A database of numerical weather prediction output, satellite data, climatology, and ground truth observations is being created in an attempt to accurately diagnose and potentially forecast sensible weather elements. The U.S. Navy's mesoscale model, COAMPS (Coupled Ocean/Atmosphere Mesoscale Prediction System), is triply-nested (81, 27, and 9 km) and run twice daily (12-hour forecasts) for three areas: U.S. west coast, Adriatic Sea, and Korean peninsula. COAMPS output parameters, coincident satellite parameters (including both geostationary and polar-orbiting data) and climatological information are extracted/computed at 45 METAR observation sites. Automated data collection routines have been written and data has been collected hourly since July, 2000. Initial data mining has begun. These techniques will help to uncover patterns among the disparate data types (related to a specific weather parameter). These relationships can assist in diagnosing a sensible weather parameter (e.g., cloud ceiling height), specifically in locations where observations are not available. Initial results related to cloud ceiling height and rain rate will be presented.