This study introduces an algorithm designed to detect barometric pressure setting (BPS) errors in large-scale flight data. The algorithm leverages Automatic Dependent Surveillance-Broadcast (ADS-B) data to detect deviations from the expected BPS values. A statistical model incorporating pilot-selected QNH values is developed to estimate local QNH variations, enabling precise error identification. Unlike previous studies that rely on Meteorological Aerodrome Reports (METAR) data, which may be limited or unavailable, particularly at smaller airports, this approach uses ADS-B data for both inferring local QNH and identifying outliers, enabling large-scale analysis. The QNH inference principle is based on a moving median over a subset of flights near a given location and time. Although a fixed time window presented promising results, a performance enhancement has been achieved by dynamically selecting the optimal moving median window size for each dataset. To validate the accuracy of the inferred QNH values, the results were first compared against METAR data. This comparison demonstrated strong agreement between the inferred and observed QNH values, establishing the reliability of the proposed method. Further validation was conducted using two well-known case studies from recent safety investigation reports. ADS-B data for the same day and area were processed, and the algorithm accurately identified the same BPS error, confirming its capability to detect and locate such errors. Following the validation of the algorithm, an analysis was conducted using ADS-B data from multiple airports of EURO-CONTROL member states over a 31-day period. The analysis identified multiple cases that require further investigation and gathered statistics on these errors. With further analysis of error patterns, the study provides insights into the potential underlying causes of these errors. This is achieved by creating new error categorization, examining the error frequencies, using string metrics such as the Levenshtein distance. By processing data on a large geographical scale, this study offers a comprehensive overview of the issue and raises awareness about the prevalence of BPS errors, regardless of their severity.
An Algorithm for Identifying Altimeter Setting Errors from ADS-B Data
2025-04-08
3837483 byte
Conference paper
Electronic Resource
English
NTIS | 1969
|Engineering Index Backfile | 1943
|