A forecast model is only as good as the data put into the model. The synoptic scale models rely on data from these sources: Rawinsondes, aircraft observations, surface observations, satellites, and other remote sensing sources. It is over remote locations and just downwind from remote locations that forecast models have the greatest error. The three most important remote location regions are mountainous terrain, desert terrain, and the oceans. Over these three locations, surface observation stations are very scattered. Upper air data is not as dense either. The locations near the U.S., that have sparse data, are Mexico, and the oceans.

Storm systems that move from the ocean or from Mexico are generally not handled as well by the forecast models as they could be. Examples include low pressures moving from Mexico and into the U.S. and a trough of low pressure moving from the Pacific Ocean and into the U.S. Over the last few years, satellite data has increased the accuracy of forecast models and allowed a better sampling of remote locations. Satellite data can also be used for numerous forecasting purposes, especially severe weather. Examples of meteorological data that satellites provide are CAPE, LI, cloud top pressures, sea surface temperatures and temperature profiles of the atmosphere. Although these indices and temperature profiles are not as accurate as the rawinsondes, the data is accurate enough to be used for forecasting purposes and adds good data to the models. Rawinsondes can only sample at point locations, but satellites can cover a broader area more uniformly. One of the primary limitations of satellite data is it can not detect meteorological conditions below clouds. Only in clear sky or nearly clear sky conditions can the satellite sensor detect CAPE, LI, temperature profile of the atmosphere, etc. You can find ALL current satellite derived products at the following web site.