Abstract
Seasonal predictions of rain or snowfall are usually too uncertain at regional scales. We suggest utilizing subsurface ocean measurements to improve long-term weather forecasts. The example we give is that regional snowfall in Japan can be predicted by a simple regression from an acoustic Doppler current profiler attached to a regular ferryboat to observe the transport of the Tsushima Warm Current. The lag correlation is shown to exceed 0.75 attributed to the simple underling marine meteorology and regional oceanography. The relationship certainly improves seasonal precipitation estimates led by the winter monsoon absorbing the latent heat from the Japan Sea. We predict there will be less snowfall this winter of 2005/2006 than in 2004/2005 despite the heavy snowfall event in the last December.
