About FungRadar
We answer when to go, not where other people found things. Below is how that works and — more usefully — what we deliberately do not promise.
The question we chose
Most foraging apps are field guides. They answer «what is this», from a photograph, and they are useful once you are already standing in front of the thing.
We chose a different question, the one that comes earlier and costs more to get wrong: is this weekend worth the drive? That question is answered by weather, season and terrain — not by photographs — and it is answerable days in advance, while there is still time to change your plans.
What the index is
A number per species, per area, per day, recomputed every six hours. It combines a month of weather history with the fruiting lag of each species, the woodland and conifer share of the ground, elevation and slope, and the phase of the current flush.
Two things about it are unusual, and both are deliberate. It is computed per species rather than as one «mushroom index» — a penny bun and a chanterelle disagree about weather and about trees, so a single number for both is an answer for neither. And it carries a confidence value beside it, which drops where we have less to go on, instead of presenting every figure with the same false certainty.
What we do not do
We do not publish a map of other people's spots. This is a product decision, not a missing feature: precise foraging locations are the one thing a forager will not forgive you for publishing, and the way we keep that promise is by not holding those locations at all.
We do not tell you that anything is safe to eat, ever, under any circumstance. Deadly species resemble edible ones closely enough that a photograph cannot always separate them, and a service that hedges on that point is worse than one that stays silent.
Where the numbers come from
Open data, all of it named and licensed: weather from Open-Meteo running on our own instance, land cover from ESA WorldCover at 10 metres, terrain from AWS Terrain Tiles, occurrence records from GBIF for testing. Each source, its licence and how much of it we have actually collected is listed on the data page — including the layers we have not collected yet, which are marked as such rather than left blank.
How honest the forecast is about itself
We test the model against independent field records and publish the result, including the part that does not flatter us: the model reads the month well and the week within it much less well, and it falls short of the threshold we set ourselves before we ran the test. That page exists because a forecast that never tells you when it is weak is not a forecast, it is advertising.