Reports of daily precipitation from stations around the world have been collected by NOAA Climate Prediction center (CPC) and utilized in various applications in real time climate monitoring and assessments, verifications of official climate forecasts, and diagnostic studies of climate variability. As a critical component of CPC's data construction efforts, these daily precipitation reports are quality controlled through a set of objective procedures before they may be used to construct analyzed fields of regional and global precipitation.

The first quality control (QC) system was developed by Higgins et al. (1996) for regional station networks over United States, Mexico, several countries in South America. After duplicate stations are eliminated, this automatic system checks suspicious daily precipitation reports of ‘0' and extreme values through comparisons with a) concurrent precipitation reports at nearby stations (buddy check), b) standard deviation of historical records, and c) radar observations (when they are available). Quality controlled reports are then used to create analyses of daily precipitation over the regions.

Recently, a new system has been developed at NOAA/CPC to perform quality control for the daily precipitation reports from the Global Telecommunication System (GTS). Based on the QC system for the regional networks, this new system examines the quality of GTS reports through comparisons with a) concurrent reports from nearby stations, b) climate statistics at the target station, c) satellite observations, and d) precipitation forecasts from the NCEP Global Forecast System (GFS). A risk index is computed for each daily precipitation report with ‘0' or extreme value as a weighted mean of the probabilities of being a bad report based on the individual comparisons. The weights are defined as a function of local GTS gauge network density, regions and seasons to quantify the reliability of the independent information used in verifying the daily precipitation reports at the targeted station. A set of tests have been conducted to compare the results from this objective system with those based on manual operations. The results demonstrated reasonable skills of our system in identifying GTS reports of bad quality. The new system has been implemented and put into real-time operations from May 2007. Details of the system will be reported at the meeting.