25.12.2014 Views

East Asia and Western Pacific METEOROLOGY AND CLIMATE

East Asia and Western Pacific METEOROLOGY AND CLIMATE

East Asia and Western Pacific METEOROLOGY AND CLIMATE

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

509<br />

To investigate the statistical property of the quasi-stationary<br />

solutions, we take a monthly running mean with a 20-day overlap <strong>and</strong><br />

regard it as a sample of monthly-mean model climate. Traveling waves<br />

are almost averaged out in the monthly means, with locked waves<br />

remaining. To examine the distribution of their probability density<br />

functions, we evenly divide a range of each monthly-mean variable<br />

into twenty intervals <strong>and</strong> count the frequency of appearance for the<br />

variable in each interval. The number of frequency divided by two<br />

hundred is the probability of this variable in each interval. Fig. 6<br />

shows the contours of the probability distributions for monthly-mean<br />

AT, 4>£i' anc * ^W2" There are two well-separated peaks in the probability<br />

distribution, i.e., it is distinct bimodal distribution. The<br />

well-defined bimodal distribution demonstrates that the system has<br />

two preferred states for winter which resulted from two different<br />

attractor basins pulling the solutions toward them.<br />

In an attempt to underst<strong>and</strong> how the system is attracted by<br />

Branch B <strong>and</strong> Branch C throughout the__seasons, we traced the annual<br />

cycles of the two sub-ensemble mean AT (starting from summer in the<br />

AT) vs. AT diagram of statistical Equilibria (Fig. 7). During<br />

summer the two mean AT are very close <strong>and</strong> lower than AT for Branch B.<br />

During early fall, the system has two possible ways to go: one is to<br />

follow a branch possessing a larger AT, <strong>and</strong> the other is to follow a<br />

branch possessing a smaller AT. In the later part of the year (i.e.,<br />

winter), the smaller AT solution stays near Branch B, while the<br />

larger AT solution can no longer be maintained in late fall <strong>and</strong> is<br />

attracted into Branch C. We interpret the split routes for the<br />

system solutions are due to transition behavior of Branch B solutions<br />

around AT = 40°K (not shown) . One type of the present solutions<br />

follows the upper AT branch of the transition solution, <strong>and</strong> the other<br />

type of solutions follows the lower AT branch of the transition<br />

solution. As we have found in the fixed AT timeintegrations that<br />

the lower AT branch of the transition solutions remain in Branch B<br />

as AT increases, the smaller AT solutions are stable <strong>and</strong>__stay near<br />

Branch B. On the other h<strong>and</strong>, as AT increases the upper AT Branch of<br />

the transition solutions cannot be maintained <strong>and</strong> thus the solutions<br />

attracted into Branch C in the following season. The self-transition<br />

from Branch B to Branch C are found in fixed AT integrations which<br />

indicates that Branch C is more stable than Branch B, Therefore, the<br />

system is attracted into Branch C rather than Branch_J3 when it is<br />

unstable. Whether the system will follow the upper AT branch or<br />

the lower SF branch in the transition range determines the system<br />

behavior in the following fall, winter <strong>and</strong> early spring seasons. We<br />

consider that the choice by the system to follow one of those two<br />

branches during the transition range is stochastic. However, the<br />

predictability of different winter types is at least one season<br />

earlier as the zone mean flow of the two type solutions is distinctly<br />

separated in fall (see Fig. 6).

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!