Weather radars scan from hilltops and tall towers, producing rainfall maps at resolutions of hundreds of meters and several minutes. In dense urban areas — where flooding risk is highest and infrastructure is most vulnerable — this resolution is often insufficient.
This paper demonstrates that cellular base stations already function as distributed radar systems. The radio signals transmitted between towers are attenuated and scattered by rainfall. By processing these signals with standard radar techniques, the authors extract rainfall measurements at spatial resolutions of a few meters and temporal resolutions of tens of seconds — orders of magnitude finer than conventional weather radar.
Note (2026-11-14; read against the paper's full text): the paper uses a commercial 5G-Advanced base station and says it demonstrates the approach with the aid of physically based simulations and actual measurements in clear-sky clutter-only conditions and during rain; the products it retrieves are weather radar moments (reflectivity factor, mean Doppler velocity and spectral width), and it reports clutter-filtered moments of satisfactory quality when compared with raw observations and with independent base stations with overlapping fields of view. It gives the resolution advantage as approximately ten times the spatial resolution and refresh rate of conventional weather radars, whereas the essay says orders of magnitude finer. The paper lists low transmitted power and limited antenna gain, and says the limited sensitivity of the system to precipitation is one of the challenges.
The main obstacle is ground clutter. Weather radars point upward, so the only returns come from atmospheric targets. Cell towers point horizontally, so buildings, trees, and terrain contaminate the signal. The paper's contribution is the signal processing pipeline that separates weather returns from the fixed-clutter background, exploiting the fact that rainfall scattering varies in time while ground clutter does not.
The infrastructure already exists. Dense urban areas — exactly where fine-resolution rainfall data is most needed — are also where cell tower density is highest. No new hardware is required, only new processing of signals that are already being transmitted for communication.
The duality is structural: a communication network and a sensor network are the same physical system viewed from two perspectives. Every signal carries information about its channel. Whether you decode the message or decode the channel depends on what you want to know.