中文摘要：

通过对自愿报告、强制报告和正常运行安全监测（NOSS）数据的威胁与差错管理（TEM）分析，获得了3类安全信息的威胁、差错和意外状态的分布特征。对比研究发现，强制报告中空中威胁、环境威胁的分布比例（55．46％；16．8％）明显高于自愿报告（32.42％；10．35％）和NOSS（19．26％；11．7％），说明强制报告信息侧重外部威胁与他人错误。自愿报告中“内部威胁-其他管制员”的比例（20．85％）显著高于NOSS（6.91％）和强制报告数据（4．20％），说明自愿报告可有效识别内部协作威胁和认知差错；NOSS对易识别、独立责任的“程序差错－文件”检出率（25％）显著高于自愿报告（2．81％）和强制报告（0％），证实了NOSS信息的客观性，但NOSS对低频度、高危害类差错的检出率显著偏低，说明小样本NOSS信息存在波动性。

英文摘要：

The paper introduced and analyzed the characteristics of several aviation safety data sources in civil aviation. As is known, safety management based on information, but safety related data is unequal to effective information. The risk control strategies calls for effectively and comprehensively use of safety data. By using the framework of threat and error management （ TEM）, operation safety data from database of Air Traffic Control （ATC） voluntary safety report system （VSRS）, ATC safety compulsory information system （SCIS） and normal operation safety survey （NOSS） archives were analyzed. Different characteristics in distribution of threats/errors/undesired states in three databases have been identified. According to statistics results of the TEM taxonomy analysis, SCIS data has detailed record of occasions in operation, especially the serious incidents, but it emphasizes particularly on ATC external threats and errors from outside i.e. from pilots. 55.46% and 16.8% reported threats from SCIS are airborne threats and circumstance threats compare to which of VSRS（32.42%; 10.35%） and NOSS（29.26%; 11.7 % ） . While internal threats are 8.40% of whole threats in SCIS data compare to that of VSRS（33.94% ） and NOSS（27.65% ）. Focusing on ATC internal threats and cognitive errors, VSRS can effectively find cognitive and co-operative threats that are not easily observed in operation, 20.85% of reported threats of VSRS are attribute to threats from other controllers, obviously higher than that of SCIS（4.20% ） and VSRS （ 6.91% ）. But VSRS still have report bias. Threats from Airborne-Pilots in VSRS （12.94%） and SCIS （ 17.65% ） are higher than that of NOSS（ 7.45% ） which is believed to be reliable and objective in recording threats. NOSS, as a thirdparty monitoring method, have advantaged in comprehensive information and small deviation. 25% of errors identified from NOSS archives are procedure errors in using documents which is solely responsible of controllers, compar