PDF《Asia》

10 m height over East Asia (see Section 3), which were obtained from SYNOP reports. The analysis period was from January

1993 to June 2002. The time interval was three hours at the temporally densest observatories. The code numbers of the present weather, ww = 07, 08, 09, 30C35, and 98, were classified as the dust outbreaks in this study. The differences among code numbers indicate the strength or temporal variation of the dust outbreak. However, this difference will not be discussed in this study. A strong surface wind was defined as one with a velocity exceeding 6.5 m/s, which is the threshold for a dust outbreak in many numerical models [e.g., Tegen and Fung, 1994]. Threshold velocities actually exhibit various values with differing land surface condi- tions. However, a constant value was used to define a strong wind in this study, because land surface conditions and the threshold velocity for each land surface condition have not yet been clarified. [7] The frequency of dust outbreaks (hereafter, dust outbreak frequency) is defined as the percentage of the number of dust outbreaks to the total number of observa- tions within a given period at each observatory and/or given region. Similarly, the frequency of strong winds (hereafter, strong wind frequency) is defined as the percentage of the number of strong winds to the total number of observations. 3. Analysis Region [8] Figure

2 illustrates the maximum monthly dust out- break frequency during the last ten years (Jan.

1993 to Jun. 2002) at each observatory. A site is considered to be a GEOPHYSICAL RESEARCH LETTERS, VOL. 30, NO. 14, 1736, doi:10.1029/2003GL017261,

2003 1 Meteorological Research Institute, Tsukuba, Japan.

2 JST Cooperative System for Supporting Priority Research, Japan. Copyright

2003 by the American Geophysical Union. 0094-8276/03/2003GL017261 ASC

2 - -

1 potential region of dust outbreaks when the maximum frequency is high. Thus, Figure

2 depicts the distribution of potential dust outbreak regions in East Asia. We repre- sented the analysis region in this study by a box in Figure

2 (33.5 NC52.0 N, 88.5 EC131.5 E);

this box indicates the north, east, and south boundaries of the potential dust outbreak region in East Asia. The Taklimakan Desert, west of the analysis region, was excluded from the analysis because steep mountains surround this desert. The observa- tory, which possesses dust outbreak potential, was also defined as having a maximum monthly dust outbreak frequency of over 4% (i.e., about one day for one month). Thus, a total of

105 potential dust outbreak observatories are included within the analysis region. The data at these observatories were used in the statistical analyses in sections 4.1, 4.2, and 4.3. 4. Results 4.1. Seasonal Variations [9] The white bar chart and line graph with white circles in Figure

3 indicate the frequencies of monthly dust out- breaks and strong winds from Jan.

1993 to Jun. 2002. Dust outbreaks occur most frequently in spring. This result agrees with previous studies [Littmann, 1991;

Parungo et al., 1994]. Two-thirds of the dust outbreaks can be observed in March, April, and May. In contrast, dust outbreaks are the least frequent in summer (July, August, and September). Strong winds also occur most frequently in spring and least frequently in summer. A weak secondary peak of strong wind frequency occurs around November;

however, the dust outbreak frequency has no peak in this season. 4.2. Year-to-Year Variation [10] The dust outbreak frequency in spring has been greater in the last three years (2000C2002) than in the preceding seven years (1993C1999) (Figure 4). The last three years (2000C2002) will hereafter be referred to as the dust-frequent years (DFY), and the previous seven years (1993C1999) as the dust-normal years (DNY). The year-to- year variation of Kosa events (Figure 1) is similar to that of dust outbreaks in East Asia (Figure 4);