I’ve been using the free version of Predict Wind for a while but had not used it to track yet to track significant weather events. I was considering to eventually subscribe to it when I started permanently cruising. I used it recently to monitor Hurricane Isaias, the accuracy was very disappointing.
Predict wind offers two public models, GFS and ECMWF and offers two proprietary models based on GFS and ECMWF. Using it to follow Hurricane Isaias the GFS and ECMWF models were as expected, they each had their own forecasts, but were relatively accurate in their own. The two Predict Wind models had the storm crossing the state of Florida and entering the Gulf of Mexico. At the same time, I was also monitoring the spaghetti models and none of which had the storm crossing the state. Even as the storm moved up the coast of Florida the perfect wind models still had the storm significantly west of all the other weather models. Less than 24 hours prior to the storm reaching my location the spaghetti models all had the storm East of my location while the Predict Wind models still had the storm passing west of me, a location difference of 100s of miles.
I will admit that this is just one observation of the accuracy of the Predict Wind models, but there was such a significant discrepancy from the Predict Wind models vs the various spaghetti models. If you were on the west coast of Florida the impact would have been at most an annoyance because of a delayed voyage or a canceling of a trip. If on the East Coast if could have been catastrophic, you could have sailed right into the storm. Now we all know you should always look at multiple sources to plan with, but if you relied on Predict Wind as your primary source, using their models, you may have sailed right into a storm.
I was considering subscribing to their service, but now I am in doubt. I will continue to use their free service and evaluate their accuracy. Maybe they are more accurate with non-named storm weather or maybe this was a one-off miscalculation. I will continue to use their service to evaluate their accuracy and evaluate if in future storms they are more accurate. It may be a mild inconvenience if Predict Wind mis-predicts the difference between 15 knot wind and 5 knot wind, but it is life threatening if Predict wind mispredicts the path of major events.