The spatial and temporal variation of the mineral aerosol concentration and its total deposition flux over the western North Pacific region were analyzed with a regional chemical transport model (AQPMS) for the period March 1994 through May 1995. Dry deposition accounted for more than 60% of the total deposition of Asian mineral dust throughout the period. The annual deposition flux was found to decrease rapidly from the coastal area (21 g m−2 yr−1) to the open ocean (0.8 g m−2 yr−1) over the study region. Sporadic dust deposition events over the HNLC region may increase the dissolved iron concentration in seawater and hence stimulate marine biological activity. The main eastward flow of mineral dust is located in the free troposphere between 30 and 40°N, above the layer of maximum concentration. These results support the intercontinental transport of Asian mineral dust across the North Pacific as background aerosol.
 Mineral dust and gaseous and particulate pollutants from the Asian continent are transported eastward over the North Pacific, especially in spring [Uematsu et al., 1983; Jaffe et al., 1999]. These natural and anthropogenic materials in the atmosphere can influence regional and global climates by altering the Earth's radiative balance [e.g., Takemura et al., 2000]. There are still large uncertainties in estimates of the emission flux, the distribution, and the radiative forcing of Asian mineral dust. The atmospheric deposition of aerosols containing iron and other essential trace elements may cause changes in the primary productivity of phytoplanktons, food web structure and the chemical composition of the marine atmosphere over the oceans. The estimated atmospheric input of mineral matter to the North Pacific (480 × 1012 g yr−1) is much higher than that to the North Atlantic (220 × 1012 g yr−1) [Duce et al., 1991]. However, the most prominent large-scale region of elevated aerosol radiances derived from the Coastal Zone Color Scanner (CZCS) [Stegmann and Tindale, 1999] was observed off the coast of northwest Africa. According to a previously published map [Duce et al., 1991], marine biological activity is expected to be affected most greatly by atmospheric dust input in High-Nutrient Low-Chlorophyll (HNLC) regions, particularly in the northern North Pacific rather than other HNLC regions with less dust such as the equatorial Pacific and Southern Ocean. Here we attempt to estimate the atmospheric transport and input of mineral dust from the Asian continent to the western North Pacific with a regional chemical transport model based on measurements of the mineral aerosol concentration and total dust deposition on the Japanese islands.
2.1. Atmospheric Measurements
 A series of continuous 7-day high-volume aerosol samples was collected on Whatman 41 filters at Sapporo (43.1 °N, 141.3°E) and Niigata (37.9°N, 139.1°E) sites on the Japanese islands during March 1994–February 1995. Mineral aerosol concentrations were obtained by determination of Al concentration using a graphite atomic absorption spectrometer. Weekly total deposition samples were also collected at the Sapporo site during January 1994–December 1994.
 A deposition collector was placed in close proximity to the air sampling system on the roof of a building (51 m above ground level) to minimize the contribution of large soil particles of local origin to the samples. The deposition samples were filtered with an in-situ filtration system using membrane filters with 1 μm pore size, and then the dried filters were weighed to determine the insoluble particle fraction in the samples. A selected deposition sample during a dust event was analysed for particle number and size distribution by scanning electron microscope (SEM) image analysis [Uematsu et al., 2000].
2.2. Model Description
 A comprehensive Air Quality Prediction Modelling System (AQPMS) is applied to simulate the spatial and temporal variations of the mineral aerosol concentration and its total deposition flux over the western North Pacific. The model has been well validated since 1995 [Wang et al., 2000; 2002]. The local emissions of mineral dust from Japan were not taken into account in the simulation. Therefore, the simulated values show only the contribution of mineral dust from the Asian continent. A size-segregated particle dry deposition scheme and below-cloud scavenging scheme are used to treat aerosol removal by dry deposition and wet deposition [Wang et al., 2002]. In this application, the model domain ranged from 75°E to 165°E and from 5°N to 60°N with horizontal 1-degree resolution. The vertical grid consists of 18 irregular levels from the surface to an altitude of 12 km. All meteorological data sets were obtained from NCEP. The boundary and initial conditions are the same as those used by Wang et al. .
3. Results and Discussion
3.1. Measured Mineral Dust Concentration and Its Deposition Flux
 During the aerosol sampling period, the mean Al concentrations at the Sapporo site and the Niigata site were 2.4 μg m−3 and 1.5 μg m−3, respectively. These mean concentrations and the spatial and seasonal trends of Al were similar to those measured at the nearby sites of Okushiri (2.0 μg m−3) and Wajima (1.1 μg m−3) during 1981 and 1982 [Tsunogai et al., 1985]. Pronounced spring peaks of the Al concentration were observed at these sites, although there were substantial background concentrations during the summer time. In the winter, the Al concentrations were low at both Sapporo and Niigata because the local soil contribution was minimum due to snow-covered ground and there was little dust transport by westerly winds from the Asian continent where mineral dust source regions were also frozen or snow covered. Anthropogenic Al (e.g., fly ash) may contribute to the samples.
 The measured and simulated insoluble particle flux at the Sapporo site is shown in Figure 1. The correlation coefficient between the measured and simulated values was obtained to be 0.627. High total deposition fluxes of insoluble particles (>0.4 g m−2 week−1) were obtained from the beginning of April to the end of May during the period of low precipitation. Over 60% of the annual dust deposition occurred in this period and dry deposition was the dominant removal process of mineral dust particles, although large deposition fluxes associated with light shower events may also have contributed. According to SEM image analysis, almost all of the insoluble particles were alumino-silicates, with particles of 5–10 μm in diameter accounting for 65% of the total particle number in the dust event sample of 25 April to 2 May 1994.
 The annual dust deposition observed was 5.2 g m−2 yr−1 at the Sapporo site. The total deposition velocity was calculated to be 0.5 cm s−1 based on the annual deposition flux and the annual mean atmospheric dust concentration of 30 μg m−3 obtained by assuming 8% of mineral dust is Al [Uematsu et al., 1983]. The atmospheric dust deposition fluxes observed in and around the Japan Sea and the North Pacific are summarized in Table 1. The direct measurements of the annual dust fluxes in this study are in good agreement with the accumulated fluxes measured from records of fallen snow and oceanic sediments, which are slightly closer to the Asian mineral dust source regions than the Sapporo site. Sporadic dust events causing high dust concentrations and large dust depositions have been observed at several places over the Japan islands [e.g., 15 g m−2 for a two day event: Tsunogai et al., 1972], but may occur on a smaller regional scale. From long term atmospheric measurements, the mean mineral dust concentration shows a clear spatial trend, with high concentrations at sites closer to source regions in the Asian continent decreasing exponentially to sites in the central North Pacific with a halving distance of 500–600 km [Tsunogai et al., 1985].
Table 1. Atmospheric Dust Deposition Over the Western and Central North Pacific
3.2. The Simulated Total Deposition Flux of Mineral Dust
 The simulated mean total deposition flux of mineral dust originating on the Asian continent is calculated to be 2.7 g m−2 yr−1, decreasing rapidly from the coastal sea (21 g m−2 yr−1) to the open ocean (0.8 g m−2 yr−1) over the western North Pacific region (0–60°N and from the Asian coast to 165°E) shown in Figure 2 and Table 2. The geographical distribution pattern of mineral dust deposition is consistent with the annual distribution of aerosol optical thickness obtained from the satellite images [Higurashi et al., 2000]. The mineral dust flux calculated for this region is more than a factor of 2 lower than the previous estimate of 5.3 g m−2 yr−1 for the entire North Pacific (which included regions far away from the sources, i.e., the central and eastern North Pacific) [Duce et al., 1991]. The total amount of mineral dust deposited to the western North Pacific region (covering 25% of the North Pacific) was estimated to be 64 × 1012 g yr−1. This value is comparable to GCM model-based estimates [Gao et al., 2001; Ginoux et al., 2001].
Table 2. Simulated Atmospheric Flux of Mineral Dust to the Western North Pacific (March 1994–Feb. 1995)
 Our regional model suggests that more than 60% of mineral dust was removed in this oceanic region by dry deposition throughout the year. For long-range transport, the air masses containing mineral dust are typically dry and are transported a great distance from the Asian continent by strong westerly winds above the marine boundary layer [Uematsu et al., 1983]. In contrast, mineral dust events observed at ground level around the eastern coastal regions of Asia do not last long because there are greater chances of them being subjected to particle removal processes within the boundary layer, such as mixing and dilution with other air mass, interaction with marine aerosols, or collision with the earth's surface.
3.3. Horizontal Fluxes of Mineral Dust
 The frequency distributions of the backward air trajectories associated with mineral dust events have been mapped to work out the general transport patterns of mineral dust particles [Merrill et al., 1989]. Vertical profiles of mineral dust layers have been measured by lidar and profiles of back scatter and depolarization ratio sometimes indicate dense layers of mineral dust particles. The altitudinal range of mineral dust transport over Japan has been observed to be from 2 to 6 km in most cases [Iwasaka et al., 1988]. However, the main horizontal flux of mineral dust associated with the peak concentration and wind speed may not be at the same altitude as the maximum concentrations obtained by lidar or by back trajectories based on mineral dust events at ground sites.
 The mean cross section of the longitudinal transport of mineral dust from west to east across the longitude 130°E during the spring of 1994 is shown in Figure 3a. The region of highest flux (>90 μg m−2 s−1) ranged from 3 to 5.5 km in altitude is centered at 35°N, although the maximum mineral dust concentration (>18 μg m−3) is located at altitudes of 1 to 3.5 km in the same region (Figure 3b). This is because the mean wind speed increases linearly with altitude from 5 – 10 m s−1 at 1 km to 30 m s−1 above 6 km (Figure 3c).
 The main eastward flow of mineral dust in the western North Pacific is located in the mid troposphere between 30 and 40°N, above the layer with the peak concentration. The region with a horizontal mineral dust flux of 70 μg m−2 s−1 is located between ground level and an altitude of 6.5 km at 120°E, 2.5–7 km at 150°E, and 4– 5 km at 165°E. These altitudinal changes of the main flow result from the gravitational settling of mineral dust particles from the dusty air masses.
3.4. Annual Mass Balance of Mineral Dust Over the Western North Pacific Region
 The outflow of mineral dust particles originating from the Asian continent and transported from 120 to 165°E are estimated to occur between 25–55°N. The inflow of mineral dust to this domain including the coastal parts of the Asian continent is 320 × 1012 g yr−1 and the total deposition amount is 84 × 1012 g yr−1. Approximately 60% of the mineral dust inflow is transported eastward crossing longitude 165°E in the free troposphere. These results suggest the possibility of the intercontinental transport of Asian mineral dust across the North Pacific and the North American continent at higher altitudes, and also suggest the large contribution of mineral dust particles to make background aerosols in the free troposphere. It is important to determine the mineral dust inflow to this region from west of 75°E, the western boundary of the model domain.
 Transport of mineral dust toward the north (28 × 1012 g yr−1) affects the aerosol quality of the arctic region. The northward flow at 55°N is centered at altitudes of 2– 6.5 km between 145 and 155°E. The distribution of deposition flux shown in Figure 2 does not reflect this northward transport due to the weak removal processes. The southward flow at 25°N is located above 3 km between 125 and 145°E.
4. Concluding Remarks
 Numerical simulation using by AQPMS successfully reproduced the variation of the mineral aerosol concentration and its total deposition flux over Japan during the period from March 1994 to February 1995, particularly in the spring. Dry deposition is the dominant process removing the mineral dust particles over the western North Pacific region because there is little precipitation in the spring. The mean annual deposition flux of mineral dust originating from the Asian continent was calculated to be 2.7 g m−2 yr−1. The total atmospheric input of mineral dust was estimated be 64 × 1012 g yr−1 for the western North Pacific (an area of 23.5 × 1012 m2), covering 25% of the entire North Pacific Ocean.
 Several sporadic deposition events of mineral dust over the HNLC region during the spring (10–20 g m−2 period−1) will increase the dissolved iron concentration in surface seawater (e.g., 0.3–0.6 nM increase in the 50 m-deep mixed layer during an event with 10% dissolution of iron in mineral particle), and this, in turn, stimulates the marine biological activity in the northwestern North Pacific.
 The main eastward flow of mineral dust in the western North Pacific is located in the mid troposphere between 30 and 40°N and above the layer of maximum concentration. Our results support the intercontinental transport of Asian mineral dust across the North Pacific and the North American continent. Therefore, Asian mineral dust particles are probably an important source as background aerosol of the free troposphere.
 We wish to thank H. Minami, N. Komai, K. Kinoshita and D. Ueno of Hokkaido Tokai University for helping with the sampling and chemical analysis. This study was partly supported by the Core Research for Evolutional Science and Technology (CREST) of the Japan Science and Technology Corporation (JST).