How the ground is read — the whole chain

Sweden’s ground, measured at one metre.

Lantmäteriet flew the country with airborne LiDAR and published the result open, under CC BY 4.0. The current forest scanning lays down one to two laser returns per square metre; the returns that find their way through the canopy to the ground are the ones that count, and they become a bare-earth grid at one metre. It is the shape of the forest floor itself, measured under the trees rather than read off the top of them.

From that height field we derive three quantities, cell by cell:

Slope: How steeply the ground falls, which sets how long rain stays where it lands.
Aspect: Which way the cell faces. A north shoulder holds its damp through a dry week; a south face takes the sun and warms first.
Topographic wetness: Where water gathers, from how much upslope ground drains into a cell against how fast it drains out again.

Those three describe the shape a flush follows: a damp, sheltered, north-leaning seam a few metres wide, running along a contour. At metre grain that seam is a feature you can walk to.

Two kilometres of the Skäralid ravine as a weather station supplies it: one flat, uniform olive tone filling the whole square. — The same two kilometres of the Skäralid ravine on Söderåsen, three times, rendered from the production terrain layer. Colour is habitat suitability for chanterelle, Cantharellus cibarius — aspect, wetness and canopy scored for that one species. Only the dark backdrop is added, so the layer stays legible away from the map.

The same two kilometres resampled to a thirty-metre grid: the gold has thickened into coarse rectangular blocks following the ravine. — The same two kilometres of the Skäralid ravine on Söderåsen, three times, rendered from the production terrain layer. Colour is habitat suitability for chanterelle, Cantharellus cibarius — aspect, wetness and canopy scored for that one species. Only the dark backdrop is added, so the layer stays legible away from the map.

What grows on it, cell by cell.

Naturvårdsverket’s national land cover map covers the same ground with canopy and forest type. Spruce, pine, birch and mixed stands, open wetland, clear-fell — each cell carries what stands on it.

Slope, aspect and wetness say what a cell does with water and light. Land cover says what is growing there. Multiply the two and you have per-species habitat suitability: chanterelle wants the damp, mossy spruce ground and the birch edges beside it, which is a wetness reading and a cover reading agreeing; bilberry wants the closed spruce and pine it actually fruits under. One hillside therefore carries five different scores, and the map draws the one you have selected at the top of the screen.

The elevation grid is one metre; the land cover is ten. Where the two meet, the habitat call is made at the coarser of them — ten metres — while the slope, aspect and wetness underneath it stay worked out at the metre the laser measured. Ten metres is about the width of the seam you are walking anyway.

That is what the gold on the map means: ground whose shape and cover suit this species. It is a habitat read, standing on published preferences for each species and on two national surveys — an expert model, and labelled as one.

Seventy thousand tiles, scored once per species.

Sweden’s elevation model is far too large to hold in one piece, so Lantmäteriet publishes it the way it has to be handled — as tiles. Each covers 2.5 by 2.5 kilometres and holds 6.25 million ground cells, delivered as a cloud-optimised GeoTIFF in EPSG:5845 through a STAC API anyone can query. Covering the country takes on the order of seventy thousand of them.

We work through them one at a time. A tile comes down, gives up its slope, aspect and topographic wetness, meets the land cover for the same ground, and goes back up as a baked suitability surface. Then the next one.

2.5 × 2.5 km

One tile of the national grid

6 250 000

Ground cells inside it

~70 000

Tiles to cover Sweden

And the scoring runs once per species, which is the part worth saying slowly. Chanterelle wants damp, mossy, north-leaning spruce ground. Lingonberry wants the dry, open, well-drained pine on the other side of the same hill. Those are different hillsides, so they get different surfaces — five species is five rasters, each computed on its own, and the map draws whichever one you have selected at the top of the screen.

Doing that work in advance is what keeps it quick in the forest. The heavy lifting is finished long before you open it; your phone pulls the small pre-baked tile for the ground you are standing on and multiplies it by this week’s conditions. The arithmetic left for the phone is a multiplication.

The weather gives you the week. The ground gives you the hillside.

Conditions arrive from the nearest active SMHI station, named on screen, so you can see which station your forecast is standing on and how far away it sits. Degree-days, rainfall and barometric pressure come from that observation network. MET Norway supplies the ten-day forecast on top, temperature and precipitation projected forward day by day.

Across the two kilometres of ravine in the images above, that weather input is effectively one value. The wet north shoulder, the ravine floor and the dry southern slope share the same rain and the same warmth in the model.

Which is precisely why the terrain layer earns its place. The weather tells you which week is worth taking. The ground tells you which side of the valley to walk once you are there.

Rain runs on two clocks.

Rain does two separate jobs weeks apart, so the model keeps two moisture windows and runs them separately. Both appear on screen.

Trigger window · weeks: Whether the ground has banked enough moisture to fruit at all this season. Answered in weeks of accumulated rain, most of it already decided by the time you look.
Expansion window · days: Whether fruit bodies are up now. Answered in the last few days of warmth and damp, and the thing that moves between Thursday and Saturday.

Splitting them is what lets a week read as loaded and quiet at once: a strong trigger with a soft expansion means the ground is charged and waiting on the next rain. That is the week to watch the strip, and the morning after the rain to go.

Both clocks are thresholded against local climatology rather than against a national constant, because Sweden spans two climate regimes — “wet for Skåne in September” and “wet for Norrbotten in September” are different amounts of water, and the percentile is what makes the same species read sensibly in both.

Five species, and where each threshold comes from.

A threshold is the number that decides when “damp enough” is damp enough for a given species. The provenance of that number varies, and you always see which kind you are looking at.

Bilberry · Blåbär
Vaccinium myrtillus
Swedish dataThresholds fitted on Swedish observations.
Lingonberry · Lingon
Vaccinium vitis-idaea
Swedish dataThresholds fitted on Swedish observations.
Chanterelle · Kantarell
Cantharellus cibarius
Published thresholds, comparable northern forests · re-tuning on Skåne ground
Porcini · Karljohan
Boletus edulis
Published thresholds, comparable northern forests · re-tuning on Skåne ground
Funnel chanterelle · Trattkantarell
Craterellus tubaeformis
Published thresholds, comparable northern forests · re-tuning on Skåne ground

Bilberry and lingonberry run on Swedish data. Chanterelle, porcini and trattkantarell run on published thresholds from comparable northern forests, being re-tuned on Skåne ground this season — and each read carries its own label, so you always know which of the two you are holding. Season one’s logs from Skåne and Blekinge are what move the mushroom three toward the first group.

Conditions × terrain, and confidence follows the weaker one.

signal = conditions ⊗ terrain

A weighted geometric mean at equal weight — a necessary-AND join, where the strong read requires both halves.

The two sides multiply rather than average, which is what makes the strong read mean something. Prime ground in a dry week reads low and stays low. A perfect week over ground that suits some other species reads low too. Gold on the map plus a good week on the strip is the state worth driving for, and it is the state the join is built to isolate.

Confidence is composed the same way and takes the value of the weakest input behind it. A well-mapped cell read against a station that reported this morning reads confident. The same cell read against a station whose last report was Tuesday reads at Tuesday’s confidence, and you can see which of the two you have. The number carries the age of the thinnest thing underneath it.

The app's map over Ljungbyhed and Slättåkra with gold kantarell habitat suitability painted across the forest, a species row along the top reading Kantarell, Karljohan, Trattkantarell and Blåbär, a five-hundred-metre scale bar, and a card headed 'Kantarell — moving' with a poor-to-prime scale and a 'Find the best spot in view' button.

The same map zoomed in to a two-hundred-metre scale bar with a 'Best ground · Kantarell' card open, describing the strongest kantarell habitat in view as moist north-facing forest, above a 'Take me there' button and a ten-day strip along the bottom marking Monday 27 as the best day.

Go and check it.

Every dataset named here is public, and most of it is free. These are the places the numbers above come from, so you can read the specifications yourself.