The purposes of this study are to observe the character of aeolian dust from the rivers of Da An, Da Jia, Da Tu and Zhou Shui, and try to find out the main factors leading to high dust concentrations in the air of the areas studied. Furthermore, a model by which the amount of aeolian dust in those areas may be calculated has been established), and a high resolution meso-scale meteorological model to enhance the accuracy of weather forecasting has also been created. Finally, receptor models were applied to chemical components of aerosols to evaluate the river dust contribution to aerosols. The section from bridge Zih-jiang to Si-bin of Zhou Shui River was selected as the study area. Image interpretation derived from SPOT imagery coupled with the aerial photos and the sites of the convex bank were used to delineate the potential areas along the riverbanks which are most susceptible to wind erosion. The amounts of PM10 emission potential can be estimated from the wind erosion depth and particle size distribution investigated at the potential areas of dust emission. The suitable vegetation sites for the potential areas of dust emission on the riverbed can be categorized into the bare sites along the main channel and the uncultivated areas at the flood plain. Rows of grass or other vegetation should be set up at the southern rims of cultivated areas along the northern bank to act as wind buffers during monsoons. For uncultivated areas at the flood plain, rows of sowed grass can be arranged at the downwind areas of the soil dust emission places for sand fixation and to form a dune for increasing the probability of natural plant invasion. With a goal of decreasing wind erosion, the dunes formed by rows of sowed grass become so-called “step-islands”, which are bases for providing seed pools for natural seeding at the potential areas of dust emission. The physical characteristics of the river sand collected from Da An, Da Jia, Da Tu and Zhou Shui Rivers were measured. The results showed that diameter of the sand at the Da Jia River was the smallest while that of the Zhou Shui River was the largest. The shape of coefficient was somewhat spherical at Da An River, but was more flat at Da Tu, Da Jia, and Zhou Shui Rivers. The threshold velocities for aeolian sand were：3.21 m/sec at Da An River, 3.0 m/sec at Da Jia River, 3.21 m/sec at Da Tu River and 3.73 m/sec at Zhou Shui River. The Zhou Shui River had the highest gradient of wind velocity, showing that the wind erosion phenomenon was most serious there. A wind tunnel experiment showed that the major motion of sand was suspension for the Zhou Shui River. This indicates that aeolian dust easily occured at the Zhou Shui River site. Arithmetic mean, multiple regressions and the minimum of conic section were used to find out the optimum cutoff points of wind speed between quadrantⅠ(high wind speed, high PM10) and Ⅱ (low wind speed, high PM10). The results showed that the wind speed threshold determined by using arithmetic mean was close to the data by using multiple regressions and conic section methods. This meant that the simplest method of arithmetic mean is acceptable to obtain the optimum cutoff points of wind speed between quadrantsⅠand Ⅱ. Data from the air quality monitoring stations showed significant positive correlation between PM10 concentration and wind speed in quadrantⅠ (high wind speed, high PM10). After the episodes of Asian dust storms during the sampling times were excluded, the remaining episodes in quadrantⅠcould be attributed to the effects of dust episodes from riverbeds. Furthermore, a significant negative correlation between PM2.5/PM10 ratio and wind speed in the data of the air quality monitoring stations was found, indicating that aeolian dust did contribute to the aerosols in the study areas. Except for NO3-, the ionic concentrations in the riverbeds were all higher than those in the street dust. However, the concentrations of metals, especially for Fe, Mn, Zn, Pb and Mo, in the street dust were all higher than those in the riverbeds. These data show that the ionic and metal concentrations could be used as fingerprints to separate dust from riverbed sources from street dust in aerosol samples. In order to compare the capability of CMB and PMF, we made some artificial samples from known source profiles. The simulation results showed that PMF can easily be used to determine sources such as tunnels, sea spray, traffic, steal mills or riverbanks. When PMF was applied on real samples collected in the study, it was found that riverbeds were the major contributing source of particular matter, especially for samples collected in Shen-Kang (Chang Hua County).

river,suspension,step-island,,suspension,particle