That could depend on what the meteorologist has on his/her disposal. They could have their own team who gather data-information and run models to arrive at a forecast for the weather, or they may simply not have that luxury and simply rely on external sources to prepare weather reports. The models might be well calibrated and tested (or not), and the external source of info could be very reliable (and vice versa).
Either way, like any other form of 'prediction', accuracy remains a challenge: you look at the most recent climate history (say, the past couple of weeks) and also historical data from prior years to have an idea how the weather is 'expected' to behave; still, there is a non-trivial amount of ambiguity and potential error that could result in erroneous weather forecast. We all remember those occasional cases where a sunny day was predicted, only to end up getting soaked in the rain!
Hope this helps. Cheers.