Analysis
Detection volume and richness
Volume and species count diverge, which is the point: the busiest days are not the most varied ones, because fewer birds sing far more often.
Species accumulation
Cumulative species by date. Where the curve flattens is where the list stops growing. With a second spring logged, the gap between years says whether it ran early.
70 species on record.
When the garden wakes
Every species combined, averaged per day, home station only. Outings are left out on purpose — a walk taken mid-afternoon would invent an afternoon peak that has nothing to do with the garden.
Activity rises above a tenth of its peak around 06:00 and falls back below it after 20:00, averaged over 56 days. Faded bars are outside that window. A relative threshold rather than a fixed count, so a loud month and a quiet one are still comparable.
Counts follow vocal activity, analyser sensitivity and ambient noise — not how many birds are present, and not when they physically wake.
Diel activity by species
Mean detections per hour, averaged across every day on record so seasonal noise washes out and the daily rhythm shows. Owls invert; most songbirds hold a dawn peak. Ordered by total detections, home and away combined.
Confidence distribution
BirdNET confidence is a network output, not a calibrated probability. Use it to pick a display threshold, never to state a false-positive rate.
Home versus away
The one place home and outing records are deliberately compared. Away-only species are what portable use adds — birds that never reach the yard.
Arrivals and departures
A migration calendar built from your own records. Once two springs are logged these become year-over-year comparisons.
Dawn chorus onset
First detection of the day against sunrise, across the season. The drift through spring is real; midsummer scatter is mostly weather.
Rarities and singletons
Species with fewer than five detections. False positives concentrate here, so each is flagged for clip review rather than counted as confirmed. This is a work queue, not a trophy case.
A low count here is not the same as a rare bird. It usually means a quiet species, a brief visitor, or a misidentification — which is exactly why these want listening to rather than celebrating.