On the characteristics of the 1993/1994
Geostationary Meteorological Satellite high cloud amount
3.1 Extended empirical orthogonal function of high cloud amount
In order to investigate characteristics of time evolution of the principal spatial structure,
extended empirical orthogonal function( EEOF) analysis with the 20-day window size and
10-day window size is applied to each year .The window size is decided by judgement that the ! 20-day can include the synoptic scale motion and then mean structure in 20 day explains the cquasi-stationary character. Now we can convert the original time-spatial function of high cloud amount to new time-spatial function reconstructed by which n is the number of windows. Then, can be expressed by the sum of the products of the time variation
function, Wit and the space variation functions, as following formula. Then, is consisted of time function by spatial function.
Where, ail is the window averaged space structure with the weight being the eigenfunction is common time structure of windows, and can be determined as the eigenfunction of covarience matrix of .Eigenvector represented by common time structure for the each window represents the time variation within the window size.
The first principle mode (Figure 3) explains the most of variability over 11% of mean total variances for 1993 and 1994. This EEOF mode exhibits relatively stationary phase, while second, third and fourth mode represent wave motion as shown by Figure 3 Figure 4 and 5 represent the time evolution of the spatial structure corresponding to the first principal mode in 1993 and 1994, repectively. The period indicated in figure is from 31 May to 8 August including; Changma season. We can perceive that the cloud band around 20N moves northward toward Korea and Japan during 20 June -19 July. Whereas, in 1994 it is noticed that the cloud band near 20N exists continuously without moving northward.
Figure 3. The spatial structure of the first extended empirical orthogonal function for the five-day mean high cloud amount in 1993.
Figure 4. Same as in Figure 9 except for 1994.
Figure 5. Plots of the spatial structure differences (1994-1993) of the first extended empirical orthogonal functions.
In order to compare between characteristics in 1993 and 1994, we have obtained the differences(1994-1993) of the spatial structure of the first mode in time. It is characterized that the strong negative value around Korea and Japan is shown during 20 June -29 July and the positive activity near 20N is sustained from 20 June. As a consequence, we can infer that the monsoon convective activity around Korea and Japan in 1994 is weaker than that of 1993, in
contrast to the character in north region of east Asia, the convective activity in lower latitude in year in 1994 exists intensely and is stationary in character. One may thus conclude that the convective activity in 1994 doesn't move northward and continue to act at the lower latitude. The climatic variability which is producing this results is characterized as the' contrasting features between 1993 and 1994. It may explain that this characteristics is forced by the constraint of equatorial and subtropical condition.
Unlike, the first mode, the second EOF for 1993 and 1994 are relatively small on the contribution to total variance, however, the mode was produced by different character in 1993 and 1994. In 1993, the wave motion was concentrated on South China Sea, and in 1994 the wave motion was prominently represented at the western and northern borderline of Northern Pacific high. The second mode was not shown in the figures, but in order to out the outstanding period of the wave motion for convective activity over east Asia, Fourier harmonic analysis was performed.
3.2 Fourier harmonic analysis
Fourier harmonic analysis is performed to determine what types of wave m ode are pronounced over the analyzed area. The prevailing mode of intraseasonal oscillation in equatorial region and the different character between 1993 and 1994 are investigated.
Harmonic function which is derived from time series of high cloud amount at each grid point is represented as a function of wave number. Figure 6 shows the power amplitude as a function of wave number obtained from Fourier hannonic analysis. The ten symbols represent the. amplitude of different area separating by two degree longitude from 130E to 148E at equator. . Symbol x at wave number 12 has higher harmonic amplitude than any other symbol in 1993. In 1994, the most dominent symbol is y at wavenumber 3.
Figure 6. The plots of the power amplitude as a function of wave number obtained from Fourier harmonic analysis. The ten symbols represent the amplitude of different area separating by two degree longitude from 130E to 148E.
Then, it is possible to express over the equatorial and subtopical oceans. From the Fourier " analysis applied to the tropical and subtropical region, it is noteworthy that the difference ..between 1993 and 1994 in wave modes is recognized.
Figure 7(a) informs that 15.4 day mode is predominant also in subtrupical region. The 6 daymode is dominant in 1994 as shown Figure 7(b). The 15.4 day mode of the most prevailing mode in 1993 is known as the quasi-biweekly mode. Like this, The most prevailing modes over the central equatorial Pacific and Indian ocean was obtained differently, in 1993 the 61 day ; mode is represent and in 1994 this mode is weak in Indian ocean. Then it is probably a ~ judicious conjecture that the different climatic features between 1993 and 1994 may be claimed .significantly in terms of the 15 day and 61 day modes at lower latitude, raising the possibility that the contrasting monsoon precipitation in 1993 and 1994 may be more fundamentally related to the interaction of intraseasonal oscillations and seasonal variation of convective activities.
Figure 7(a). The prevailing modes of the intraseasonal oscillations over the equatorial and subtropical oceans for the period of 1 April 2 October 1993.
Figure 7(b). Same as in fig. 13 except for 1994.
