Miljøprojekt, 914; Teknologiudviklingsprogrammet for jord- og grundvandsforurening – Diffus jordforurening og industri

Diffuse Soil Pollution and Industry

Summary and conclusions

This data report presents the data collection and results of the investigation of diffuse soil pollution around a former rolling mill on Amager with a view to determining suitable investigation strategies for mapping of diffuse soil pollution around industrial sources. Furthermore, data (lead and cadmium) from a former investigation by the County of Copenhagen of an area around an industrial point source is included in the report.

The data rapport is one of seven reports prepared in connection with a project initiated and supervised by Agency of Environmental Protection in the city of Copenhagen under the Danish EPA technology program. The overall objective is to prepare methods to optimise and simplify technical investigations by the environment authorities in connection with mapping of diffuse soil pollution at the legislative Knowledge Level 2.

Test area

The test area around the former rolling mill includes present or former industrial sites, especially along the shoreline road (Amager Strandvejen), as well as areas with reclaimed land and waste landfills The rolling mill was in operation from 1908 – 1979, and housing was established west and south of the mill in the 1920’s –1940’s.

Conceptual model

The diffuse soil pollution around the rolling mill is expected to be described by

a deposition model, whereby the diffuse soil pollution is caused by deposition from the former stacks at the mill

and

a contribution model whereby small random contributions from housing contribute to the pollution load over a period of more than 70 years.

Hypotheses

Deposition model
It is assumed that the pollutants are heavy metals such as copper, aluminium, PAH, phthalates, PCB and dioxins. It is assumed that the pollutant load in the soil surface decreases with distance from the source, and that the affected area extends to at least 500 m in all directions from the pollution source. It is assumed that the area immediately down wind from the prevailing wind direction east-northeast of the rolling mill is more affected than other areas. It is assumed that soil pollution decreases with depth. Therefore, it is assumed that the test area is more contaminated than the background levels in rural areas.

Contribution model
It is assumed that the pollutant load is caused by many small contributions over time and that the contaminant levels are higher than for uncultivated rural areas. It is assumed that contaminant parameters are heavy metals, PAH, dioxins, PCB and oil. It is also assumed that the contamination constitutes a variable and random load in the top soil and that the soil pollution decreases with depth. It is therefore assumed that the test area is more polluted than background levels in rural areas.

Pollutant loads due to traffic (the line model) along the major roads can also affect the area.

Experiment plan

To document the contaminant levels, a series of samples are taken from sampling areas randomly distributed across the test area. This is to ensure that the sampling areas are placed at different distances to each other (30 – 500 m), although a higher sampling density is maintained close to the rolling mill. In each 100 m² sampling area, soil samples are taken from 1 – 5 positions. To determine variation in depth, samples are taken at different depth intervals down to about 1m.

The objective is to determine the variation in concentration levels across the test area by use of a geostatistical analysis, where data pairs at different distances are compared. Observations close to each other are expected to be more alike than observations at greater distances. The geostatistical analysis illustrates the spatial distribution (here in two dimensions) across the test area. The concentration level can be estimated, and a tendency to decreasing or increasing concentrations in certain directions can be evaluated. Furthermore and possibly more important, this technique estimates the uncertainty of the estimate of the concentration levels.

Indicator parameters are measured in all samples, and certain samples are selected and analysed for other parameters.

Results

Historic fill is found to a depth of at least 1 m across the whole test area.

Generally, it is concluded that the contaminant pattern across the test area is well described by the contribution model, but that the deposition model can describe the area close to the rolling mill (within about 250 m). However, small areas with lower or higher levels are observed, and these are assumed to represent point sources of pollution or areas where soil is removed in connection with renovation or soil excavation. Generally, it is noted that that contamination is highest in the upper soil layer (contribution from the surface), but can penetrate in depth.

Generally, a direct effect from traffic (i.e. high levels of lead or BaP in soil samples taken close to roads) has not been demonstrated, but PAH composition in samples taken close to the road indicates anomalies compared with other samples. Generally, the soil quality criteria (JKK) are exceeded for lead, cadmium, zinc and BaP in the investigated area, and the soil intervention limits (ASK) are exceeded for BaP. Copper content is elevated, but generally less than JKK.

Only results for copper, cadmium and zinc show a tendency to decrease with distance from the rolling mill, but high and low values are also seen for adjacent samples, and this obscures the picture.

The other main contaminants - lead and BaP - show more consistent although variable concentration levels across the test area. Some smaller areas with high or low values can be delineated.

It is assessed that the documented diffuse soil pollution can be ascribed to contributions (lead, zinc, BaP) from housing (maintenance, materials, domestic heating etc.) as well as deposition (cadmium, copper, lead and zinc) from dusts and emissions from stacks and machines at the rolling mill.

In the data report for historic fill in urban areas, it is assessed that the contaminant levels depend on the age of the housing area. Comparison of concentration levels in the test area with the levels measured in other housing areasshows that lead levels are slightly elevated in comparison with areas with a comparable housing age in Copenhagen. However, BaP levels are equivalent to the housing age related levels in other areas.

A possible effect from the rolling mill can be seen most clearly for cadmium, copper and zinc in that the concentration levels are very different in comparison with other housing areas.

It is assumed that the contamination with BaP and lead is caused by contributions from housing, while the rolling mill contributes to higher loads of heavy metals, including copper, zinc and cadmium. JKK for copper is not exceeded.

Polychlorinated biphenyls (PCB) are generally not found in soil samples from areas with urban housing (PCB is not detected in 90% of samples). Low content of phthalates are measured, but levels are largely below the Danish soil quality criterion (JKK) of 250 mg/kg DW (Phthalates are not detected in 30% of the samples, and only one samples had a content of more than 1/250 of JKK). All the analysed soil samples had a low content of dioxins (1 - 20 ng international toxic equivalents (ITE)/kg DW). The pesticide content (only persistent pesticides) was analysed in 10 soil samples, but only low levels of DDT and parathion were found in four of the 10 samples. Soil quality criteria JKK (0.5 and 0.1 mg/kg DW respectively) are not exceeded.

Experience concerning the investigation strategy

Standard parameters
The contaminants of importance for diffuse soil pollution in historic fill in urban areas are lead and BaP. Supplementary analysis for other heavy metals (cadmium, copper, and zinc) and sum of PAH will contribute to a better description of the distribution of contaminants and uniformity across an area and will enable identification of a possible effect due to emissions from an industrial source.

Concentration levels and soil cover layers
The deposition of airborne contaminants presents a load on the top soil surface including the turf. It is assumed that the contamination in the upper soil layer (0 – 5 cm) will be greater than in the lower lying soil layers. However, there are problems associated with sampling of the upper growth layer (turf), and it may be desirable to take samples from the soil layer just under the turf (2- 10 cm). Likewise, it is assumed that soil in lawns is more contaminated than soil in plant or vegetable beds, where cultivation of the soil causes mixing of soil from different depths.

Generally, no significant difference is seen between samples taken in beds or in lawns or between samples taken at 0 – 5 cm’s depth or at 2- 10 cms’s depth.

It is therefore concluded that soil samples can be taken either in beds or in lawns, and that samples at 2-10 cm’s depth represent the surface soil pollution.

Geostatistics
The geostatistical analysis means that a mathematical description of how concentrations in the individual samples are related to each other as a function of distance can be determined. A map can then be drawn up with an estimate of concentration and confidence limits for every position across the area.

The probability that soil concentration level in any position exceeds the JKK or is less than the soil intervention level (ASK) can be calculated based on the derived semivariogram and the estimate for concentration and standard deviation.

The advantage of deriving a spatial relationship is that the concentration level is described as a continuum across the area, and areas with anomalous concentration levels and variance can be shown on the map. In other words, the estimate for the concentrations level and confidence interval as well as the probability that the soil at a given position across the area exceeds the soil quality criteria, is based on the actual measurements close to the position in question – i.e. the spatial correlation describes how measurements close to each other are more alike than measurements at greater distance.

Based on the experience from the investigation, the range for the correlated data is about 400 – 2000 m, and it is necessary to measure samples from at least 40 – 90 positions. It is probable that more positions need to be sampled to delineate the areas contaminated by the deposition and to reduce “noise” in atypical areas. Since the sampling density must provide both an adequate description of the area and reveal variation within small distances, a sampling density of about 200 positions per km² in the area to be mapped is recommended. A less dense sampling pattern can be used to delineate the extent of the affected area due to a point source (deposition) and along boundaries.