Today's warning systems for floods or droughts require sensor applications that provide a vast array of information. Current systems provide insight into either the water level or the river velocity. In order to obtain additional parameters for the characterisation of the flow behaviour in a non-contact manner, a frequency modulated continuous wave (FMCW) Radar with chirp sequence is used. Since these sensors provide range and velocity information but also enable gathering of additional parameters. The presented measurements in this contribution were performed at four different rivers with a commercially available radar sensor. The results of the classical post-processed two-dimensional fast Fourier transform are used as basis for an image-wise processing approach to obtain additional features for classifying the behaviour of the river surface. For this purpose, the frame-wise enveloping velocities are depicted in a time sequence. Due to processing the detected reflection patterns in relation to time, a characteristic pattern for river flow profiles can be extracted. By reducing the information using time averaging, characteristic features for different flows can be extracted from the spatial envelope velocity distribution. In particular, the resulting insights lead to characteristic features for single flow distributions. That enables novel monitoring systems with new possibilities to classify rivers using radar sensors.