Loading…
A vertically pointing radar sends a narrow beam straight up. Anything that flies through it — a migrating thrush, a bat, a moth — is detected, measured and classified. This is a field of night sky with those animals moving through it.
Every mote is a real animal. When it was detected, how high it was flying, how fast it was moving over the ground and which way it was heading are all measured by the radar.
Where it appears across the frame is not. The beam is only tens of metres wide at flight height, so the radar cannot tell whether an animal passed directly overhead or a little to one side — every entry point here is arbitrary. Each mote also travels in a straight line, which these tracks are close to (a straightness index averaging 0.92) but never exactly.
The radar follows each animal only briefly — typically about twelve seconds, over some two hundred metres of flight — so the frame is sized to match. A mote is at its brightest over the stretch that was actually tracked, and fades either side of it, where its path is inferred from the heading and speed that were measured.
Animals higher up drift across more slowly, as they would to someone lying on their back watching the sky. That is a depth cue rather than a measurement: the effect is real, but it is drawn gentler than true perspective so that low and high animals stay legible together. The numbers beside each mote are the measured values — height above ground, and speed over the ground — not the speed you see on screen.
The headline figure is a migration traffic rate: how many birds and bats crossed a one-kilometre line each hour. It is not the number detected. The beam widens with height, so it samples far more sky at 1500 metres than at 200, and a raw count of echoes would make a night of high-flying birds look quieter than a night of low ones. The radar computes a correction for every echo, and this is those corrections summed over the last half hour. The plain count of detections over the same half hour is shown beneath it. The height chart is corrected the same way, which matters: the beam is narrowest close to the ground, so a chart of raw detections badly understates how much passes low overhead. Detections below fifty metres are left out altogether: that close to the ground the correction runs away, and what it magnifies is mostly local birds, insects near the antenna and clutter rather than migration. The mean height printed under the chart is weighted the same way, so it describes where the traffic was rather than where the detections happened to fall.
The sky holds far more than you see here. The beam is a narrow column, so the radar only detects animals that happen to fly through it — a small sample of everything crossing overhead. Every mote is something that was really detected, and on a good night the airspace above the site holds many times that number.
Insects make up roughly nine echoes in ten. Drawn individually they would bury the birds, so they appear as the soft green glow the birds move through, and as a count rather than as animals.
The sky behind them is not decoration. It is painted from the sun's real elevation at the site and the moment being shown, so dusk deepens through civil, nautical and astronomical twilight and into night at the times it actually did — which is also when the birds appear.
Data from a Swiss BirdRadar MR2. Times are shown in local time at the radar, with UTC alongside. Heights are above ground level. A night is named for the evening it begins — "15–16 May" runs from midday on the 15th to midday on the 16th, so the whole night of flying stays together instead of being split at midnight.