Thread by @aDissentient

The argument that global warming is a serious problem rests on the output of climate models, vast computer simulations of the atmosphere, the oceans and the biosphere. But are climate models good enough to inform policymakers? This thread argues they are not. Not even close. 🧵

Climate models are *really* big. Millions of lines of code attempting to reproduce the physics of a highly complex system, creating an artificial world, its surface divided into cells of up to 100 km × 100 km in size, multiple layers of atmosphere above, and ocean below.

Unfortunately, some features of the climate operate at smaller than this. For example, clouds are much smaller than a typical grid cell. In these circumstances, a simplified model has to be used instead. This is called “parameterisation”.

However, parameterisation can lead to unphysical results. In a recent @netzerowatch paper, Willis Eschenbach explains that sometimes the proportion of the cell that is cloud covered can become negative! netzerowatch.com/climate-models…

Eschenbach outlines similar problems. For example, NASA’s Model E includes a bit of code to make virtual melt ponds on the polar ice caps refreeze when the physics programmed into the system fails to do so at temperatures below -10°C.

There are lots of issues along these lines in Model E. Sometimes wind speeds and temperatures go far outside reasonable bounds. The model even fails to conserve energy and mass, and there is a crude fix in place for this…

It’s not surprising then that climate models tend to be not very good at reproducing the real climate. For example, most overestimate warming in the tropical troposphere, a part of the atmosphere that is supposed to be diagnostic of global warming.
https://t.co/VR4wlVNZ7fthegwpf.org/publications/c…

Given the difficulties climate models have with clouds, it’s no surprise that they get rainfall wrong too. Most climate models suggest that almost all rainfall at low latitudes will be drizzle. Observations show that’s it’s less than half. https://t.co/FHh7xAOFtOrepository.library.noaa.gov/view/noaa/4471…

So when you are told that your area will become wetter or drier or struck by drought, remember that climate models are really bad at rainfall.

There is a lesson here for policymakers. If climate models don’t get key features of the climate correct, if they don’t conserve energy, if they are full of crude fixes to stop their output looking daft, they are of *no* relevance to the policy process.

Unfortunately, they are being used to justify an unprecedented economic and social upheaval, which is leading us towards disaster. https://t.co/2b3EdVAFHkthegwpf.org/publications/t…