Miljøprojekt, 911; Teknologiudviklingsprogrammet for jord- og grundvandsforurening – Strategier for kortlægning af diffus jordforurening i byområder - Delrapport 1

Strategies for Mapping of Diffuse Soil Pollution in Urban Areas

Summary and conclusions

This report presents the investigation strategies for investigation of the diffuse soil pollution in urban areas. The project was initiated and supervised by the 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.

The report is based on a review of previous experience concerning diffuse urban pollution carried out in Phase I and on results of actual trials of the strategies on test sites in Copenhagen and in Ringsted in Phase II.

Since the objective of the project is to simplify mapping of areas with diffuse soil pollution the investigation must document with a high degree of certainty that “the area as a whole is contaminated” and that it is possible to interpolate between measured points and be certain that the soil at these points also is contaminated. This requires documentation that there is “a consistent pollution pattern and a common history of land usage” and that the area is contaminated - in other words that a majority of the measurements exceed the soil quality criteria.

Conceptual models

While traditional investigations are based on a contaminant model involving transport from a point source to the surrounding environment, diffuse soil pollution will typically comprise contributions from many random sources.

The basis for an investigation strategy for a diffuse contaminated area is the conceptual model for the actual geographical area - a model describing the mechanism by which the soil pollution may have arisen. It is important to delineate areas with a common historical development, since comparable diffuse soil pollution can be expected in these areas.

In the Phase I report summarising the present knowledge concerning sources of pollution and experience acquired by investigation of diffuse soil pollution, five pollution models were defined, which are used to identify five overall types of strategy. The five models are the deposition model, the line source model, the surface addition model, the contribution model and the soil fill model.

Physical investigation

The physical trials of the investigation strategies included three of the five conceptual pollution models: deposition model, contribution model and the line model, since the overall investigation strategy for the contribution model can also be used for the surface addition model and the soil fill model.

In connection with the physical trials a range of characteristics and typical values for contaminants and levels and distribution patterns in the soil profiles were identified. Characterisation means that an investigation strategy can be focused and optimised so that only a minimum of control measurements is required.

Furthermore, a geostatistical data treatment was applied to assess the probability that the soil quality criteria or another limit value is exceeded.

Investigation strategies

On the basis of experience with physical trials of the investigation strategies for mapping of diffuse contaminated areas , the following systematic steps were defined:

Preparation of the historical description for the area, including actual information as well experience with similar areas.
Drawing up of one or more pollution models to describe the diffuse soil pollution in the actual area.
Geographical delineation of the area to be investigated in sub-areas with a common history within which a comparable exposure to pollution can be expected according to the same conceptual model
Definition of hypotheses, which need to be clarified during the course of the investigation.
Choice of a suitable analytical parameters and measurement techniques.
Drawing up of an experimental plan

The historical description and the conceptual model
A checklist to assist in the preparation of the historical description and assess potential contaminants and spreading to the environment according to a conceptual model is presented.

Delineation of the area
It is important to delineate areas with a common history and exposure to diffuse soil pollution, since it is within these areas that a comparable level of pollution can be expected.

It should be noted that experience with the trial of investigation strategies indicated that the investigation area should have a certain size, at least 0.2 – 1 km², if a geostatistical approach is to be used.

Hypotheses
To ensure that the investigation strategy provides a sufficient basis for decisions on mapping, but at same time is goal-orientated (unnecessary work is avoided), a number of hypotheses - which can be accepted or rejected - are defined.

In the report, a number of examples are given which can be used in connection with the different pollution models. The most usual hypotheses will, however, be:

The collected data is in agreement with the conceptual model for the delineated investigation area.
The soil quality criteria is (is not) exceeded.
The soil intervention limit is (is not) exceeded.

Analytical parameters and measurement techniques
The most important contaminants found in diffuse soil contamination in urban areas in the historic fill (contribution model) and the contaminant load due to traffic (line model) are lead and benzo(a)pyrene. Additional analyses for other heavy metals (cadmium, copper, zinc) and sum of PAH contribute to a better description of the contaminant distribution across an area and confirm that the contaminant levels are identical across the area. For diffuse soil contamination due to emissions from a point source (deposition model), spreading of a contaminated media (surface model) or in soil fill (fill soil model), other contaminants can be present depending on the actual source of pollution.

As analytical method for lead, a laboratory method such as ICP-AES or AES is recommended. As analytical method for benzo(a)pyrene, a GC-MS-SIM method is recommended.

Experimental plan
In the report experimental plans for the five concept models are outlined, stating the number and density of sampling points. For a geostatistical analysis, a density of at least 200 sampling points / km² or a minimum of 40 sampling points is recommended. Likewise it is also recommended that the investigation area is at least 0.2 -1 km².

Most diffuse soil pollution (deposition model, line model, and surface model) is found in the soil surface, and contamination is in the worst case mixed down to the level of cultivation. The soil contamination is highest in the upper 0 – 30 cm and decreases in concentration with depth. For historic fill, the soil contamination is found in the upper soil layers, but may also continue to greater depth in older urban areas. Likewise, the actual depth in soil fill (fill soil model) may vary greatly.

It is recommended that most samples are taken from the soil layer under turf at 2- 10 cm’s depth, which represents the contaminant levels relevant for assessment of risks by skin contact, and a suitable number of samples are taken at greater depth in the potential contact zone according to land usage - usually 1 m or to intact soil.

It is recommended that samples be taken, if possible, from undisturbed soil. A distinction should be made between vegetable or flowerbeds where digging of the soil and addition of new soil or peat can occur, and undisturbed soil where digging does not occur., for instance turf and uncultivated soil.

It is recommended that soil samples should be taken as discrete and not as composite samples, so that samples taken close to each other can be compared to samples at greater distance, which are assumed to be less alike.

Execution and sampling

It is recommended that a thorough site inspection (walk over) is performed with a view to assessment of environmental and access conditions for sampling sites (avoidance of detrimental sites). It is also recommended that the actual sampling be documented by photography, and that observations are noted which can be of importance in the interpretation of deviant results.

Data treatment

It is recommended that the following descriptive statistics be given for results from all depth intervals:

Number of data
Average
Minimum, median and maximum
Quantiles (e.g. 10, 25, 50, 75, and 90%)

Values exceeding the soil guideline limit or the soil intervention limit can be shaded, so that it is possible to see if for example 90% of the data are over or under the soil guideline limit (JKK) or if the contaminant levels fall or rise in depth.

It is recommended that diffuse soil pollution that can be described with a deposition model, a contribution surface model or a soil fill model be treated by geostatistical techniques. The geostatistical analysis makes it possible to estimate concentrations across an area by kriging and to estimate the uncertainty of the estimate. Furthermore, the probability that the soil at a given position is above the soil quality criteria and below the soil intervention criteria can be calculated.

In connection with the geostatistical analysis, it is important to analyse if the data is correlated. Correlated data taken in positions close to each other are more alike than measurements for samples taken at positions far apart. A precondition for the analysis is that the data follow a normal distribution or log normal distribution or that these conditions are more or less fulfilled in sub-areas (regionalised).

In the report, various relevant techniques, including graphical methods to assess the data in connection with the geostatistical analysis, are presented. A flow diagram to show the different steps that can be achieved in the geostatistical analysis is defined. Not all areas are suitable for geostatistical analysis.

If no spatial correlation can be found, the probability for exceeding the soil guideline limit can be estimated for the whole area based on a distribution function for the normal distribution and the average value and standard deviation for the area.

In a few situations it is not possible to estimate the contaminant levels for the area, and investigations are required, with a greater level of detail, possibly down to sampling of each land plot.

For diffuse soil pollution, which can be described with the line model, it can be expected that the soil pollution decreases with distance at right angles to the line sources. Soil pollution is expected to decrease with depth in the soil profiles, and it is recommended that a simple graphical presentation of concentration with respect to distance from the source be used.