Help build a trustworthy spring recordDocument a migration night. Keep sensitive places protected.
Conservation groups and field observers can submit a dated public institutional source that explicitly documents actual movement. The source enters a review queue; no submission changes the tracker automatically.
Submit source evidence ↗01 · SUBMITBroad region and dateSupply a public source link. No GPS, crossing directions, photo upload, or contact details needed.
02 · REVIEWHuman source verificationAn editor checks actual movement, independent event date, authoritative sourcing, and habitat privacy.
03 · PUBLISHDated evidence, not a predictionApproved reports enter a region-level historical archive after automated integrity tests.
Long-term research qualityAre there enough observations to validate the predictor?
Historical event coverage and monitored non-movement nights must be recorded before any claim of statistically validated migration probability.
—Reviewed movement reports
—Spring seasons covered
—Broad regions covered
Calibration status: evidence collection. No validated migration probabilities claimed.
Historical archive: no independently reviewed movement records yet. An empty record is not evidence that salamanders did not migrate in earlier years.
This is an independent educational tool, not an official reporting program of the Michigan Vernal Pools Partnership or participating organizations. For formal vernal-pool research and monitoring, use the Michigan Vernal Pools Partnership.
A missing report is not a quiet nightMeasure the nights people checked.
We need both monitored movement and completed watches where no salamanders were encountered. These are specific surveys, not regional absence claims. Nothing submitted goes public without separate original-source review.
Log a completed monitoring night ↗Reviewed monitoring sessions—Completed, source-checked surveys only
Movement encountered—During recorded monitoring effort
None encountered—Not evidence of region-wide absence
Observer-hours logged—Observers × completed minutes
What counts as useful monitoring?
A completed, repeatable watch by authorized observers, with recorded duration, effort, a documented movement or zero-encounter result, and a source that the editor independently checks. A casual sighting or skipped night does not meet this standard.
Model verification statusCollecting evidence · Not calibratedAn as-issued forecast archive begins in spring; held-out evaluation across monitoring seasons remains unavailable. The seven-night tool is a weather-and-thaw guide, not a validated probability.
For already-authorized monitoring only. Never enter traffic, trespass, move animals, disturb woodland pools, or publish sensitive habitat locations. This is an independent reporting initiative, not the Michigan Vernal Pools Partnership's official monitoring program.
The forecast, frozen before the nightWhat did the tracker predict at the time?
Each March–May afternoon, the public NWS-and-thaw outlook is saved as it was issued, with a timestamp and its seven-night, nine-region assessments. Forecast history is never reconstructed from observations after the fact.
Historical forecast evidenceBegins next springScheduled for spring afternoons · no past forecasts invented
Archive history is checked without requesting additional weather data.
—Completed-day issues saved
—Permanent missed spring days
—Days with partial region coverage
—Issues with limited NOAA thaw evidence
Issued forecast days—One frozen issue per Michigan date
Seven-night regional assessments—Known-status outlooks preserved
Matches to monitored nights—Same-day issued forecast + reviewed survey
Both survey outcomes matched—Movement / none encountered during survey
No archived spring forecast issues yet.The daily capture schedule begins in March. October weather cannot be used to recreate a March prediction, and historical forecast scores will not be invented. If a scheduled day is missed, the gap remains visible in the archive.
Not a prediction-accuracy score. Region-level weather suitability is only matched to separately reviewed, site-specific monitoring effort. A zero encounter at one site does not prove regional absence. The forecast remains uncalibrated; reliable accuracy estimates require held-out seasons, comparable monitoring, and detection-bias checks.
.jpg?width=900)
