In the battle between models and reality, the models are struggling. At least some are. So this chapter of the DOE Report begins with a brief explanation of how climate models are assembled and explains why modelers have to make a lot of guesses and assumptions. It’s because they are trying to simulate subgrid scale processes, important things that happen on too small a scale for the computer models to represent them accurately. So there is a lot of averaging and improvising to mimic their results, rather than studying what’s actually happening and how to replicate their processes, and different modelers make different decisions along the way about how to kluge what cannot be done accurately. The result is a mess in which the settled science of climate modeling doesn’t just disagree about little things. Instead the models diverge widely on big, basic ones like what the current average temperature is. It’s not just bizarre but impudent that we keep being told how much warmer it is than in 1850 to two spurious decimal places when top models generate estimates of how warm it is now that span a range of about 2.5 degrees C, more than double the estimated warming over the 20th century. (And don’t get us started on the precision of temperature 175 years ago.) But even if they get the current level of temperatures wrong, surely they get warming trends for the very recent past right, don’t they? Er, not so fast.
Some models seem to. The DOE Report reproduces results for surface warming since 1980 that shows low-ECS models successfully tracked observed warming through 2022. But the medium- and high-ECS models predicted too much warming, in some cases way too much. Which is a very important point because we’re talking about the records since 1980, not 1880, so a period when climate scientists have the modeling inputs they need: CO2 emissions, other air emissions, solar changes, etc. And they have terabytes of climate data with which to build their models, and temperature observations so they can check at the answer and adjust the models accordingly.
It should be the mathematical equivalent of child’s play. And yet even so, most models predict far more warming than actually occurred. Moreover, the ones that get the answer reasonably correct have something important in common: they assume the climate is not overly sensitive to greenhouse gases. Which means they project minimal climate changes over the 21st century not the catastrophe that grabs headlines and grants.
It gets worse because the mismatch between models and observations goes beyond the surface warming record. As we’ll discuss next week when we look up, way up, into the troposphere and beyond.
