Data quality improvements for FAA

摘要

Effective communication among air safety professionals is only as good as the information being communicated. Data sharing cannot be effective unless the data are relevant to aviation safety problems, and decisions based on faulty data are likely to be invalid. The validity of aviation safety data depends on satisfying two primary characteristics. Data must accurately represent or conform to the real world (conformance), and it must be relevant or useful to addressing the problems at hand (utility). The FAA, in efforts to implement the Safety Performance Analysis System (SPAS), identified significant problems in the quality of the data which SPAS and FAA air safety professionals would use in defining the state of aviation safety in the US. These finding were reinforced by Department of Transportation Inspector General and General Accounting Office investigations into FAA surveillance of air transport operations. Many recent efforts to improve data quality have been centered on technological solutions to the problems. They concentrate on reducing errors in the data (conformance), but they cannot adequately address the relationship of data to need (utility). Sandia National Laboratories, working with the FAA's Airport and Aircraft Safety Research and Development Division and the Flight Standards Service, has been involved in four programs to assist FAA in addressing their data quality problems. The Sandia approach has been data- driven rather than technology-driven. In other words, the focus has been on first establishing the data requirements by analyzing the FAA's surveillance and decision-making processes. This process analysis looked at both the data requirements and the methods used to gather the data in order to address both the conformance and utility problems inherent in existing FAA data systems. This paper discusses Sandia's data quality programs and their potential improvements to the safety analysis processes and surveillance programs of the FAA.