This study uses information theory to enhance rainfall retrieval from attenuation data of commercial microwave links (CML). The framework enables evaluation of the performance of CMLs as rainfall sensors in a probabilistic manner and assessment of information content of an arbitrary variable for the rainfall retrieval, e.g. synoptic type. The study shows that using the information theory concept can also directly improve data processing of attenuation data, in this case dry-wet classification.

Commercial microwave links as rainfall sensors have been investigated and evaluated in numerous studies with gauge-adjusted radar used for reference for rainfall observations. We evaluate collocated commercial microwave links, which are thus exposed to identical atmospheric conditions. This set-up enables the exploration of inconsistencies in observations of independent sensors using data from a real telecommunication network. The sensors are in agreement and are homogeneous in their behaviour.

An original dataset of microwave signal attenuation and rainfall variables was collected during 1-year-long field campaign. The monitored 38 GHz dual-polarized commercial microwave link with a short sampling resolution (4 s) was accompanied by five disdrometers and three rain gauges along its path. Antenna radomes were temporarily shielded for approximately half of the campaign period to investigate antenna wetting impacts.

Commercial microwave links operating at E-band frequencies are increasingly being updated and are frequently replacing older infrastructure. We show that E-band microwave links are able to observe even light rainfalls, a feat practically impossible to achieve by older 15–40 GHz devices. Furthermore, water vapor retrieval may be possible from long E-band microwave links, although the efficient separation of gaseous attenuation from other signal losses will be challenging in practice.