Data quality checking for single station meteorological databases

Abstract

In the past decade individual and networks of automated meteorological stations have been installed throughout the United States and many other countries. For a variety of reasons, the data collected are being archived in databases; however, quality control/quality assurance procedures, when employed, vary greatly. As a start to possible standardization, screening rules for hourly and daily data values are proposed for quality checking micrometeorological data from individual base stations that record solar irradiance ( SI), precipitation ( P), barometric pressure ( P b), vapor pressure ( e), wind speed ( u 2), wind direction ( θ 2), air temperature ( T a), and three soil temperatures ( T s0, T s1, and T s2). Three types of screening rules are considered: (1) high/low range limits (LIM), (2) rate-of-change limits (ROC), (3) continuous no-observed-change with time limits (NOC). Daily data from historical meteorological records for Ames, IA (30 years) and Treynor, IA (26 years) were available for developing climatic based dynamic data screening rules. Otherwise, instrument specifications and theoretical models were used to develop screening rules for the remaining measurements. Hourly and daily data from well maintained, automated weather stations at Walnut Creek, IA (9 months) and Treynor, IA (1 year) were used to evaluate and refine the screening rules. Daily data are not flagged often. The most common flag, on either time-scale, was on vapor pressure when its value exceeded the 95% relative humidity calibration limit of the sensor. Hourly SI often exceeded a computed extraterrestrial radiation value, particularly at sunset. Rule 1 (LIM) is mainly invoked via observations outside the sensor ranges; rule 2 (ROC) flags abrupt changes; rule 3 (NOC) flags unusually steady periods in the data stream. When used as part of a total field operation and data processing system, these rules improve the data quality and may help with data exploration.