You're going to have to present a stronger case than "if you read the paper that they publish, they talk about error margins."
Error margins are introduced because they do not have the computing power (nor models of sufficiently good quality) to compute every molecule in the atmosphere. Instead they have to divide the planet into a grid of cells, then model the cells as the smallest unit in the simulation. The processes that occur within these cells are only approximated and not actually computed, therefore there is some margin of error. This is typical practice for simulations and doesn't prevent them from being highly accurate.
Furthermore, scientific theories yield predictions, "which then turn out to be false" ALL THE TIME. In fact it too is standard practice for the field of science. Science never establishes certainty, it just hopes to paint the most accurate picture possible (nevermind the specifics of what accurate means (to save me from writing a philosophy of science essay to inform you some more)).
At the end of the day your argument that the simulations outstrip the predictory power of the data is refuted. Not only don't they exceed the data, but they've been increasing the quality of their data on a daily basis.
PS- you may be interested to note that some climate models have refuted existing data, which initially was taken as evidence against the model, until scientists figured out that the model was more accurate than the data! Turns out the weather balloons they were using were sensitive to atmospheric conditions which influenced the readings they generated. Sometimes models are more accurate than the data!