Overview
Data sources
Flightcall draws on two distinct eBird data products, each suited to a different question.
Species phenology is derived from the eBird barchart export, a summary of reporting frequency across a county or region for each of 48 calendar weeks (4 weeks per month, every month). Each value is the fraction of complete eBird checklists in which a given species was recorded. These are the same numbers that drive the familiar colored bars in the eBird species barchart tool.
Hotspot rankings are derived from the eBird Basic Dataset (EBD), the full, checklist-level export for the county. This gives individual observation and checklist records, which allows occurrence rates, counts, and seasonal breakdowns to be calculated per hotspot.
Only complete checklists, those submitted with "All Species Reported" checked, are used in any calculation. Protocol checklists where the observer may have recorded only selected species are excluded, so that absence of a species on a checklist can be treated as a genuine non-detection.
Species phenology
How migration patterns are classified
The core of the phenology pipeline is valley detection: finding stretches of the year where a species is effectively absent, and using those gaps to infer its seasonal pattern.
What counts as a valley? A valley is a run of 4 or more consecutive weeks where detection frequency falls below both 15% of the species' annual peak frequency and below 0.5% in absolute terms. Both conditions must hold: a species with a very high peak that dips to 10% of that peak is not considered absent.
The number and seasonal placement of valleys determines the classification:
No valleys detected. The species is consistently present throughout the year with no prolonged absence period. Examples: Canada Goose, Black-capped Chickadee, Downy Woodpecker.
One valley, occurring in winter (December–February). The species is absent in winter but breeds or is otherwise present spring through fall. Examples: Eastern Wood-Pewee, Yellow Warbler, Sora.
One valley, occurring in summer (May–August). The species is absent in summer but present fall through spring, wrapping around the year. Examples: American Tree Sparrow, Hooded Merganser.
Two valleys — one in summer, one in winter. The species is only reliably present during spring and fall migration windows. Examples: Solitary Sandpiper, Tennessee Warbler, Rusty Blackbird.
Fewer than 10 weeks of any detectable presence, or a peak frequency below 0.5%, or a valley pattern that doesn't fit the above types. Timing is simplified to first appearance, peak, and last appearance.
Species phenology
Calculating arrival, peak, and departure
Once a species is classified, its presence window is defined by the valley boundaries: the weeks between the end of one absence period and the start of the next. Within that window, three timing landmarks are calculated:
Arrival is the first week inside the presence window where frequency exceeds a threshold of 10% of peak or 0.1% absolute, whichever is higher. This filters out the earliest, sparsest reports and targets the week when the species becomes genuinely findable.
Peak is simply the week with the highest detection frequency within the presence window.
Departure is the last week inside the presence window that clears the same threshold used for arrival. For two-passage migrants, spring and fall timing are calculated independently within their respective passage windows.
Week numbers follow the eBird barchart convention: 48 weeks per year, 4 per month, starting January 1–7. Dates shown in the app correspond to approximate calendar ranges for each week slot.
Hotspot guide
How hotspots are ranked for each species
For each species at each hotspot, the primary metric is occurrence rate: the fraction of complete checklists from that hotspot on which the species was recorded.
Occurrence rate
= detections at hotspot ÷ total complete checklists at hotspot.
A rate of 0.40 means the species appeared on 40% of complete checklists.
Hotspots with fewer than 10 complete checklists on record are excluded from rankings: there is not enough data to produce a reliable rate.
Hotspots are also given a lift score: the hotspot's occurrence rate divided by the county-wide baseline rate for that species. A lift of 2.5 means the species is 2.5× more likely to appear on a checklist at that hotspot than at an average location in the county. Lift scores help identify truly special birds rather than simply common species at busy hotspots.
Context
Caveats and limitations
Observer effort is not uniform. eBird data reflects where birders go, not necessarily where birds are. Well-watched hotspots accumulate more checklists and therefore yield more stable rates; lightly birded areas may appear less productive than they actually are.
Detection frequency is not abundance. A frequency of 60% means the species was recorded on 60% of checklists — it says nothing about how many individuals were present. Common, conspicuous species will have higher frequencies than secretive ones even if equally numerous on the ground.
Phenology windows are historical averages. Arrival and departure dates represent central tendencies across the full span of eBird data for the county. Year-to-year variation can be substantial, and climate trends may shift patterns over time. The data is not updated in real time.
Species exclusions. Unidentified species (anything ending in "sp."), slash taxa (e.g., "Greater/Lesser Scaup"), and hybrids are excluded from all calculations. Only taxa with a full 48-week frequency array are processed.
All underlying data is sourced from eBird, a project of the Cornell Lab of Ornithology, and is used in accordance with eBird's data use policies.