Spatial regression as a technique for assessing the quality of temperature data

A spatial linear regression technique was developed by the High Plains Regional Climate Center to assess the quality of daily temperature data. A station's data is compared to those data from neighboring stations to identify suspect observations. The technique is being implemented as part of the operational quality assessment of daily temperature data at the National Climatic Data Center. This paper describes the evaluation of the technique for operational use, as well as compares the utility of the technique to other assessment tools. Emphasis is placed on the relationship between parameter estimation and the degree of manual intervention needed in a mostly automated system.