Evaluation of 33 Years of Quantitative Precipitation Forecasting at the NMC

View More View Less
  • 1 Weather Forecast Branch, Meteorological Operations Division, National Meteorological Center, Camp Springs, Maryland
© Get Permissions
Full access

Abstract

The National Meteorological Center (NMC) initiated Quantitative Precipitation Forecasts (QPF) and an intensive QPF verification program in 1960. These forecast products have evolved from a manual effort, relying on extensive forecast experience to one that placed much greater reliance on the interpretation and modification of numerical models.

Verification graphs show steady improvements in forecast accuracy, especially for the longer-range forecasts, which in this context am those in the 24–60-h range. During the 1960s the Threat Score (TS) for day-2 forecasts for 1 in or more of precipitation averaged approximately 0.07. During recent years, that score has nearly doubled, and the 36–60-h period forecast in 1993 had a TS comparable to that for the 12–36-h period during the 1960s. Improvement in accuracy is probably related to a number of diverse factors including improved numerical models, increased forecaster knowledge of the strengths and weaknesses of the operational models, and an increased understanding of precipitation processes. The verification results have been used to track individual and group progress.

Abstract

The National Meteorological Center (NMC) initiated Quantitative Precipitation Forecasts (QPF) and an intensive QPF verification program in 1960. These forecast products have evolved from a manual effort, relying on extensive forecast experience to one that placed much greater reliance on the interpretation and modification of numerical models.

Verification graphs show steady improvements in forecast accuracy, especially for the longer-range forecasts, which in this context am those in the 24–60-h range. During the 1960s the Threat Score (TS) for day-2 forecasts for 1 in or more of precipitation averaged approximately 0.07. During recent years, that score has nearly doubled, and the 36–60-h period forecast in 1993 had a TS comparable to that for the 12–36-h period during the 1960s. Improvement in accuracy is probably related to a number of diverse factors including improved numerical models, increased forecaster knowledge of the strengths and weaknesses of the operational models, and an increased understanding of precipitation processes. The verification results have been used to track individual and group progress.

Save