Rougly said, the chaos you are talking about is averaging out over larger spatial and temporal scales. The errors in climate models are coming from
A) necessary simplifications (we can only resolve the earth to a scale of a few kilometers or so and need to use subgrid models to approximate processes that happen on smaller scales such as cloud formation) and
B) unknown or not well enough known parameters and model behavior and
C) Unknown input quantities (most prominent: future man-made CO2 emissions)
There is thus quite some disagreement between different climate models[1] and uncertainty is sometimes reported based on the amount of disagreement between these models.
This is a great explanation, thank you. If I'm reading that chart correctly, out to 2060 or so, the uncertainty is high enough that both high and low temperature change scenarios overlap, but the overall risk is still moderate to high, is that right?
A) necessary simplifications (we can only resolve the earth to a scale of a few kilometers or so and need to use subgrid models to approximate processes that happen on smaller scales such as cloud formation) and
B) unknown or not well enough known parameters and model behavior and
C) Unknown input quantities (most prominent: future man-made CO2 emissions)
There is thus quite some disagreement between different climate models[1] and uncertainty is sometimes reported based on the amount of disagreement between these models.
[1] See https://www.ipcc.ch/report/ar6/syr/downloads/report/IPCC_AR6..., page 17, in the high emissions scenario the uncertainty range lies between 3 and 5 °C.