Seasons, snow and glaciers
A climate over centuries with cold and warm periods; snow that lies as voxels in the town and on the hills, heavier on the heights, firn, glacier ice that creeps into the valley, grinds its bed, pushes the forest over and drops moraine at its snout, permafrost, and the fertility all of it leaves.
The ice age scenario, three and a half degrees colder than the site: the snow lens shows what lies on the ground. Under the era lens the snow, firn and ice are voxels the people walk on and nobody builds on.
The simulation’s climate has a site baseline and a year’s noise. The engine’s epochs are the clock above that: two slow cycles, a drift, and a scenario of keyframes (from year 180 to 320 it is 1.6 degrees colder; after 500 it warms) applied to the baseline once a year. Every rule that reads the weather feels it: harvests, heating, the snow line, the rivers.
The cryosphere makes snow a thing in the world. Per column it keeps snow, firn and ice as millimetres of water and a frozen-ground clock, steps them monthly from the month’s weather, and writes them into the voxels on the natural ground. Snow that survives the ablation season is firn; a deep enough bed of firn turns, some of it each year, into glacier ice; ice thicker than a threshold creeps toward its lowest neighbour a little each month, so a summit’s snow ends up as a tongue in the valley; everything melts by degree-days, ice slowest, and the meltwater is handed to the water model column by column, the river that comes down the mountain when a cold century ends. Ground below freezing for two years is permafrost.
The heights catch more of it than the valley floor: every cell of height adds a share to the fall, up to a cap, so a cold century whitens the summits long before the fields. Moving ice grinds its bed, about a metre of rock a century under an ordinary valley glacier, carries what it grinds, and drops it where it melts, so the snout of a glacier is a heap of stones; once a year that work is made real in the ground, a lowered bed under the ice and moraine below it. Ice deep enough to push a standing tree over does, so the forest in front of an advancing glacier goes down and the ice moves on over the ground. Into a lake or the sea, it calves and is water.
In the town. Snow lies on the fields, the yards, the streets and the open ground of the town as well as on the hills, and on the roofs and the trees. Nobody shovels it by rule: the people’s feet do. A column walked often enough in a few days has its snow pushed onto the ground beside it, so the paths the town actually uses open up between two banks, grey with hard-packed snow, and a street nobody walks stays under the fall; a square walked on every side has its snow carted to its edge, so a hard winter leaves big piles round the squares. The next fall lies on the paths until they are walked again, and the banks go with the thaw. A road crew clears the snow off the street it is laying. A deep winter loads the roofs: a building already in poor repair may have its roof brought in by the weight of the snow, which leaves it standing roofless with scaffolding up for the repair, and one worn right through comes down.
Fertility is one number per column, recomputed once a year from snow-free days, permafrost, saturation, streams, slope, soil and the summer’s warmth, and moved a share of the way each year so a cold decade shows as land going off. The simulation multiplies each field’s harvest by it.
The land tool runs the same models over decades with no town in the way; cargo land-probe does it headless, a line a decade.
In the code: crates/engine/src/climate/epochs.rs crates/engine/src/climate/cryosphere.rs crates/engine/src/climate/fertility.rs · playground: land