They are observations plus modelled corrections. Here's a simple thought experiment. How would you accurately measure the average temperature in a kitchen? Res

They are observations plus modelled corrections. Here's a simple thought experiment. How would you accurately measure the average temperature in a kitchen? Residential? How about commercial? How many sensors would you need? Where would you locate them? How about in a gymnasium when it's empty? Or during a basketball game with packed stands? How about the State of California? The USA? The world? Climate scientists have to think about all of these things, and sadly there isn't one right answer. So they end up making assumptions that are based on their models. That's right, the model drives the measurement! Supposedly those techniques are robust, but naively it feels like there's definitely room for some pretty significant mistakes to be made.