Zenbitz:Thoughts on Temperature and Weather

Temperature and Weather
The harsh environment is a big part of this game, and some effort will be taken both in mechanics and programming design to get the needed effects.

As a simplification weather zones or "bands" will be created corresponding roughly to latitudes. I would estimate that no more than 4-5 of these are necessary given the game map. It is possible, but not necessary that the North Atlantic coast of Norway/Sweden/Denmark may be given special treatment. Generally speaking, a zone will always be colder (on average) than one further south. However, the local temperature may fluctuate (in a consistent manner) from world map tile to world map tile. Modeling front systems (pressure changes) and storms is probably outside the scope of the game, but who knows, we might find a meteorologist/programmer with some spare time.

Effective temperature at a given position on a local map is determined by the temperature of the world map tile, modified by the local wind speed, using a standard "wind chill" chart. See below for determination of wind speed, direction, etc.

Seasons
For story reasons, most of the map will require at least 2 seasons - freeze and thaw. The "Thaw" season is required to allow basic crops to grow. The northernmost regions of the maps (and even most of the map later in the game) may be permafrost/single season. If we manage the art assets, we could have 3-4 seasons in the warmer parts of the map.

Assuming a two season model - both ends of the "thaw" are flanked by rainy weather resulting in Snow->Rain->Mud in the "spring" and "Rain->Mud->Snow" in the "fall". We can assume for now 6mo Freeze/6 mo thaw, subject to change.

Could a simple cosine wave can model the average temperature throughout the year? With period 1 year and amplitude of 35 deg C, centered on -10 (so +25 to -45). January 1 or there abouts would be -45, July 1st +25. Subject to local randomized fluctuations?

Freeze & Thaw
Roughly October->April, with a minimum at -45 in the middle, and -10 at the ends. So the "Thaw" season (when avg. T > 0 C) would be roughly mid-May - mid August. This needs to be confirmed with typical taiga temperature ranges I.R.L.

Local Variations
If map tiles are considered to be "cells" (2D) then the temperature for each cell can be chosen from a distribution representative of the season and weather zone. These cells can then be "smoothed" by averaging over over neighboring cells. I would recommend a single iteration of averaging. Wind speed and direction can be handled in a similar manner, with the prevailing wind DIRECTION determined from actual real world data, and random vector variations applied to this. The average min/max temperature should vary smoothly from day to day. The wind speed can be more random.

Cloud cover and precipitation
In an analogous manner to temperature and wind speed, the cloud cover should be chosen from a realistic distribution, and should be smoothed to be continuous across the relevant part of the world map. This cloud cover and temperature along with the season (and a random factor) can determine both precipitation and what type. This should be a simple table look up with a random variation. Note that it can get too cold to snow. The fact that it was raining yesterday should influence the chance that it rains today, and etc. The season should be the primary determinant of precipitation level, although this can be indirect via cloud cover + humidity.

Diurnal variation
It's colder at night, generally.