Extended Abstract (24K)2.1
Quality assessment of monthly temperature data
David A. Robinson, Rutgers Univ., Piscataway, NJ; and J. Parlagreco
Daily quality assessments of surface air temperature observations from National Weather Service stations, among others, are unable to recognize subtle, sometimes not so subtle, errors in reported temperatures. Some of these errors are better seen when examining monthly means of maximum and minimum temperatures.
We have developed a straightforward method to identify suspicious monthly means. Standard deviations of differences between monthly means from two nearby stations were computed using a base period of 1961-2000. The quality of an observation from a given station in a particular month and year is assessed using comparisons of monthly differences from approximately 8 nearby stations. When the given station has a difference of greater than 1.5 standard deviations with respect to at least half of its neighboring stations, the observation at that site is considered suspect. Further inspection is done by manually comparing results for each station in a region and by looking at results from recent months to see if a pattern has emerged. Problems at the station may lie in the temperature unit itself, they may be due to persistent observer error, or they could signify a station move in recent years.
This paper will discuss the assessment methodology and provide examples using stations throughout New Jersey.
Session 2, Data Reliability and Usability
Monday, 13 May 2002, 10:30 AM-4:30 PM
Next paper