An Improved QC Process for Temperature in the Daily Cooperative Weather Observations

Abstract

TempVal is a spatial component of data quality assurance algorithms applied by the National Climatic Data Center (NCDC), and it has been used operationally for about 4 yr. A spatial regression test (SRT) approach was developed at the regional climate centers for climate data quality assurance and was found to be superior to currently used quality control (QC) procedures for the daily maximum and minimum air temperature. The performance of the spatial quality assessment procedures has been evaluated by assessing the rate with which seeded errors are identified. A complete dataset with seeded errors for the year 2003 for the contiguous United States was examined for both the maximum and minimum air temperature. The spatial regression quality assessment component (SRT), originating in the Automated Climate Information System (ACIS), and TempVal, originating in the NCDC database, were applied separately and evaluated through the ratio of identified seeded errors to the total number of seeds. The spatial regression test applied in the ACIS system was found to perform better in identifying the seeded errors. For all months, the relative frequency of correct identification of wrong data is 0.72 and 0.83 for TempVal and SRT, respectively. The goal of the comparison was to evaluate quality assurance techniques that could improve data quality assessment at the NCDC, and the results of the comparison led to the recommendation that the SRT be included in the NCDC quality assessment methodology.