J6.3 Comparison of Manual and Automated Quality Control of Operational Hourly Precipitation Data of the National Weather Service

The National Weather Service (NWS) Office of Hydrologic Development operates the collection and dissemination of real-time Hydrometeorological Automated Data System (HADS) and other precipitation gauge data to users at River Forecast Centers (RFCs) and Weather Forecast Offices (WFOs). As most of the data are delivered to the users with minimal quality control (QC) in order to shorten data latency, forecasters at RFCs apply significant effort toward QC to insure proper hydrologic forecasting. Thus there is a great need for automated gauge QC. The NWS Lower Mississippi RFC (LMRFC) has been generating quality codes of hourly precipitation data available within the service area, and made available to WFOs and to the National Climatic Data Center (NCDC). The dataset (called “LOUZE” data) is unique as it represents a refined class of quality control algorithms. As a general rule, the results of manual editing are considered to be of significantly higher value and accuracy than those resulting from automated technique. But manual editing is susceptible to human inconsistencies, biases, and just plain errors that automated processes are not. Three NOAA agencies, NCDC, Earth System Research Laboratory (ESRL) and National Severe Storm Laboratory (NSSL), are engaged in automated QC practices for precipitation data. The NCDC’s objective is to apply emulated manual QC algorithms to historical HADS hourly data (Kim et al. 2008); the NSSL’s objective is to refine the next generation quantitative precipitation estimate (Q2; Vasiloff et al. 2007) to use radar-based QPE as a tool to improve QC of precipitation data; and the ESRL’s Global System Division (GSD) focuses on the quality history of gauge stations used in precipitation assimilation and verification (Tollerud et al. 2005). Even though the objectives of these LOUZE data users are different, from these studies many lessons toward