Content uploaded by Michel Garcia
Author content
All content in this area was uploaded by Michel Garcia on Jun 21, 2019
Content may be subject to copyright.
Content uploaded by Michel Garcia
Author content
All content in this area was uploaded by Michel Garcia on Jun 21, 2019
Content may be subject to copyright.
Technical Day: Geostatistics applied to contaminated soils –23/01/2019 1
Using a geostatistical approach to
characterize and visualize complex
contaminated zones: Identifying the
potential origin of contaminants among
several neighboring industrial sites
Michel H. Garcia1, Bertrand Vidart2, Jean-Baptiste Mathieu1
1KIDOVA, 2AECOM
Technical Day: Geostatistics applied to contaminated soils –Paris –January 23, 2019
Technical Day: Geostatistics applied to contaminated soils –23/01/2019 2
Organic contaminants in non-aqueous phase (NAPL)
Context Methodology Feedback and conclusions
Results
Technical Day: Geostatistics applied to contaminated soils –23/01/2019 3
Organic contaminants in non-aqueous phase (NAPL)
Benzene (LNAPL)
Context Methodology Feedback and conclusions
Results
Soil contaminant
saturation
Groundwater
contaminant
concentration
// groundwater flow groundwater flow
KIDOVA –Transpol 2002
Technical Day: Geostatistics applied to contaminated soils –23/01/2019 4
Naphthalene (DNAPL)
Organic contaminants in non-aqueous phase (NAPL)
Context Methodology Feedback and conclusions
Results
Soil contaminant
saturation
Groundwater
contaminant
concentration
// groundwater flow groundwater flow
KIDOVA –Transpol 2002
Technical Day: Geostatistics applied to contaminated soils –23/01/2019 5
Organic contaminants in non-aqueous phase (NAPL)
Geostatistics as an investigation science?
Context Methodology Feedback and conclusions
Results
Smoking can seriously harm you and others around you…
Technical Day: Geostatistics applied to contaminated soils –23/01/2019 6
Context Methodology Feedback and conclusions
Results
Relationship between industrial and environmental agency about
managing soil & groundwater contamination in a production site
Long history of studies and works carried out by the industrial
•Several surveys to assess the soil and groundwater contamination by
several organic contaminants
•Pumping to get rid of and confine the groundwater contamination
•Assessment of contaminant masses still in place, carried out by AECOM
without geostatistics
Decisions with the environmental agency for next step
•Considering the contaminant masses estimated in the soils
•Considering the complexity of the spatial distribution of contaminants
•By taking into account the uncertain origin of the contamination
•Different industrials at different times
•Current presence of other industrials in the neighboring plots
Technical Day: Geostatistics applied to contaminated soils –23/01/2019 7
Expectation from the geostatistics
Better understanding of the contamination
•Better representation / visualization of the residual contamination source
shape
•In particular
•Identification of possible contamination pathways
•Identification of the potential contamination source locations
Checking the estimation of soil contaminations
•Estimations of contaminant masses in the 3 zones of interest
•Two in the plot corresponding to identified source zones
•One deep under the neighboring plot
•Distinction above and under the groundwater (saturated / unsaturated zone)
•Quantification of the uncertainty on estimated contaminant masses
Context Methodology Feedback and conclusions
Results
Technical Day: Geostatistics applied to contaminated soils –23/01/2019 8
Presentation of the case study
•Current situation: two production areas in two neighboring
plots owned by two industrials
•Multiple contaminants: BTX, VOCs, alcohols, ketones
•Available data: contaminant grade data measured in
laboratory, PID measurements on site
Context Methodology Feedback and conclusions
Results
Technical Day: Geostatistics applied to contaminated soils –23/01/2019 9
Data
Context Methodology Feedback and conclusions
Results
Plot 1
88 soil data from samples
Topography
Water table
987 PID measurements
Plot 2
Toluene (ppm) Prob(Toluene > 10 ppm)
Technical Day: Geostatistics applied to contaminated soils –23/01/2019 10
Context Methodology Feedback and conclusions
Results
Technical Day: Geostatistics applied to contaminated soils –23/01/2019 11
Smoothed 2D histogram as correlation models
Relationship between PID and grade data
Context Methodology Feedback and conclusions
Results
Cumulative distribution
of toluene for PID = 0.01
Proportions
Proportions
Cumulative distribution of
toluene for PID = 10 000
Technical Day: Geostatistics applied to contaminated soils –23/01/2019 12
Context Methodology Feedback and conclusions
Results
Geostatistical approach
Estimation by kriging of the probability of exceeding a grade
threshold
•Grade data measured in laboratory (hard data)
•Prob = 1 if grade > threshold
= 0 else
•PID data (soft data)
•Prob derived from the correlation model
Grade
Threshold
Prob
Cumulative distribution of
contaminant grade given
the PID measurement
Technical Day: Geostatistics applied to contaminated soils –23/01/2019 13
Context Methodology Feedback and conclusions
Results
Geostatistical approach
Variogram of the probability of exceeding the grade threshold
•Toluene for threshold of 10 ppm
Hard probabilities Soft probabilities
Technical Day: Geostatistics applied to contaminated soils –23/01/2019 14
Threshold on toluene
•57 grade data > 10 ppm
Proportions
Cumulative distribution
of toluene(> 10 ppm)
Context Methodology Feedback and conclusions
Results
Selection of a grade threshold
Proportions
Threshold 10 ppm
Cumulative
distribution
of toluene
Technical Day: Geostatistics applied to contaminated soils –23/01/2019 15
Threshold on toluene
•17 grade data > 100 ppm
Context Methodology Feedback and conclusions
Results
Selection of a grade threshold
Proportions
Threshold 100 ppm
Proportions
Cumulative distribution
of toluene(> 100 ppm)
Cumulative
distribution of
toluene
Technical Day: Geostatistics applied to contaminated soils –23/01/2019 16
Context Methodology Feedback and conclusions
Results
Technical Day: Geostatistics applied to contaminated soils –23/01/2019 17
Context Methodology Feedback and conclusions
Results
Estimating probability maps
Probability that contaminant grade > 10 ppm
•Estimation by kriging of the probability of exceeding the
threshold using the grade or PID data
Grade data (hard) PID measurements (soft)
Technical Day: Geostatistics applied to contaminated soils –23/01/2019 18
Context Methodology Feedback and conclusions
Results
Identifying contaminant pathways
Associated with high enough probabilities that toluene grade
> 10 ppm
•Sensitivity analysis on probability cutoff
Probability cutoff = 80% Probability cutoff = 70%
Surface at the
probability cutoff
Technical Day: Geostatistics applied to contaminated soils –23/01/2019 19
Context Methodology Feedback and conclusions
Results
Interpretations
•Possible contamination pathways
•Vertically = gravity, horizontally = water table and groundwater flow
•Potential contamination source locations
Left: toluene grade data
and probability cutoff of
80%
Right: PID data and
probability cutoff of 30%.
Technical Day: Geostatistics applied to contaminated soils –23/01/2019 20
Context Methodology Feedback and conclusions
Results
Technical Day: Geostatistics applied to contaminated soils –23/01/2019 21
Context Methodology Feedback and conclusions
Results
Achievement of the study objectives
Better understanding of the contamination
•Better representation / visualization of the residual contamination source
shape
+
•Identification of possible contamination pathways
•Identification of the potential contamination source locations
Checking the estimation of soil contaminations
•Estimations of the contaminant masses in the zones of interest
•Quantification of the uncertainty on estimated contaminant masses
Technical Day: Geostatistics applied to contaminated soils –23/01/2019 22
Context Methodology Feedback and conclusions
Results
Exploitation of the results by the industrial
Vis-à-vis the environmental agency
•Validation of the contaminant masses estimated by zone
•Validation of mass balances in the vadose and saturated zones
•Better description / 3D visualization of the residual contamination source
•Complexity of the shape
•Effectiveness of a potential approach by natural attenuation
•Confirmation of the interest in a joint approach involving all industrials
Vis-à-vis the neighboring industrial
•Better highlighting the relevance of a joint approach
Internally
•Communication on the complexity of contaminated soil management
Technical Day: Geostatistics applied to contaminated soils –23/01/2019 23
Thank you for your attention
Context Methodology Feedback and conclusions
Results