Aitik, owned by Boliden Mineral AB, is a copper mine which is located near Gällivare in northern Sweden. The Aitik mine is the largest open pit mine in Sweden. The annual production of Aitik mine in 1968 was 2 million tonnes of ore and this production was increased to 34.3 million tonnes in 2012. The waste materials are deposited in the tailings dam (Figure 1). The impoundment is spread over an area of 12 km². Strong dust storms often occur when it is dry and windy. The measurement of dust generation was done by Big Spring Number Eight (BSNE) sampler.

Big Spring Number Eight (BSNE) samplers were designed by Fryrear [15] . Because it is a passive sampler and requires no power supply, it is cheap and widely used in field experiments. By far, it is the most popular passive sampler in Aeolian research [16] . Figure 2 shows a picture and the design of BSNE sampler. The sampler is constructed of 28-gauge galvanized metal, galvanized 18-mesh screen, and stainless steel 60-mesh screen. Dust-laden air enters through the opening, which has an area of 20*50 mm², and discharges through the 60- mesh screen. The air speed is slowed down inside the sampler because of enlarged space, and it allows the dust to settle in the collection pan. The 18-mesh screen helps to reduce the breakdown of deposited material and loss of very fine particles by reducing the movement of collected material. A wind vane attached at the rear allows the opening to face into wind direction.

BSNE sampler is very robust and able to collect a large amount of sediments, up to tens of grams or more [16] . The flow speed near through the opening is only about 5% lower than that of undisturbed flow, and thus can be considered approximately isokinetic [17] . Efficiency of BSNE has been studied by [15] , Stout and Fryrear [18] , Shao et al. [17] , Goossens and Offer [19] , Goossens et al. [20] , Sharrat et al. [21] , Sharrat [22] and Goossens and Buck [16] , for a wide range of wind velocity and grain size. Most of them came into the similar conclusion. The efficiency of BSNE decreases with decreasing particle size because fine grains easily follow the slight wind deflection at the sampler opening. Another reason is fine particles are prone to escape from the meshes where the inflow gets out of the sampler. For coarse grains, the efficiency is rather high. Goosseens and Offer [16] measured the overall efficiency between 86% - 96% for sand, and about 40% for 30 µm particle. Goossens and Buck [16] used efficiency of 17.5% for PM10 which is the average of Sharrat et al.’s [21] measurement between 10% and 25%. In this research, the efficiencies which used to correct the measurement are: 90% for particles > 30 µm, 40% for particles between 30 µm and 10 µm, and 17.5% for PM10. Also, researchers have noticed a slight decrease in efficiency due to increasing weed speed and this slight decrease can be neglected.

The measurement was done by exposure-profiling method and the collected dust particles were total Suspended Particulate matter (TSP). The method measures the passage of airborne particulate matter immediately downwind from the source by simultaneous, multipoint sampling of particulate concentration over the effective cross section of the dust cloud. The statistics of wind direction for the year 2001-2010 in Aitik show the wind was dominantly from west (Figure 3). Therefore, the BSNEs were installed along the downwind dam section (dam A-B in Figure 4). Due to the limited number of samplers, the measurement was done for only half length of dam A-B and the dust generation for other half dam was assumed to be the same as the measured values.

To measure the horizontal variation of dust concentration, five BSNEs (sampler a, b, c, d, e) were installed at the height of 1.5 m along dam A-B from the middle to the end dam A-B (Figure 4). The distances between BSNEs were 103 m, 115 m, 350 m, and 500 m. At the same location of the middle BSNE, two more BSNEs (sampler f and g) were installed at the height of 2.5 m and 3.5 m to measure the vertical variation of dust concentration. The collectors were faced into the tailings dam perpendicular to dam A-B, and the wind from NW-W

was required to drive the dust into the collectors. It was assumed to be none dust collection when wind direction was beyond 45 degrees to the both sides of NW-W direction. A weather station was installed as well to monitor the wind direction and speed. The measurement was done for the period of 2013-10-12 to 2013-10-29. The template is used to format your paper and style the text.

SWEEP is a process-based computer model coupled with a simple graphical user interface that simulates wind erosion for a single day storm event. The model estimates threshold friction velocity and starts the simulation of wind erosion when friction velocity exceeds the threshold. Friction velocity is calculated from log-law wind speed velocity and aerodynamic roughness. Aerodynamic roughness depends on oriented roughness, random roughness, and leaf and stem area. Threshold friction in SWEEP is defined as the velocity at which numerous aggregates begin to saltate [21] . It depends on soil aggregate, aggregate density, crust cover, loss material on crust, surface roughness, bio mass, soil moisture content and soil wilting point water content. All the input parameters are classified to field dimension, crop parameters, soil parameters, and weather factors. Since the tailings dam is a mining waste impoundment, crop parameters do not need to be considered in the modeling. Table 1 lists the input parameters for SWEEP model, and the Aitik tailings dam was assumed to be a 2*6 km² rec-

tangular field. The simulation was done for the selected dates shown in Table 1, and those days were the days with required wind directions measured by the weather station. Tailings samples were taken for particle size analysis, bulk density and moisture content measurement.

SWEEP was used to simulate the wind erosion for five dates when the recoded wind was in the required directions. It was assumed that for each day, one single wind erosion event took place. Table 2 shows the simulated material loss from the Aitik dam for the five simulated dates. Under natural conditions soil motion during wind erosion was classified into three categories, namely creeping, saltation and suspension [24] . Chepil [25] gave the division of grain size for the three categories. Creeping particles, ranging between 0.5 - 2 mm are the largest movable surface soil and rolling across the surface by wind forces. Medium sized particles of 0.1 - 0.5 mm in diameter are saltating or jumping across the surface. Finer grains smaller than 0.1mm are subjected to long time suspension in the air. In the simulated result (Table 2) SWEEP estimated material loss for saltation and creep loss, suspension loss, and PM10 loss. Saltation and creep particles are typically deposited locally, whereas suspended particles can be lifted into air and carried great distances. Suspended grains can thus be detrimental for off-site air quality and PM10, even is a small fraction of suspended particles, can, however, pose a respiratory hazard to human health in particular. The total material loss, saltation and creep loss, suspension loss, and PM10 loss were 4.4941 kg/m², 0.042 kg/m², 4.4559 kg/m², 0.264057 kg/m² respectively.

Figure 5 shows downwind discharges of material loss for 2013-10-14 and the downwind discharges for the

other days had similar patterns. The typical behavior of downwind discharge of a wind eroded field is that the suspension keeps increasing with downwind distance, even though saltation and creep reach a maximum capacity at a certain downwind point. This is because under a certain wind speed, there is a maximum capacity for saltation and creep fraction that uses up all the wind force under the wind speed. Whereas suspended particles are released by impacts of the jumping and rolling saltation and creeping grains and as long as downwind fine sized particles are not depleted completely the suspension continues increasing. As can be seen from Figure 5, saltation and creep loss reached the maximum capacity approximately at 800 m downwind the upwind edge of the tailings dam. Since suspension is released by the impacting of saltation and creeping particles, to control the dust generation wind breaks could be suggested to be installed at this distance to reduce the saltation and creep and it would in turn reduce the suspended emission. Figure 6 shows the simulated loss, wind speed and moisture content for the simulated days.

Since the moisture content was nearly constant, the dust generation was mainly affected by wind speeds and the higher the speed was the more dust generated.

The BSNEs were continuously collecting dust particle during the period 2013-10-12 to 2013-10-29. Wind was the driving force for the collectors to receive dust particles. The collectors were facing to the NW-W direction. Thus, the wind from NW-W was the required diving force for dust collection. In this test, BSNEs were assumed not to collect dust when wind was beyond 45 degrees to both sides of NW-W direction. The weather station monitored the following dates with required wind directions: 2013-10-14, 2013-10-15, 2013-10-19, 2013-10-20, and 2013-10-21. BSNEs were not collecting any dust for the other days because of unfavorable wind directions. BSNE collected the suspended particulate matter. Figure 7 shows the particle size distribution of the collected dust particles. The majority was below the size of 0.1 mm and the dust particles were slightly finer than the tailings particles. Figure 8 shows the mass of dust particles collected in BSNEs. It indicates that the dust mass was

relatively high in the middle of the dam edge and diminished with height and to the end of the edge. Therefore, approximately the cross section of the dust cloud had a height of 4m and the width close to the length of dam A-B. The total dust generated for the five days was calculated by integrating dust masses over the cross section of the dust cloud. Since the number of installed collector was limited some assumptions were needed in order to carry out the integration. The assumptions were that 1) dust masses for the other half length of the dam was the same as the dust masses for the measured half dam; 2) horizontal variations of the dust mass had the same pattern at different heights; 3) vertical variation of the dust mass had the same pattern at different distances. The integrated value was then corrected with sampler efficiency and the sub-variation of wind directions during the five days. The total suspended loss from the Aitik dam for the five days was 2300812.5 kg, which divided by the area of the tailings of the dam of 12 km² resulted in 0.475 kg/m².

The value had a big difference with simulated total suspended loss for the five days (4.4559 kg/m² in Table 2). Even though Hagen (2004) evaluated WEPS model in cropland and concluded the simulated values were in reasonable agreement with the measured ones, the model has never been evaluated for fields other than agriculture land. Many reasons contributed to the disagreement of simulated value and measured value in this study. The measurement of wind erosion of a vacant area by exposure profiling method requires enough number of collectors to cover the cross section of dust cloud. However, in this study the number of the collector was very limited and only small area of the cross section was sampled. A lot of assumptions had to be done to calculate the total suspended loss and this could result in tremendous inaccuracy. Also a considerable amount of sample might have lost since the machinery works interrupted the measurement and destroyed some of the collector during the measuring. In addition, although in the SWEEP model the input parameters are important to estimate the erosion losses, the internal empirical coefficients and equations may affect the simulation resulting in an overestimation or underestimation result. In order to use the model for mining industry more tests are needed to validate the modeling result and calibration methods could be useful in adjusting the internal coefficients and empirical equations.