OK so I have to know, what are the actual uncertainty levels of the forecasts with and without the dropsondes? Are we talking like 50% and 5% or what?
It really depends on the synoptic situation and where the break in the ridge is forecasted to occur.
If the break in the ridge is going to occur over the CONUS - then we are talking 5% because we have tons of data over land. IF we are talking eastern Atlantic than 30% perhaps because we have zero data - but we have a dominant ridge there.
If we are talking east coast and the fall when we start getting trofs moving over the area more frequently - then we can get a 50% error because we have 1) no data over the oceans other than Bermuda and 2) we are in a transition season.
So - it can be a nightmare. To top it off - the NHC wants to have the SMALLEST error as possible on their forecasts. Now - you may think - well - DUH - that's what EVERYONE wants - but that is not ALWAYS the best forecast.
For example - they usually ALWAYS split the model difference - they will forecast a model blend, rather than stick to a "model of choice" - because simply - over the years it's been proven that that has better odds of being closer. But - it's not always GOOD for emergency managers.
Back several years ago the Euro model and its ensembles were taking a hurricane into Florida and the GFS (and its group) was taking a storm into Mexico. It all depended on whether a trof would come and pick it up (Euro) or not (GFS). The one thing is WOULD NOT DO was go straight to the NW into Lousiana...which was the consensus of the model tracks.
SO - what did the NHC do? You guessed it. They put out a track taking it into NOLA...and we have to start evacuation procedures for NOLA! KNOWING it wouldn't go there and that that was the one place it would NOT go. Try explaining THAT to Janet Napolitano.