Extended Abstract (284K)7.2
Strategies for evaluating quality control procedures
Imke Durre, NOAA/NESDIS/NCDC, Asheville, NC; and M. J. Menne and R. S. Vose
Users of quality-controlled meteorological data need to know whether the QC procedures have inadvertently removed true climate extremes and whether errors remain that might compromise the analysis. This information can be provided via a thorough evaluation of the type I and type II errors, i.e., the degree to which a QC process identifies good observations as erroneous and the extent to which known errors remain undetected. This paper outlines three components of such an evaluation process. The approach relies on manual inspection as a tool for (1) the selection of appropriate thresholds for individual procedures, (2) the examination of patterns in flagged values, and (3) the determination of the type I and type II error rates. An "extremes check" for daily precipitation totals is used as an example for illustrating the approach. the methods discussed have been applied to several data sets available from the National Climatic Data Center, including the Integrated Global Radiosonde Archive and the Global Historical Climatology Network.
Session 7, Data Quality Control / Metadata
Thursday, 18 January 2007, 8:30 AM-9:45 AM, 207A
Previous paper Next paper