The accurate estimation of precipitation quantities arriving at the ground in complex terrain such as the Alps is complicated by many factors. The orography interacts with atmospheric flow and thus influences the spatial and temporal distribution as well as the microphysics of precipitation. Although weather radar can provide high-resolution measurements of precipitation, their view is sometimes blocked by the relief in which case measurements from higher altitudes need to be extrapolated to the ground level. When extrapolating the radar measurements aloft for quantitative precipitation estimation (QPE) at the ground, these must first be corrected for the vertical change of the radar echo caused by the growth and transformation of precipitation (VPR correction). Many existing operational algorithms for QPE and VPR correction assume the vertical structure of precipitation to be spatially and temporally homogeneous. However, given the variable nature of precipitation this assumption may not hold, especially in mountainous areas. This thesis seeks to contribute to the improvement of QPE by radar in the Alps through the analysis of the spatio-temporal variability of the polarimetric radar signals. More specifically, this is done through contributions to questions on radar monitoring and stability, the characterisation of the spatio-temporal variability of the melting layer and the study of the potential for the inclusion of polarimetric radar variables in a more localised vertical profile correction approach. A method based on spectral analysis is used to provide some new perspectives on radar hardware monitoring using the polarimetric signals returned by a single bright scatterer. It is shown that valuable information on the state and stability of the radar hardware can be obtained if different scales of variability and several polarimetric variables are considered. The same approach is used for the characterisation and comparison of the spatio-temporal variability of the melting layer on the relatively flat Swiss plateau and in a large inner Alpine valley in the Swiss Alps. Based on the results of this study it appears that the smaller spatial scales contribute more to the total spatial variability of the melting layer in the case of the Alpine environment. Finally, building on the availability of polarimetric data and a hydrometeor classification algorithm, a new framework for the application of machine learning methods to study the vertical structure of precipitation in Switzerland as well as a more localised vertical profile correction method is proposed. It is shown that models which include information on hydrometeor proportions better represent the observed patterns of vertical change in precipitation and that these models can predict from altitudes between 500 to 1000 metres higher than models based on only reflectivity data.