A trial production model of cockpit voice recorder (CVR) was evaluated in laboratory and flight. The trial equipment is the same size and electric interfaces as the conventional CVR but has four times more recording capacity. The recorder uses a recording system with two cassette tape decks that is preserved the last two hours of recordings.

This paper describe the performance of the improved equipment, the methods and results of the test and the quantitative analysis of the syllable articulation accuracy both of the new and conventional CVR's recordings.

This paper proposes an analytical evaluation method for establishing a radar separation minimum in air traffic control (ATC) environments where a monopulse SSR (secondary surveillance radar) is used for surveillance. The SSR, namely Long Range SSR (LSSR), is supposed to be used for the surveillance of an oceanic airspace and its coverage is extended to 250 NM.

As a relative measure of the collision risk due to azimuth errors of the LSSR, the close approach probability (CAP), i.e., the probability that a pair of aircraft overlaps in azimuthal direction is used for evaluating the safety of the surveillance system. The CAPs for both the LSSR and the conventional SSR are estimated through the modeling of azimuth distribution errors. These CAPs are compared each other from statistical view points.

The results show that the CAP for the LSSR is significantly smaller than that of the conventional one under 5 NM radar separation minimum environment.

ICAO has studied the possibility of reducing the vertical separation minimum above flight level 290 from 2,000 ft to 1,000 ft on the basis of a collision risk model.

This paper describes results of calculation on the passing frequency, which is a key parameter of the collision risk model, for a Japanese domestic airway. The estimates of the passing frequency obtained by three methods are compared.

Then the safety of reducing the vertical separation minimum is evaluated by using the estimated parameters and the collision risk model. The results are as follows

1) No remarkable difference is observed among passing frequency values estimated by three methods.

2) These passing frequencies for opposite direction traffic are approximately 3.

3) An estimated collision risk based on the above mentioned value is 4×10^-7 [accidents/flying hours].