The scientists can do models, but the models are only as accurate as the data and if theres something you're missing (like some parameter or effect you don't know about), then the model might not be very accurate either.
Eh, this is true for simple models, but when looking at something like climate...
We are talking about a genuinely chaotic system. As in
Chaos Theory. As in;
Chaos: When the present determines the future, but the approximate present does not approximately determine the future.
We simply can't create a model of reality that is anything but an approximation. The most complex and precise model has to approximate individual atoms to general rules of fluid dynamics or some other generalisation. Climate models in general are based on grids of cells (
this site has some nice illustrations), approximating the interactions between them based partially on observed, partially on calculated behaviour.
You can get reasonable predictions out of them, don't get me wrong, but there is a reason you don't make weather forecasts (even on a national level) for longer than seven days into the future. It's why most grand climate models (particularly the IPCC reports) are general temperature range projections, as in the lower image in that last link.
@palsch: Yet politicians do that all the time.
Only in a few fields are the outcomes of their decisions so widely ranging, have such a substantial impact on lives or are so irreversible. I'd say that environment/climate are
maybe second to decisions about war in that field. Economic decisions (the usual area where politicians show contempt for facts) are trivial in comparison, reversible in months or years and usually having only second-order deaths involved.