Garden with Insight v1.0 Help: autocorrelation
The values of some variables recorded through time tend to resemble each other. For example, the amount
of food you eat in any minute of the day is autocorrelated, or correlated with itself (auto) through
time. Any system where there is some feedback such as a repository that is depleted (the amount of food in
your stomach) tends to have some autocorrelation. Autocorrelation is sometimes called a lag
effect, because the effect of each action has affects that lag after it has occurred.
This simulation uses autocorrelation to generate daily temperature and
solar radiation. The repository here is heat from the sun stored in the earth. If you examine the
temperature and solar radiation parameters for any climate in the
simulation, the temperature peak tends to lag behind the solar radiation peak by about 80 days.
The autocorrelation matrix used in the simulation (which predicts today's radiation and temperature based in part on yesterday's) was
calculated by examining climate data from one thousand climates over thirty years to find common trends.
It is not a perfect predictor of climate variation, but it does simulate weather data similar to what
one would find in a particular climate.
How it works: