkning i ugn (Larsson, 2008). Lerhalten i ett jordprov bestäms genom en hydrometeranalys som anger mängden lera i förhållande till övriga kornstorlekar i provet. Proverna som har studerats i denna studie var av varierande kvalitet med avseende på varvighet och både glaciala och post-glaciala leror har undersökts. Resultatet visar att det är svårt att studera skillnader mellan ljusa och mörka variationer i leror som inte har en tydlig varvighet och att stora variationer förekommer i de undersökta parametrarna för mörka såväl som ljusa skikt. / When geotechnical laboratorial analyzes are executed, piston samples from different earth depths are commonly used. These samples will represent the soil at the given depth. The point at which the sample is taken could be seen as representative in a homogenous layer of clay but in an inhomogeneous layer, such as a varved glacial clay, this point could occur in a variation that is not representative for the whole layer. To find out if these structural variations will affect a clays mechanical properties clays from the Stockholm region have been analyzed. The clays water content, liquid limit and clay content has been determined through laboratorial analyzes and then compared with each other. The liquid limit is defined as the water content when a clay is transitioning from plastic to liquid consistency and is determined by the fall-cone method (Axelsson & Mattsson, 2016). The water content is determined through drying in a drying oven and gives the relationship between the soil’s solid mass and the mass of the water (Larsson, 2008). The clay content in a soil sample is determined through hydrometer analysis and gives a value on the amount of clay in relationship to other fractions. The samples which have been studied were of different quality with respect to how distinguishable the varves were and both glacial and post-glacial clays have been analyzed. The result show that it is difficult to analyze differences between light and dark variations and varvs in clays which does not have distinct layering and that vast variations occur in all of the analyzed parameters for both dark and light variations.

La terre compactée est considérée comme un mélange granulaire dans lequel l'argile joue un rôle de liant mais cette dernière exhibe une forte interaction avec l'eau. Pendant la durée de vie en service, la terre compactée est soumise aux changements de l’humidité relative. En raison de ces changements des conditions ambiantes perpétuels, la teneur en eau dans la terre varie, impactant leur performance mécanique. Le présent travail a ainsi pour but d’étudier l’impact de l’humidité relative et de la teneur d'argile sur le comportement mécanique de la terre compactée. Il se réalisera au travers d’études expérimentales et d'une modélisation constitutive. Dans la première partie de cette thèse, quatre terres régionales de provenances et de teneurs d'argile différentes sont identifiées. Une étude comparative a été réalisée entre le double compactage statique et le compactage dynamique. En parallèle, trois types d'essais spécifiques : essais de succion par la méthode de papier-filtre, essais de retrait et essais d'absorption d'eau, ont été menés pour donner des indications préliminaires quant aux effets d'interaction entre l'eau et l'argile. Dans la deuxième partie, l’impact de l’humidité relative et de la teneur d'argile sur le comportement de cisaillement a été étudié, prenant en compte des cycles de chargement-déchargement. En adoptant une définition particulière de la contrainte effective de Bishop, il a également été observé que les états de rupture dans le plan (p'-q) pour tous les échantillons sont alignés approximativement à une ligne droite unique passant par l'origine, quelque soit la succion et la pression de confinement. Sur la base des résultats expérimentaux, un nouveau modèle constitutif a été développé pour la simulation du comportement mécanique de la terre compactée. Ce nouveau modèle a ainsi été formulé dans la cadre de la mécanique de l'endommagement des milieux continus et de la théorie de Bounding Surface Plasticity. / Compacted earth is regarded as a granular mixture in which clay plays a role of binder but it also exhibits an important interaction with water. During their service life, compacted earth can be subject to large changes in relative humidity. Those perpetual changes of environmental conditions induce continuous changes of water content of the earth that impact significantly its mechanical performances. The present work aimes at studying the mechanical behavior of compacted earth with respect to relative humidity and clay content. It involves an extensive experimental study and a constitutive modelling. In the first part of this thesis, four kinds of local earth are identified with different clay contents. A comparison of compaction method was then conducted between a double static compaction and dynamic compaction. Three types of specific tests: suction test by filter paper method, shrinkage test and sorption-desorption test were carried out, thereby providing a preliminary insight on the interaction effects between clay and water. In the second part, the impact of clay and moisture contents on the shear behavior of compacted earth was investigated taking into account loading-unloading cycles. Adopting a particular definition of Bishop's effective stress, failure states of all samples were observed to lie approximately on a unique failure line crossing the origin in the (p'-q) plane regardless of matric suction and confining pressure. Finally, based on the above experimental results, a new constitutive model was proposed, based on the theories of Bounding Surface Plasticity and continuum damage mechanics, aiming to simulate mechanical behaviour of compacted earth.