First stages of the remediation of an area affected by a toxic spill from a pyritic mine in Aznalcóllar (Seville, Spain)

 

I. Ortíz, F. Martín, M. Simón, I. García, J. Fernández, R. Belver, E. Fernández, J. Aguilar and C. Dorronsoro.

 

 

On 25 April 1988 there was a breach in the dam of a settling pond containing tailings from a pyritic mine located in Aznalcóllar (Seville, Spain). As a result there was an important 45x10⁵ m³ strongly acidic toxic spill of tailings and water that overflowed about 40 km of the banks of the Rivers Agrio and Guadiamar, affecting 4.4 km². The water and the tailings penetrated the soils, which were covered on the surface by a variable layer of tailings, with an average thickness of 5 cm.

 

The soils were mainly contaminated by Zn, Pb, Cu, As, Cd, and TI.

 

Due to the importance of the contamination, all the estates affected were expropriated and an environmental organisation plan was drawn up dedicating the whole area to a National Park, specifically excluding any possible agricultural or livestock farming use.

 

Three months after the spill, there was an extensive clean-up operation in the area. When the tailings layer was mechanically removed, the soil surface layer was also removed, leaving the soils unevenly stripped.

 

As time passed, the tailings that had penetrated the soils oxidised, and the insoluble sulphides changed to soluble sulphates, freeing significant quantities of heavy metals and associated elements. There was a second clean-up in the most contaminated areas in an attempt to reduce this secondary contamination.

 

The soils affected by the spill were basic or neutral, but in no cases were they pH<7, and there was one important difference between them. The soils in the high part of the basin did not contain carbonates, whereas the middle and low basin soils were carbonated (with an average content of 8%). As a result, the environmental impact was far more serious in the high part of the basin, since the carbonates neutralised the acidity of the spill and to a large extent precipitated the contaminants in the rest.

 

As a corrective measure, carbonates were added to the soils in the whole basin (2 kg m^-2).

 

A sampling network was set up to monitor the contamination with a 400 m² centred square grid with 100 sampling points. The first soil samples were collected in autumn 1988, after the spill’s tailings had been removed, but without any other alterations to the soils, and the second in autumn 1999, after the second clean-up and after the area had been recarbonated by lifming. The 0-10 cm horizon was always sampled.

 

The soils in the basin were grouped into five different types with a k-means cluster analysis of the properties most related to precipitation and adsorption of the contaminating elements, such as pH, carbonates and texture. The progress of the contamination was analysed in each group and the influence of the soils’ properties was studied with a factorial analysis.

 

The analyses show that the CaCO³ content is higher in the second sample (1999) in comparison with the first (1998), a clear demonstration of the addition of carbonates. However, the increase was minor, especially in the soils that had no, or hardly any, CaCO³ (approximately 25% of the soils). On the other hand, in the soils where the CaCO³ content exceeded 6%, adding this material hardly changed the soils’ behaviour.

 

PH shows hardly any change in both samples, which confirms what has been stated in the previous paragraph; in other words, the addition of carbonates was insufficient in the acidic soils to produce a considerable increase in the pH and ensure the immobilisation of the contaminants, whereas it was not necessary to add the limestone amendment to the carbonated soils.

 

The conductivity analysis on both sampling dates shows a clear decrease in conductivity in 1999 in comparison with 1998, especially in the soils of CE>4 dS m^-1, which would be the most contaminated soils. This decrease could be the response to the additional clean-up of the most contaminated soils, as it affects the soil surface and that is where salts tend to accumulate (the climate is semiarid); or it could be a consequence of an insolubilization process of the toxic elements over time.

 

As far as heavy metals are concerned, in general, the total content of the different contaminating elements tends to decrease over time. The decrease is mainly produced in the most contaminated soils, which, in theory, should be attributed to the additional clean-up work.

 

Cd had an average value of 3.0 mg kg^-1 in the 1998 samples, while the average value for the 1999 samples dropped to 2.2 mg kg^-1, and in no cases exceeded the critical intervention level of 15 mg kg^-1 for National Park soils.

 

The Cu content was 188 mg kg^-1 in 1998, and the intervention level of 500 mg kg^-1 was only exceeded in 2% of the soils. The second sampling shows a greater contamination decrease (158 mg kg^-1), and in no cases was the critical level reached.

 

In the first sampling there was an average value of 2.7 mg kg^-1 for TI, and the intervention level was exceeded in 12% of the soils, and in the second sampling the value was 2.2 mg kg^-1, with the critical level reached in only 5% of the soils.

 

With regard to Pb, its average contents fell from 385 mg kg^-1 in 1998 to 316 mg kg^-1 in 1999. The seriously contaminated soils changed from representing 19% in 1998 to 16% in 1999.

 

The Zn content especially decreased in the soils that had more than 600 mg kg^-1 in 1998, so that in 1999 only 22% of the total of the soils in the basin exceeded 1000 mg kg^-1, while in 1998 this level was exceeded in 32% of the soils. The soils indicated average contaminations of 947 mg kg^-1 in 1998 and 722 mg kg^-1 in 1999.

 

The total As content also decreased between the first and the second samplings (the average value of the soils in 1998 was 156 mg kg^-1, and it dropped to 122 mg kg^-1 in 1999). However, the number of soils that exceeded the intervention value of 100 mg kg^-1 increased, which was 45% in 1998 and changed to 52% in 1999.

 

The concentrations of these elements were determined in water-soluble extracts and EDTA bioabsorbable extracts in order to specify their degree of hazard. In general, the EDTA soluble or extractable concentrations tend to decrease over time, which indicates that they are changing to non-bioavailable forms.