Veryard & Ebdon¹⁾, and Reed²⁾ showed that east and west winds appear cyclically in the tropical stratosphere. Because one cycle is about 26 months, which is a little longer than 2 years, this phenomenon is called the quasi-biennial oscillation (QBO).

Lindzen & Holton (1968)³⁾, and Holton & Lindzen (1972)⁴⁾ showed that this phenomenon is caused by internal waves that propagate upwards from the tropopause. Because the sign of the coriolis parameter changes across the equator, two kinds of waves exist on the equator; one that propagates eastward, and another that propagates westward. Although the mechanisms of these waves differ from each other, the waves carry the momenta of the directions of their phase velocities. As these waves are attenuated, these momenta remain as mean flows. The rate of attenuation is significantly affected by the mean flow, and the two waves disappear at different locations. As a result, the waves accelerate the mean flow in the opposite direction, thereby producing the QBO.

Plumb and McEvan⁵⁾ examined this mechanism with a tank experiment in 1978. Density stratification was produced using salt water to represent the stratosphere. After internal gravity waves were excited on the bottom of the tank with a water pump, the induced mean flow changed its direction with a period much longer than that of the gravity waves.

Although the record of this experiment was widely distributed as 16-mm films, there has been no report of a double check on this experiment during the past 20 years. Plumb & McEvan visualized mean flow using floats, but not the internal gravity waves carrying momenta.

In this paper, we carried out another QBO experiment to confirm the oscillation of the mean flow using almost the same method as Plumb & McEvan. Furthermore, the internal gravity waves were also visualized and their rates of decrease are compared with the theoretical results of the WKB approximation.