Abstract : The integration of the weather effects into decision support tools and real time traffic management strategies represents a critical need for all road operators. The motivations are clear because of the significant effects of adverse weather on road safety and drivers' behaviors. At a safety level, the increase of the crash frequency and severity has been highlighted by several studies. This increase of the crash risk does not concern only extreme weather events, such as winter storms, but also recurring events like rain. The changes in drivers' behaviors (decrease of speeds, headways) andtraffic flow dynamics (speed, flow, density) lead to significant consequences from a mobility point of view : thus, rain represents the second largest cause of non recurring congestion (15 \%) after incidents.In spite of this context, the effects of adverse weather on traffic are not well quantified and, above all, not integrated into traffic modelling and estimation. The presented thesis research aims at contributing to a better understanding of the meteorological effects on traffic by focusing on precipitation events at interurban sections. From a literature review of the meteorological impact on traffic, we have underlined a need of a standardized methodology. Such a standardized methodology for the rain impact quantification is proposed and applied to real data. It enables aquantification of the rain effects at different levels, according to the scale of representation : at a microscopic level, the statistical analyses highlight changes in drivers speeds, time headways. Those effects reflect on the macroscopic level of traffic flow with changes in speed, flows, and, in a general way, in all the parameters composing the fundamental diagram of traffic. Hence, the empirical results pave the way for integrating the meteorological phenomenon into traffic modelling.Next, we propose a theoretical contribution to traffic modelling, based on a Vlasov formulation, which enables the derivation of a two equations macroscopic model. The proposed model offers a relevant framework for the integration of a meteorological parameter. Regarding the numerical discretization, we propose a fractionnal step method allowing to deal successively with the source terme and the homogeneous part of the system. We develop a Lagrange+remap scheme for the homogeneous part of the system. The model behaviour is illustrated through several numerical experiments which highlight the model features faced with changing meteorological conditions.In the last chapter, an effort towards future online applications is put forward. Within a Bayesian framework for data assimilation, the goal resides in the online estimation of the traffic state vector given current measurements. Based on real world data, some scenarios show the benefits of the integration of the meteorology into such approaches. Thus, a better knowledge of the weather impact on traffic leads to its integration into traffic models and will enable the improvement of decision support tools for road operators. The proposed work opens perspectives for the development ofweather-responsive traffic management strategies.