volume ) is plotted as a function of the density deviation. In experiment A, B and C, the total of forcing areas and the intensity of the density flux are kept unchanged. As the forcing area is clustered closely (C → B → A), the peak of the histogram shifts toward a larger density deviation.
The reason of this peak shifts is suggested as flows: We first compare experiments A and C. The border between the convected region and the exterior of experiment A
is almost one forth of the total length of the borders of experiment C. The effects of the non-convected light water hardly reach the interior of the large forcing disk in experiment A. Therefore, heavier water is formed near the center. However, the smaller disks in experiment C are easily mixed with the exterior water. Since the forcing disks are located very close in experiment B, the non-convected water becomes heavier than that of experiment C.
In experiment D, the forcing disk is almost same as the area of the square in which small forcing disks are clustered in experiment B. Because of a difference in density flux intensity, plume development is more slower than experiment A. In experiment B, each forcing disks gives larger density flux than the same area in experiment D. In other