La construction d'ouvrages de protection répond à la nécessité de se protéger d'aléas naturels comme les phénomènes d'érosion ou de glissement de terrain. Si le béton est classiquement utilisé pour la construction de tels ouvrages, le bois est régulièrement employé sur des aléas engendrant des contraintes faibles ou modérées. Pour cela, les praticiens s'appuient sur deux approches différentes, le génie civil et l'ingénierie écologique, qui requièrent, dans ce contexte, la mise en œuvre d'une structure construite sur la base d'un agencement de rondins de bois d'essence locale. Il existe cependant des lacunes de connaissances freinant la prescription de ces ouvrages. Les verrous scientifiques concernent notamment l'évaluation du niveau de dégradation, au sens des propriétés mécaniques, des rondins pour le suivi des ouvrages existants.L'objectif de cette thèse est de développer deux méthodes d'évaluation du niveau de dégradation de rondins de sapin blanc à partir de la spectroscopie proche infrarouge (SPIR) et de l'analyse vibratoire. Dans ce travail de thèse, le niveau de dégradation d'un rondin est défini à partir du taux de perte en module d'élasticité ou en module de rupture par rapport à leur valeur à l'état intact.Les résultats obtenus ont montré que les méthodes développées permettaient d'évaluer le niveau de dégradation de rondins aussi bien en conditions semi-contrôlées qu'en conditions naturelles. La méthode par SPIR évalue le niveau de dégradation à partir de modèles de prédiction des indicateurs de référence $Dw_{MOE}$ et $Dw_{MOR}$ basés sur les statistiques multivariées. La méthode par analyse vibratoire mesure le taux de perte en module d'élasticité dynamique ($Dw_{E_n}$) des rondins, qui se caractérisent par leurs imperfections géométriques.Ce travail contribue à approfondir les connaissances sur l'évaluation du niveau de dégradation du bois. Les résultats obtenus permettent d'envisager une application sur des structures réelles. Ces deux méthodes sont effectivement adaptables pour effectuer les mesures textit{in-situ}. / The construction of protective works responds to the need to protect themselves from natural hazards such as erosion or landslides. If the concrete is conventionally used for their construction, wood is regularly used on natural hazards generating low or moderate stresses. For this, practitioners rely on two different approaches, civil engineering and ecological engineering, which require, in this context, the implementation of a structure built on the basis of an arrangement of logs. However, there are knowledge gaps hampering the prescription of these works. In particular, scientists locks concern the assessment of the level of decay, in the sense of mechanical properties, of logs for monitoring existing structures.The aim of this thesis is to develop two methods for assessing the level of decay of silver fir logs from near infrared spectroscopy (NIRS) and vibration analysis. In this thesis, the level of decay is set from the rate of loss in modulus of elasticity (MOE) or modulus of rupture (MOR) in relation to their intact value .The results showed that the methods allowed evaluating the log level of decay both in semi-controlled and in natural conditions. The NIRS method assesses the level of decay from prediction models of $Dw_{MOE}$ and of $Dw_{MOR}$ based on multivariate statistics. The method by vibration analysis measuring the loss rate in dynamic MOE ($Dw_{E_n}$)of the logs, which are characterized by their geometric imperfections.This work helps to deepen knowledge on the assessment of the level of decay of wood. The methods may be considered for application on real structures. These two methods are actually suitable for textit{in-situ} measurements.The objective of this thesis is to develop two indicators, DwNIRS and DwE1, from two complementary methods, the near infra-red spectroscopy (NIRS) and the vibration resonant method (VRM), to assess decay-extent of small-diameters silver fir logs decayed by a microbial community. These indicators are intended to help to practitioners in the diagnostic of the structures. The work is based on a mechanical definition of the decay-extent. A reference indicator DwMOE is defined for this purpose from the normalized loss in modulus of elasticity (MOE) of the logs between intact and decayed states.The thesis is composed of three parts. The first part explores the ability of NIRS to assess decay-extent in semi controlled conditions. Small-diameter logs have been degraded in a greenhouse and monitored at different decay levels. The indicator DwNIRS is calculated from a prediction model using multivariate statistics to predict the reference indicator DwMOE. The second part studies VRM. This method allows measuring dynamically the modulus of elasticity of logs at different decay levels. The third part is dedicated to the comparison of both methods in semi-controlled conditions and also in natural conditions.The results show that both methods allow identifying the effects of the microbial community on wood from early stages of decay. The indicators DwNIRS and DwE1 follow faithfully DwMOE even if the intrinsic variability of wood properties coupled with those bring with the microbial activity alter the precision. Thus, the definition of decay classes from DwNIRS and DwE1 limits the undesirable effects of the variability and brings closer the methods from those already used to assess decay in natural conditions. Finally, the classifications obtained from logs decayed in natural conditions are comparable and allow considering an application on real structures. Both methods are effectively adaptable to in-situ measurements with a limited degradation impact on structures.