Accounting for Uncertainties in NOAA Atlas 14 Precipitation Frequency Estimates

The Hydrometeorological Design Studies Center (HDSC) within the Office of Water Prediction of the National Oceanic and Atmospheric Administration’s (NOAA) National Weather Service (NWS) has been updating precipitation frequency estimates for various parts of the United States and affiliated territories in volumes of NOAA Atlas 14, Precipitation-Frequency Atlas of the United States since the 2000s. All NOAA Atlas 14 estimates are accessible through the Precipitation Frequency Data Server, along with 95% confidence limits and other supplementary information.

NOAA Atlas 14 precipitation frequency estimates are calculated from sample data and represent expected values, but there is a high probability that true values lie above or below the sample estimates. The purpose for publishing confidence limits on NOAA Atlas 14 estimates is to help users recognize that actual values might be different from expected values and to encourage them to look at a range of possible scenarios in their designs.

The widths of confidence intervals between the upper and lower confidence limits are affected by a number of factors. Confidence limits in NOAA Atlas 14 are calculated using a simulation-based method which accounts for some of those factors, such as a degree of confidence, frequency and sample size. However, it fails to account for additional factors, such as distribution selection and parameterization method. In addition, the current NOAA Atlas 14 approach assumes stationarity in the annual maximum series data used in the analysis, and therefore does not characterize uncertainty in the presence of non-stationary climate conditions.

HDSC has been assessing the potential impacts of different sources of uncertainty on NOAA Atlas 14 estimates at scales of interest to engineering designs, particularly with regard to non-stationary climate conditions relative to other factors. Initial findings will be discussed in the presentation.