Mushroom Forecasting Alerts
Estimate when and where a mushroom species is likely to fruit, and send an alert.
Concept. Not built; this page describes a proposed design.
What it would do
- Combine weather forecasts, recent weather history, and past sightings.
- Estimate, for each map grid cell and day, the chance that a chosen species is fruiting.
- Let a user pick species and an area, and send an alert when the chance passes a threshold.
How it would work
- Collect past sightings with date and location for each species.
- Join each sighting to the weather in the weeks before it: rainfall, temperature, soil moisture.
- Add fixed site data: elevation, land cover, and likely host trees.
- Fit a model per species that predicts the chance of a sighting from those inputs. A simple option is a logistic regression or gradient-boosted trees; a Bayesian model could add uncertainty ranges.
- Run the model daily on the weather forecast and publish a map.
Data sources
| Data | Possible source | Update rate |
|---|---|---|
| Weather forecast | NOAA National Digital Forecast Database | Hourly |
| Weather history | NOAA or local weather stations | Daily |
| Soil moisture | Public soil moisture products, or own sensors | Daily |
| Sightings | iNaturalist, Mushroom Observer | As submitted |
| Terrain and land cover | USGS elevation, national land cover data | Rarely |
Outputs
- A map layer of fruiting chance per grid cell (target: about 1 km cells).
- Alerts by email or push notification for chosen species and areas.
- A daily summary listing areas with a high chance.
Risks and open questions
- Sightings show where people look, not where mushrooms grow. Popular trails will look like hotspots.
- Many species have few records. The model may only work for common species.
- Forecasts could send many people to the same small area. Sensitive or protected sites may need to be hidden.
- An alert is not an identification. The app must not suggest that anything found is safe to eat.
- How to check accuracy: hold out one recent season of sightings and compare predicted against actual.