What the forecast is built on

Sentinel-1 and Sentinel-2 imagery at 10 m per pixel. Forest is a gate, not a bonus: a cell with no trees does not grow woodland mushrooms however good the weather is. Conifer share matters because species disagree about it — the taiga tick lives in conifers, the European one avoids them. Wetland is kept apart from open water on purpose: a bog is land, and treating it as water would erase the boreal forests of Karelia and Western Siberia from the map, along with the cranberries and cloudberries that only grow there.

108 m per pixel, which is ample: we correct temperature by height across an H3 cell about 1.2 km wide, so the argument is about hundreds of metres, not single ones. We deliberately did not take the Copernicus GLO-30 COG version — it needs GDAL, several hundred megabytes of dependencies, for data we read once.

Blackfly larvae cling to stones in streams and riffles: they need current to breathe, and stagnant water starves them of oxygen. Mosquito larvae need the opposite — a puddle. We were tempted to use bog cover as a stand-in for both and rejected it: in June, by a taiga river, bog cover is low and the substitute would have suppressed blackfly exactly where it is worst. A bad proxy is not slightly less accurate; it answers with the wrong sign. HydroRIVERS is derived from terrain rather than mapped by hand, which matters: hand-mapped coverage follows population density, and a stream layer built from OpenStreetMap would report no blackfly in the taiga.

Real boundaries rather than circles drawn by hand. Until this is collected the site falls back on the hand-drawn shapes, which for 76 of 77 regions means 'no protected areas' — and that is a statement we would rather not make.

We run our own instance rather than calling the public endpoint. Soil moisture at depth matters more than rainfall: mycelium responds to what reached it two weeks ago, not to yesterday's shower. Soil temperature, not air temperature, decides whether a tick has emerged.

Nobody collected these for us, which is the point. For each find we compute our index on that day in that place, then on other days in the same place. If the model is right about timing, the day of the find should rank above a random day. The result, and what it does and does not prove, is published in full.

Read district by district rather than region by region, because the difference is large: Moscow Oblast is endemic in two districts out of sixty-four. Calling the whole region endemic would frighten six million people; calling it clear would stay silent where the risk is real. This register covers Russia; the European sections need their own sources, and until they exist those sections stay unpublished rather than translated.