For approximately 30 years, the National Weather Service (NWS) has operated a national verification program in which selected NWS forecast products are routinely verified. This sample of forecasts is chosen to represent the quality of public and aviation weather services provided by the NWS. Verified forecasts include maximum/minimum (max/min) temperature, probability of precipitation for a 12-h period (PoP), precipitation type, snowfall, wind direction and speed, cloud amount, ceiling height, and visibility. Forecasts produced by each NWS Weather Forecast Office (WFO) are evaluated for one or more cities in the WFO area of responsibility, and the skill of the human forecast is generally compared to that provided by the Model Output Statistics (MOS) guidance for the same weather element.

While the goals of the verification program remained constant during these 3 decades, many of the details changed substantially. Data collection, for instance, went from a manual process to a completely automated system. The aviation forecasts of wind direction, wind speed, ceiling height, and visibility were issued at different initial times during the period of record, according to the rules for the aviation terminal forecasts. Observations of cloud amount became problematic when human observers were replaced by reports from the Automated Surface Observing System (ASOS), complemented by satellite-based estimates of clouds. Reports of precipitation type were also affected to some degree by the introduction of ASOS. ASOS was unable to report snowfall, a situation exacerbated when the METAR standards for surface observations were implemented, thus eliminating the requirement for snowfall reports. Unlike these other elements, however, the characteristics of the max/min temperature and PoP forecasts and their verifying observations stayed relatively constant during the period. Thus, the time series of the max/min temperature and PoP verifications provide one look at the evolution of skill in NWS public weather forecasts from the early 1970's to the present.

Although many changes have occurred during this 30-year period in both the observing system and the organizational structure of the NWS, a core set of approximately 90 stations in the contiguous United States is available for analysis. In this paper, we look at the skill of the NWS max/min forecasts compared to the MOS guidance for four forecast periods of the public weather product. We also examine the skill and the reliability of the PoP forecasts for three forecast periods. As expected, the skill of the NWS max/min temperature and PoP forecasts has increased significantly during the 30-year period. While the skill of the first period forecasts has leveled off during the last decade, the skill of the third period forecast is now comparable to that of the first period forecast provided 20 years ago.