Xiangjiang River undertook 60% population and 70% GDP of Hunan Province. It suffered from over 60% pollutants of Hunan. It was the most seriously polluted river by heavy metals in China. Dongjiang Lake is one of the water sources of Xiangjiang River. Therefore, pollution status of Dongjiang Lake area is very important to the heavy metal treatment of Xiangjiang River.

An abandoned tailings reservoir was taken as the research object. It is located in Qingjiang town, Zixing County, Hunan Province. It was built in 1980s and closed in 2008. Due to the lake of design documents and standard regulation, parameters of this tailings site were inadequate. Based on site survey, it was found that tailings had been discharged directly on hillside. Tailing dam was constructed with waste rocks. The thickness of accumulated tailings was approximately 3 m. On the east of this tailings site, a path led to X023 county road. A river beneath the feet of tailing dam flowed into Dongjiang Lake. Minimum distance between the tailings dam and Dongjiang Lake was less than 2 kilometers, as shown in Figure 1 .

Ground penetrating radar (GPR) is a high-frequency electromagnetic technique which is used in geophysical explorations. It involves the emission of an electromagnetic wave that travels through the medium. The signal is reflected by sharp changes in the electromagnetic properties of the materials, which is recorded by the radar system. The travel time, frequencies and amplitude characters of the GPR wave are recorded.

Relationship of travel time, depth of reflection and wave velocity is shown in Equation (1).

$t=2H/v$(1)

In the formula, h represents distance, m; v represents velocity.

The most common mode of GPR operation is single-fold, fix-offset reflection profiling as illustrated in Figure 2(a) . This mode of operation gives rise to date such as shown conceptually in Figure 2(b) .

Six survey lines were arranged on reclaimed tailings site considering site conditions ( Figure 3 ). Five of those lines (from line No. 1 to No.5) were set as X direction, the other one (line No.6) was set as Y direction. Respectively, the length of survey lines from No.1 to No.6 was 85 m, 94 m, 88 m, 66 m, 48 m, and 130 m.

Reclaimed tailings site could be divided into three layers (reclaimed soil layer, tailings layer, and subsoil layer) according to ingredient. It’s known that the deepest depth of tailings layer was only about 3 m. Antennas frequency of 100 mHz were selected for the testing in this situation.

Considering the estimated depth (3 m) of tailings layer, GPR antennas of 100 mHz were selected, and time window was set as 335 ns. During the date processing, electromagnetic wave velocity in wet sand was set as reference velocity. It presented subsurface information in approximately 10 meters.

There was quite clear interface between reclaimed soil and tailings layer showed in each survey line’s result. There were little changes of thickness in reclaimed soil. Thickness ranged from 0.6 m to 1 m (seen in Figure 4 ).

There was an anomaly detected (shown in Figure 5 ) by radar near the intersection of line 1 (60 - 65 m) and line 6 (85 - 90 m). The anomaly was about 3 meters deep. It was speculated to be a geometrical shape of tubiform based on the form of reflection. It was estimated to be pipe lines installed before.

It was found that uneven distribution of a large number of scattered small reflector appeared at depths range from 0.5 m to 1.5 m in each survey line. Because tailings dam was constructed with backfill materials, they were speculated to be reflection of that structure. such as gravels, bricks and other wastes. Survey line 5 was closest to the filling dam, so it was found of the largest number of that scattered small reflections. As shown in Figure 5 , there was an inclined reflection presented in survey line 5. It could be caused by the structure of tailings dam as well.

When GPR was working near the wire, it was interfered. Significant signal disturbance appeared in the results of three survey lines. After excluding the interference of wires, there were rarely abnormal reflections received under the depth of 3 m, as shown in Figure 6 . Therefore, it was concluded that there was rarely interface of two different medium in terms of electromagnetic properties. In other words, it could be speculated that subsoil had been mixed up with tailings. Downward migration of heavy metals occurred in this tailings site.

In order to verify the results of GPR detection, two points along survey lines were selected as the sampling holes (shown in Figure 7 ). Samples were analyzed by X-ray fluorescence method to determine chemical composition. The depths and numbers of each samples were shown in Table 1 . National standard GB15618-1995 of China was set as reference to determine contamination degree in all samples.

Concentrations of four strong toxic heavy metals (As, Cd, Cr, Pb) were statistically analyzed. Analysis results indicated:

1) Total amount: The total amount of the four elements was illustrated in Figure 8 . It could be found that there were also heavy metals in surface soil, and the concentration were the lowest, but peak value of heavy metals concentration was not found in deepest samples.

2) As: Arsenic was found in both sampling holes (trivalent arsenic). Concentration of arsenic in all the samples was found exceeded class III soil standard specified in GB15618-1995. The highest concentration of As in samples was 98.5 times limit value of dry land ( Figure 9 ).

3) Cd: There was no cadmium found in four samples. The number was 1-1, 1-3, 2-1, 2-5, respectively. Cadmium concentrations of remaining samples exceeded the standard value. Highest value of cadmium concentrations was 21,000 times standard value. Cadmium in soil usually existed in the form of cadmium carbonate, it was difficult to be dissolved and migrated with water flow. Therefore, it was not evenly distributed in the soil. It was indicated that distribution of Cd was random in the contaminated soil area ( Figure 10 ).

4) Cr: Concentration of Cr in all samples was lower than level specified in national standard, Cr component was distributed in the soil randomly, Owing to the insolubility of chromium oxide. There was no Cr found in some samples (Number 1-3, 2-2, 2-3, 2-4, 2-5 showed in Figure 11 ).

5) Pb: It was found in all samples. Concentration of Pb was not substandard in surface soil samples. However, Concentration of Pb in other samples was all exceeded to standard ( Figure 12 ).

The concept of pollution degree (W) was introduced to reflect soil pollution effectively caused by heavy metals. Critical concentration of national soil quality standard is shown in Table 2 .

$W=\left(\frac{\text{Concentration of test data}}{\text{Critical concentration}}\right)-1$ (2)

On the basis of calculation results, pollution degree of soil was divided into different ranks. W < 0 (degree I): no pollution, 0 < W < 1 (degree II): light pollution, 1 < W < 2 (degree III): moderate pollution, W > 2 (degree IV): heavy pollution. After the calculation, Table 3 below was got:

Based on the analysis and calculations above, in general, this land was heavily polluted. The concentration of As, Cd and Pb in this area was much higher than normal.