Automated quality control for hydro-meteorological sensor data

By Rick Hagenaars

The TAHMO network performs hydrological and meteorological observations using automated weather stations. These observations are used for example by weather forecasting and decision support systems, which rely on the quality of the data that is provided to them. Sensor recalibration, proactive maintenance and both automated and manual quality control are essential to achieve high quality data. The spatial and temporal variability in the measured parameters present challenges for the quality control and quality assurance procedures, on which this thesis will focus.

Quality control of meteorological data is generally split into multiple components occurring at different time spans. As soon as the data is assimilated statistical tests are applied to identify data outliers, changes in variance structures and jumps in data values. During this research it’ll also be attempted to use correlations between parameters. This first component is run near real-time and provides a basic level of quality assurance that is used to deliver the data to systems which require data as early as possible. Once data is acquired from multiple stations in a specified proximity, spatial checks will identify inconsistent data in relation to nearby stations. The last automated quality procedure consists of comparing the data from ground stations to radar data. The combined automated tests will set quality flags which will help to determine which data need further analysis during the manual quality control.