I thought that science made conclusions based on observations, not that it made observations based on conclusions
It is completely normal that conclusions lead us to doubt our observations and do new measurements. Observations can be misleading unless we have a proper theory with appropriate measurements.
As for the current problem: Computer models are not reality, but - apart from data from the past - they are the best we have.
Science works like this:
First you observe something.
Then you try to come up with an explanation. That explanation needs to make some predictions (or it would be useless).
Then you test those predictions. Usually by making experiments.
Now, here we have a problem. We don't have a second earth to experiment with. And even if we had, the timescales involved are too large to make experiments in real-time.
So instead we use climate models. But these models are not reality. These are models we come up with on our own. How do we do that? Well, people that are generally learned about the subject try to think about anything that could possibly affect the climate. They then create a mathematical model and see if it's predictions fit the known data (i.e. data from the past must predict the present). The better it does, the better the model.
Now it turns out that we have some wrong data. Obviously that wrong data will have lead to climate models that do not predict reality.
This means we have to alter our current climate models to fit the new data. Someone will have to come up with an idea what exactly is wrong with the models and how to fix them, of course. But that is no different from how the models were created in the first place.
It also does not mean the models are not useful. They are. As long as they accurately predict current climate from past data, we can assume that they will also predict *future* climate from *present* data.
But, yes, our predictions can at best be as accurate as our observations. And if we measure wrong, that is a problem.
Disclaimer: IANACS - I am not a climate scientist