Also known as: waterfall, waterfall plot, scrolling spectrum
A waterfall display is the live, scrolling spectrogram at the heart of nearly every SDR application: each new row is a fresh power spectrum across the tuned frequency span, colored by intensity, and the rows scroll so time flows down (or up) the screen.1 It is the panel operators actually watch — a signal that flicks on shows up instantly as a bright vertical streak, and its color, width, and shape hint at what it is before a single bit is decoded. Practically, the waterfall turns an invisible slice of spectrum into something you can read at a glance.
How it works
A waterfall is a spectrogram rendered incrementally in real time. The receiver takes blocks of I/Q samples, applies a window function, runs an FFT, converts each bin to power in dB (a PSD estimate), and draws that vector as one horizontal line of colored pixels. The line is pushed onto the display and the rest scroll to make room, so the image is a moving history a few seconds to a few minutes deep. Several controls shape what you see:
- Color map and range — power is mapped through a palette between a floor and a ceiling in dB. Setting the floor just above the noise floor and the ceiling near the strongest signal gives the most contrast; a poorly set range either washes out weak signals or saturates strong ones.
- FFT size — more bins give finer frequency resolution (you can split two close carriers) but cost more computation and blur fast events, the same time–frequency trade-off as any spectrogram.
- Averaging and scroll rate — averaging several FFTs per row lowers the grassy variance so weak steady carriers emerge, at the price of temporal smearing; the scroll rate sets how much history fits on screen.
Because a persistent-but-weak signal accumulates as a continuous line, the human eye picks it out of noise far better than it could from a single spectrum — the waterfall’s time integration is a free processing gain for detection.
In practice
The waterfall is paired with a spectrum (“panadapter”) trace and is the primary way SDR users find, identify, and tune signals: click a streak to set the receive frequency, judge modulation by the streak’s width and texture, and watch a trunked system’s control channel sit as a steady line while voice channels blink on and off across the band. It is also the fastest way to spot interference, birdies, and images.
A few artefacts are worth recognising because they are display features, not signals. A DC spike draws a permanent bright line at the exact center of a zero-IF receiver’s span, from residual DC offset rather than a real carrier. An IQ image ghosts a real signal to the mirror side of center and slides the opposite way when you retune. “Birdies” are internal oscillator harmonics that stay fixed on the display regardless of antenna. And a strong nearby transmitter can smear across the whole waterfall when the front end overloads. Learning to discount these keeps a waterfall from sending an operator chasing phantom channels.
Relevance to SDR
Waterfall displays are ubiquitous in SDR software — GQRX, SDR#, SDRangel, CubicSDR, and web-based receivers all center their UI on one. For monitoring trunked systems it is invaluable for locating control-channel carriers and seeing simulcast or interference conditions. GopherTrunk itself is a headless decoder rather than a graphical SDR, so it does not draw a waterfall; users typically identify candidate frequencies in a waterfall-based tool first, then hand the frequency to GopherTrunk to decode, and inspect captured I/Q in a waterfall when troubleshooting a stubborn signal.
Sources
-
Spectrogram — Wikipedia, on the time-frequency display that the scrolling SDR waterfall renders in real time. ↩