"As you learned earlier, meteorologists compensate by implementing a technique called ensemble forecasting, which gauges the sensitivity of a computer model's prediction to the way it's initialized. Specifically, meteorologists make minor changes to the initialization of a lower-resolution version of a specified operational model called the control member. For example, the control member in the GEFS has lower resolution than the operational GFS model. At any rate, the control member is run using this slightly different initial state. Then meteorologists tweak the initialization of the control member yet another time, in a slightly different way, and run the control member again using this new initial state. This process of "tweaking" the initial conditions of the control member is repeated a number of times (for some models, several dozen times), yielding a set of ensemble members.
If all or most of the ensemble members come up with basically the same numerical prediction for a specific forecast day, meteorologists have a relatively high degree of confidence in that day's forecast. If, however, the tweaked model runs predict several noticeably different scenarios for the day in question, then forecasters have a fairly low degree of confidence in the numerical prediction." (Source)