This work involves automatic quality control and quality assurance of data for very large scale rainfall and stream flow measurements collected on a national scale. The work is applicable to sensor networks implemented on a large scale. The technical contributions include the development of robust interpolating multivariate spatio-temporal models that can provide expected values by condition on identical measurements taken locally in space and time, as well as on different but related measurements. Actual measurements are then compared to these expected (interpolated) values to check the quality of measurements both in real-time and batch mode. Outliers are flagged for manual checking. The impacts of this study include, among others, substantial reduction in the manual effort normally required to check data designed to improve its quality.

Keywords: Data quality; Interpolation; Multivariate spatio-temporal models; Sensor networks

Biography: Dr. Ross Sparks (Australian Accredited Statistician) has research interests that include statistical process control, applied multivariate analysis and applied statistical modelling. In the area of statistical process control Ross has published more than 25 papers in international journals. In the area of applied multivariate analysis and applied statistical modelling, Ross has published 16 papers in refereed journals that are cited in the Scientific Citation Index. He was worked for CSIRO Australia for the past 20 years, and prior that he lectured at universities for more than 13 years.