The regional differences in the characteristics of dust events (i.e., all phenomena related with dust; dust in the production and suspension phases) in East Asia were investigated by use of surface meteorological data from March 1988 to May 2004. Referring to the topography and land cover type (e.g., Bare Desert, Semi Desert Shrubs, Grassland, Cultivated Land), East Asia was divided into 11 regions. Dust outbreaks (i.e., dust in the production phase) frequently occur in the Gobi Desert, the western part of the Loess Plateau, and the Taklimakan Desert, whose land cover types are Bare Desert and Semi Desert Shrubs (i.e., Desert). Although the result of frequent dust outbreaks is common among these regions, the frequency of floating dusts (i.e., dust in the suspension phase) differs greatly. Floating dusts are frequently observed in the Taklimakan Desert because most of the dust particles are blocked by the steep topographical surroundings such as the Tibetan Plateau, Pamirs, and Tian Shan Mountains. On the other hand, floating dusts are seldom observed in the Gobi Desert and Loess Plateau. In these regions, although dusts are produced by synoptic disturbances, they are immediately swept away by these same synoptic disturbances. We clarified the regional differences in the relationships among dust outbreak, surface wind, and land surface condition by analyzing the correlation between dust outbreak frequency and strong wind frequency at a threshold wind speed ofa constant, 6.5 m sec−1. Our analysis showed that the surface wind is primarily responsible for dust outbreaks in Desert regions in March and April. On the other hand, land surface conditions are the determining factor for dust outbreaks in Grassland regions. Land surface conditions affect dust outbreaks in May even in Desert regions, but almost no impact ofwind is seen. An examination of the effect of snow cover on the threshold wind speed, using the parameterization of Kurosaki and Mikami (2004), shows snow cover to have marked effects on dust outbreaks in Inner Mongolia and the Hexi Corridor in March, and in Northern Mongolia in April. The seasonal and year-to-year variations in dust outbreaks, and the mean threshold wind speed of dust outbreak are also compared among regions.
To observe aeolian dust in high mountainous areas, two sets of automatic weather station including visibility meters have been installed on the slopes of the Tianshan Mountains and the Kunlun Mountains. The two observation sites are at a height of about 2500 m above sea level, and those slopes face to the Taklimakan Desert. The aeolian dust climbs the slopes with anabatic wind in daytime. Horizontal mass transportation of dust over the slopes on a fine day was estimated using the data set of visibility and wind speed, and its daily and seasonal variations were analyzed. Before this analysis, visibility was converted to dust concentration via a relationship determined by visibility meter and an optical panicle counter in a laboratory. It has been recognized that the mass transportation over the slopes reflects activity of the dust outbreak at the foot areas.
Observation of number concentration of aerosol particles in the south of the Taklimakan Desert, China, was conducted using optical panicle counter (OPC). Observations were made in the edge of the small oasis Qira, which is located in the southern Tarim Basin, during the springtime in 2002, 2003, and 2004. For the comparison of regional changes, observation at gobi desert, which is 13.7 km southeast of the oasis site, was also conducted in July 2002. It was found that the number concentration of aerosol particles, especially in the coarse particles,showed significant increases after the duststorm. Averaged number concentrations of aerosol particles in Apri1 2002, March 2003 and March to April 2004 indicate that 1) a local maximum number concentration at 2 to 3 μm diameter is commonly found and 2) the coarse particle larger than 5 μm diameter is dominant in the each volume size distribution. However, annual variation of number concentrations during springtime is large. This is because the emission of dust particle of each year strongly depends on the magnitude and the number of duststorm event. The time variations of number concentration of aerosol particles at the oasis site and the gobi site showed a good coincidence with those of coarse particles larger than 1 μm diameter. It is suggested that these variations are not controlled by the local circulation induced by the oasis effect but by the diurnal circulation between the Kunlun Mountains and the Tarim Basin.
Aerosol sampling and meteorological observations were carried out from March 2001 to April 2004 at Aksu, in the northern part of the Taklimakan Desert in north-west China. There were three intensive observation periods (IOP): 11-24 April 2002, 16-26 March 2003 and 10-17 March 2004. The observed maximum concentration of total suspended particles (TSP) was 5.2 mg m−3 during the dust event of April 2003. Variations in the TSP and PM11 (panicles smaller than 11 μm) show that high atmospheric concentrations were observed during dust events that occurred mainly in spring and summer. According to the meteorological observations, dust storms or blowing sandldust occurred not only in spring, but also in summer and autumn, though floating dust occurred mainly in March and Apri1. The frequency of dust events was consisted with the results of dry deposition. During spring and summer, the size-distribution of aerosols is unimodal with a maximum in the range 3.3-7.0 μm, while in winter the size-distribution shows a bimodal pattern with two peaks; at 3.3-7.0 μm for soil particles and 0.65-1.1 μm for anthropogenic panicles. From the results of size distribution of soluble ion concentration, anthropogenic particles consist mainly of ammonium sulfate and ammonium nitrate, which result from coal combustion for heating during winter. Concerning chemical composition, aerosols from Aksu contain a significant amount of evaporate minerals such as calcium carbonate (calcite), sodium chloride (halite), and calcium sulfate (gypsum and anhydrite). These evaporates are common minerals in the saline soils which are widespread around observation site, and it is suggested that they are provided into atmosphere from the ground surface together with soil particles. Major element composition for aerosols in Aksu show a lower SiO2 content and higher FeO, K2O and MgO content compared with that for the saline soils (ψ < 125 μm), the loess deposits in the Tarim Basin and Central Loess Plateau, and Taklimakan Desert sands (ψ < 125 μm). This suggests that minerals rich in Fe, K and Mg, such as clay minerals and mica, come selectively from the soil surface into the atmosphere as mineral aerosols. This is because their particle size is generally smaller than that of quartz, the main constituent minerals in the soil.
In order to elucidate the variation and characterization of dust particles that originate in the arid and semi-arid regions of inland Asia and are transported to Japan, we established a sampling network system in east China and Japan and collected aeolian dust under the Japan-China joint project “ADEC”. Monitoring of the total suspended particle (TSP) and size-segregated dust concentrations were carried out in Beijing, Qingdao and Hefei in China, and Fukuoka, Nagoya, Tsukuba and Naha in Japan. The seasonal variations showed the general trend that the aerosol concentration was high in spring and low in summer, though it sometimes became high in winter in Qingdao. The aerosol size distribution was bimodal. One peak at around 0.5 μm diameter corresponds to panicles of anthropogenic origin, in which the main components are black carbon, NH4+ and SO42−, and some elements such as Pb, Cd, Sn, Sb and Bi are abundant. The other peak at around 4-5 μm diameter corresponds to mineral dust of soil origin, in which the main soluble components are Na+, Ca2+, NO3− and Cl−. The coarser particles increased when a dust event occurred. The IOP program was conducted in the spring of 2002 (IOP-1) and 2003 (IOP-2). The dust concentration during IOP-1 was high in China, particularly in Beijing. The average dust concentration during a dust event decreased in the order of Beijing > Qingdao > Hefei and Beijing > Fukuoka > Nagoya > Tsukuba > Naha. However, there were few observable dust events during IOP-2. From the results of continuous monitoring, the dust concentration during periods without dust events seemed to be almost equal at all monitoring locations in Japan. This suggests that the segregation of dust occurs in China and fine dust particles are transported to Japan. Seasonal variations of TSP, PM11 (paniculate size is less than 11 μm), PM2.1 (less than 2.1 μm) and mineral matter (larger than 2.1 μm) contents were studied. It was shown that these values decrease with distance from the source area.
Aerosol samples, which are dust suspended in the atmosphere, were collected in Qira, Aksu, and Shapotou in 2001 and Qira, Dunhuang and Shapotou in 2002, which are located between north-western and central China, one of the possible source areas of Asian dust. Their chemical and isotopic compositions, mainly Sr and Nd isotopic compositions, in the total suspended particles (TSP) and finer fraction (ψ < 5 μm) in the TSP were analysed. There is a slight difference in the chemical compositions of TSP and the ψ < 5 μm fraction that can probably be attributed to the relative increase of silt and/or clay in the ψ < 5 μm fraction. 87Sr/86Sr ratios were 0.71770-0.72030 in the TSP and 0.72349-0.72737 in the ψ < 5 μm fraction over all sampling sites, although εNd(0) values were relatively constant, from −11.3 to −9.3. The isotopic characteristics due to the variation of particle size are clearly seen in Rb-Sr and Sr-Nd isotopic correlation diagrams, which are usefu1 aids to identify the sources of Asian dust. In addition to the possible isotopic change with particle size, results of Sr and Nd isotopic measurements suggest that the broad area from the north-western to the central part of China, i.e. from the Taklimakan Desert to the Central Loess Plateau, is characterized as a single possible source component of Asian dust. Clarifying the correlation between geochemical features and particle size is an important requirement that may allow the correlation of long-range transported Asian dust to possible source materials.
In-situ observations and sampling of aerosol particles from 800 to 2000 m altitude, the upper mixed layer and lowermost free troposphere, were carried out over Mt. Raizan (Fukuoka, Japan) in the spring of 2003 using an optical particle counter and a cascade impactor borne in an unmanned radio-controlled airplane. In the mixed layer, mineral particles, sea salt particles, as well as the mixture of the two were dominant among the coarse particles, whereas sulfate particles dominated the fine particles. In the free troposphere, mineral particles dominated both the coarse and fine particles. The volume concentrations of particles including mineral constituents were derived using the number concentration and the composition of aerosol particles. The volume concentrations of particles including mineral constituents in the mixed layer and free troposphere were estimated to be 9.7 × 10−6−7.7 × 10−5 cm3 m−3 and 1.4 × 10−6−5.2 × 10−6 cm3 m−3, respectively. The number fractions of mineral particles mixed with sea salt of all the mineral particles in the mixed layer were much higher than those in the free troposphere. These results suggest that mineral particles mixed with sea salt were formed in the marine boundary layer. The weather record suggests that the formation process through the clouds, which has previously been considered an eflicient formation process for mixed particles, cannot explain sufficiently the observed abundance of mineral particles internally mixed with sea salt.
We observed broadband snow albedos in the visible and the near infrared spectral regions with snow pit works of several-day intervals, during the winters of 2001/2002 and 2002/2003 at Shinjo, Japan. We examined the dependence of albedos on snow grain size and on concentration of snow impurities, comparing observations and theoretical calculations using a radiative transfer model for atmosphere-snow system. The comparisons revealed that the snow was contaminated by strong absorptive impurities such as soot additional to moderate absorptive impurities such as mineral dust. Snow albedo reduction after snowfalls (snow aging effect on albedo) observed in both spectral regions corresponded to the growth tendency of snow grains and the increasing concentration of snow impurities with elapsed time after snowfalls. Measurement of the atmospheric aerosols above the snow surface using a laser optical particle counter suggested that wet deposition of atmospheric aerosols caused snow impurities of more than 1 ppmw in mass concentration.
We analyzed the time series data of dust vertical profiles, observed using polarization elastic lidar, from Apri1 8th-21st, 2002, during the first intensive observation period of the Aeolian Dust Experiment on Climate impact (ADEC), in Shapotou, Ninxia Province in China. We conducted the analysis with consideration to the upper atmospheric soundings and surface weather data. The results are summarized as follows. 1) A well-mixed dust layer dominated in the boundary layer sampled from around 11 a.m. until around 1 or 2 a.m. local time (Beljing time). Under well-mixed conditions, the mixed layers grew well with dry convections in the boundary layer, especially during the day and in the early evening. 2) Dry convections developed deeper during synoptic low-pressure conditions because of the systematic upward motions prevalent in the low-pressure systems. Vertical wind shear affected growth of the mixed layer as well, that is, weak vertical wind shear condition was favorable for the growth of the mixed layer. The deepest mixed layer was observed on Apri1 13th at a thickness of about 6-km above ground, when the pressure was the lowest, the vertical wind shear was very weak, and the atmosphere was dry. The aspect ratios (horizontal/vertical scale ratios) of the convections decreased under low-pressure and weak-shear conditions. On the other hand, the aspect ratios of the convections increased dramatically under high-pressure and strong-shear conditions, because convection growth was suppressed by systematic down-drafts of the high-pressure system and by strong vertical wind shears.
Daily and monthly aerosol optical thickness (AOT) over the ocean, derived from the Geostationary Meteorological Satellite (GMS-5) visible data, were compared with those determined from the sun photometer measurements in March and Apri1 2002. Time variations in the daily AOT derived from the GMS-5 showed good agreements with those from the sun photometer measurements. Correlation coefficient between the GMS-5 and the sun photometer derived daily AOTs was 0.85 for 61 samples taken at three sun photometer observation sites; Ryori, Yonagunijima, and Minamitorishima in Japan. The difference of monthly AOT, the GMS-5 minus the sun photometer, in March and April 2002 were −0.03 and 0.03 at Ryori, −0.01 and 0.18 at Yonagunijima, and 0.01 and 0.07 at Minamitorishima, respectively. Monthly regional AOT, defined as the median of the daily AOTs over 30° latitude × 30° longitude region, was calculated in latitude from 60°N to 60°S and in longitude from 80°E to 160°W for March and Apri1 of 2002 and 2003. In the southern hemisphere, monthly regional AOT was less than 0.1; about 0.1∼0.2 in 0°N∼30°N; larger than 0.2 in 30°N∼60°N with a maximum of 0.4∼0.5 in the western region near the Asian continent. AOTs in 2003 were generally smaller than those in 2002 in six regions between 110°E and 160°W of the northern hemisphere, where aerosol loading is most likely affected by the Asian dust. The numbers of regions, in which the AOT increased/decreased/equivalent from 2002 to 2003, were 3/8/1 in March and April, with the AOT changes from 2002 to 2003 to be as large as −0.04∼0.03.
Convective-scale transport of mineral dust in a severe weather setting is investigated with the approach of three-dimensional cloud-resolving simulations coupled with a dust emission-transport modeling. The simulations are intended to explicitly represent convective- and cloud-scale processes (such as updraft/downdraft, surface cold pool, precipitation) in a squall-line-type convective system, and are performed in an idealized setup in order to focus the primary mechanisms for convective-scale transport of dust within a squall-line system. Initialized based on an observation in a severe duststorm case over the Gobi Desert in China, the cloud model well simulates an observational feature of the squall line and the associated duststorm in spite of a simplified model setup. Dust is emitted by strong surface winds associated with a well-developed surface cold pool, and is contained and mixed within the cold pool: a high dust concentration of greater than 10 mg m−3 is induced. Owing to a high subgrid-turbulence mixing at the leading edge of the cold pool, the contained dust is transferred out of the cold pool and is entrained into the updraft region at the cold pool edge. Dust is then transported upward by the convective updraft which is continuously regenerated at the cold-pool leading edge, and spreads laterally in the cross-line directions at upper levels by system-scale circulation. Rearward dust transport relative to the leading edge of the system is pronounced at upper levels, according to the prevalent front-to-rear flow typically found in the squall-line systems. This study suggests that the representations of convective-scale transport processes should be adequately updated in order to improve the accuracy of the regional-scale to global-scale numerical predictions.
The mesoscale atmospheric environment of a dust event in the Tarim Basin, Taklimakan Desert, one of the key sources of aeolian dust in Northeast Asia, is investigated. Dust events in the Taklimakan Desert, including blowing dust and suspended dust, have long-lasting attributes, whereas in the Gobi Desert, dust events occur in phase with the passage of synoptic cyclones. Numerical simulations are conducted for a well-defined dust event on 12-15 April 2002. The Regional Climate Model of the Meteorological Research Institute (MRI-RCM) is employed. The MRI-RCM horizontal grid size is 20 km. The model reasonably well simulates time variations and spatial distributions of the surface wind field. Three types of mesoscale flow are revealed by the simulation. Sequential formation and/or coexistence of these flows can cause wind intensification and active dust emission from various areas across the basin. This gives a possible explanation for the long lasting dust events in the Tarim Basin. A closer examination shows that the development of these mesoscale circulations is closely related to the large-scale flow field behind the surface low-pressure system.
A nested regional meteorology/dust-transport model was applied to an Asian region including the Taklimakan Desert. A three-grid system with a resolution of 81 km, 27 km and 9 km was designed and model calculation was conducted over the whole period of Apri1 2001 to examine the typical springtime meteorology and some dust episodes over the Taklimakan Desert region. The performance model was examined based on meteorological observation at WMO SYNOP station surrounding the Taklimakan Desert and ADEC special observation site at Qira. The model results reproduced complicated airflows within the Tarim Basin, strong down slope winds from the Tianshan Mountains when meteorological disturbances crosses over the Taklimakan region, and a strong easterly flow from the Hexi Corridor side. Composite analysis when the Tazhong site simulated an easterly wind was conducted and showed that the percentage of the occurrence of this condition was approximately 50% of the full simulation period. The scale of the easterly wind zone is 400 km in the Y direction, 1000 km in the X direction and 2-3 km vertically. Simulated high dust concentration shows a good correlation with this easterly wind zone. The model’s averaged dust concentration fields showed that the surface concentration ratio was approximately 0.6 between the eastern and western side of the Tarim Basin. However, the modeled ratio showed an opposite relation with the SYNOP visibility, thereby indicating that the current high wind type dust emission scheme may be insuflicient to simulate the suspended dust phenomena in the western part of the Tarim Basin.
Dry deposition of Asian dust has been estimated during the intensive dust events in spring 2002 (Period I: 15-24 March 2002, Period II: 4-13 Apri1 2002) using the Asian dust aerosol model (ADAM). The emission module including log-normal distributions of particle size of different soil types of Gobi, Sand, Loess, and Mixed soil in the Asian dust source regions, has been used to calculate the amount of dry deposition. The dry deposition estimated by ADAM is compared with that calculated by using the optical particle counter (OPC) observed at Anmyeondo in Korea. The correlation between the modeled and the calculated dry deposition, is found to be 0.78 and 0.54 in Periods I and II, respectively. The dry deposition velocity depends on the particle size with a minimum value of 0.0272 cm s−1 and 0.0269 cm s−1 in the size range of 0.82-1.35 μm in diameter in Periods I and II, respectively. The averaged spectral distribution of dry deposition indicates that the larger particles are more contributed to the total dry deposition amount than the smaller particles due to the large deposition velocity oflarger particles. Consequently, the dry deposition rate is found to be higher in regions closer to the Asian dust source regions. The estimated total dry deposition amounts over the South Korea, North Korea and Japan are found to be about 558, 955, and 729 k ton, in Period I, respectively and about 240, 364, and 196 k ton, in Period II, respectively.
We have developed a dust aerosol module integrated into a global aerosol transport model, called MASINGAR (Model of Aerosol Species IN the Global AtmospheRe). The dust module treats dust-emission processes based on the saltation-bombardment theory. A global soil texture database is used to determine the size distribution of parent soils. Erodibility factors for vegetation cover, snow cover, land-use type, and soil type are considered. Dry and wet deposition processes are dependent on particle size. To obtain a climatological global dust emission, deposition, and spatial distribution, we performed a 25-year simulation which is assimilated to the 1979-2003 condition, using a nudging scheme. The simulated annual mean global emission flux and atmospheric dust load (with 0.2 ≤ D ≤ 20 μm) are 2149 (1817-2339) Tg yr−1 and 17.9 (15.6-20.4) Tg. The surface concentrations and depositions are compared with observed data, and the dust emission flux is compared with previous model studies. The model reproduces the simulated seasonal variation of dust concentration at stations near East Asian dust sources, where most previous models have underestimated the amounts and amplitudes. The global simulation results indicate that the quantitative estimation of the dust budget is sensitive to the size distribution of the emitted dust.
Spectral albedo and nadir reflectance of desert surfaces were measured with spectrometer for spectral region from 0.35 to 2.5 μm in western and central China in spring of 2001 and autumn of 2003. Desert surfaces were mainly gobi and/or sand dune around Qira in southern Taklimakan Desert, Aksu in northern Taklimakan Desert, Ejina in Badain Jaran Desert, and in Tengger Desert. The measured spectral albedos are low (0.05-0.11) at ultraviolet wavelengths but it rapidly increases to 0.2-0.3 at the wavelength about λ = 0.6 μm at any site. The wavelength at which the albedo reaches to the maximum is around λ = 1.8-2.2 μm, where the albedos range from 0.37 to 0.49 among measured sites. Nadir re flectance also basically shows a similar spectral distribution with that of the albedo, while the values are lower than albedos at any wavelength. Standard deviations of albedo and reflectance calculated from measurements for different surfaces at each site indicate that gobi surfaces around Aksu and Ejina have large variability compared with other surfaces. These datasets would be usefu1 for the studies of climate simulation, remote sensing, radiative forcing, and so on.
A sky-radiometer network was developed in order to investigate the characteristics of aeolian dust on the way of transportation from the source region to the Japan area. From the analysis of the sky-radiometer data, optical thickness in the visible to near infrared regions and size spectrum from radius 0.0l to 5 μm can be obtained. Five observation sites are set at Aksu, Qira, Shapotou, Qingdao, and Beijing in China, and four observation sites at Naha, Fukuoka, Nagoya and Tsukuba in Japan. The data of six observation sites (Qira, Shapotou, Naha, Fukuoka, Nagoya and Tsukuba) are automatically transferred to Meteorological Research Institute (MRI) through telephone lines. The following results were obtained from the data analysis of intensive observation period (IOP) in Apri1, 2002. (1) The data with Ångström exponent between 0.0 and 0.5 corresponds to the dust event day data. (2) The contribution of the particles with radius greater than 0.5 μm, which correspond to coarse dust particles, to the total optical thickness frequently exceeds more than ∼70% in the source region, and exceeds in Qingdao and sites in Japan on the dust event day. (3) The contribution of particles with radius greater than 0.5 μm to the total volume is more than ∼80% on the dust event day. (4) The retrieved volume spectrum in the source region, Aksu and Qira, is not dependent on the optical thickness. This means that the floating aerosols mainly consist of dust particles in the source region. (5) The total volume observed in Japan sites is one third of that in the source region. (6) When the size distribution for the coarse mode (r > 0.5 μm) is approximated by log-normal size distribution, effective radius reff is 2.1 to 2.3 μm in China sites and 1.6 to 1.8 μm in Japan sites. rg is about 0.7 μm at Aksu, Qira, and Shapotou and about 0.5 μm at other sites (ln rg is the center of log-normal size distribution). The width of size distribution (σg) is scattered between O.67 and O.87. These difference of size distribution among the observation sites are caused by the modification of air mass including aeolian dust; coarser dust particles are partially removed during the transportation, and the air mass including aeolian dust as a main component are partially contaminated by the aerosol into the atmosphere from anthropogenic activities. The observation network of ADEC caught this change clearly.
According to records of meteorological stations of the Japan Meteorological Agency (JMA), the occurrence of Kosa dust events increased during 2000-2002. This was verified by the analyses of aerosol optical depth at 500 nm (τ500) and Ångström exponent (α) obtained on a minutely basis by sun photometers in JMA from 1998 through 2002. The significant aerosol event (τ500 > 0.3) was classified into “a significant large particle event” (α ≤ 1.0) and “a significant small particle event” (α > 1.0). “Significant large particle event days” in Ryori were clearly more frequent from 2000 to 2002 compared with 1998 and 1999, and corresponded with the frequencies of Kosa dust events in China and Japan. In Yonagunijima, Japan, however, “Significant large particle event days” were not corresponded with the frequencies of Kosa dust events in China. Moreover, the occurrence distribution of the Ångström exponent (α), an index of particle size distribution, in Yonagunijima showed that typical Kosa events were rare, while significant small particle events occurred frequently. This indicates the dominant influence of relatively small particles on the large spring aerosol optical depth in Yonagunijima, rather than large particles such as Kosa dust. Trajectory analysis suggested that these small particles originated from urban pollution in the east coast of the Asian continent and Taiwan and/or smoke from biomass burning in Southeast Asia.
Sensitivity experiments of direct radiative forcing (RF) caused by mineral dust (MD) for the optical and physical properties of MD, surface albedo, solar zenith angle, and cloud cover were conducted using the Streamer-based Radiative Transfer Model for ADEC Sciences (SRTMAS). The atmospheric and dust profiles were simulated with a chemical transport model at four locations: the sea off Japan, the desert in Tarim Basin, the Sahara Desert, and snow in Siberia. The optimum calculation condition to estimate RF was also tested for the effect of the number of streams used in SRTMAS on the accuracy of RF by MD and the influence of all aerosols other than MD on RF by MD or all aerosols. When several types of aerosols are contained in the atmosphere, the test results showed that the RF by MD should not be calculated from the difference in net radiation between MD-only case and an aerosol-free case, but the difference in net radiation between a case containing all aerosols and a case containing all aerosols except MD. The experiment results by this method confirmed that the sensitivity of instantaneous RF in the shortwave (SW) region at the top of the atmosphere (TOA) to the refractive index strongly depends on surface albedo. Namely, the effect of the difference in the MD model on instantaneous RF is significant over high albedo surfaces and is relatively small over the sea. This is because the multiple reflection between the atmosphere (dust) and surface enhances light absorption by dust particles over high albedo surfaces. Over desert surfaces, the instantaneous RF in SW at TOA produced both positive and negative values within the possible refractive index range of MD. The diurnally averaged RF in SW at TOA also produced both positive and negative values in the possible range of desert albedo. The effect of the spectral distribution of surface albedo on RFs in SW for the TOA, surface, and atmosphere was examined for various surfaces. It was found that for small dust particles with an effective radius of less than 0.6 μm, RFs by MD changed depending on the difference in surface type even if the broadband albedo was the same. This is due to the contrast of spectral distribution between albedo and the extinction coefficient of MD. The vertical positional relationship of cloud cover to dust layer was also very important for RF at TOA in all spectral regions over desert and sea surfaces. However, the effect of cloud cover was generally small over snow surface because cloud albedo was close to the underlying snow albedo.
There are many uncertainties in the quantitative assessment of radiative effects due to atmospheric dust aerosol and the optical properties of dust particles contribute large to them. Numerical sensitivity experiments to evaluate the impacts of optical characteristics on the radiative forcing have been performed in this study. The experiments involve the effects of refractive indices, single scattering albedo, asymmetry factor and optical depth of the dust particles. An updated data set of refractive indices representing East Asian dust and the data set recommended by World Meteorology Organization (WMO) are used in our calculations for comparison. A k-distribution model for solar and thermal radiation transfer is employed in the calculation of radiative forcing. It is found that comparing with the WMO model, East Asian dust model has stronger scattering and weaker absorption at solar regime, which leads to higher negative forcing at the top of atmosphere (TOA) in this study. The more important is the signs of radiative forcing at TOA over certain regions could be reversed for the two dust models, which emphasizes the essentiality of accurate measurements of optical properties of dust aerosols for quantitatively estimating their radiative forcing.