"Passive samplers to quantify micropollutants in sewer overflows:
accumulation behaviour and field validation for short pollution events"
Authors: Mutzner Lena, Vermeirssen Etiënne L.M., Mangold Simon, Maurer Max,
Scheidegger Andreas, Singer Heinz, Booij Kees, Ort Christoph
Baysian Inference for Passive Sampler Calibration Data
-----------------------------------------------------------
17.04.2019
*****************************************************************************************************************
ReadMe to run models
https://doi.org/10.25678/0000CC
Citation:
Mutzner, L., Vermeirssen, E.L.M., Mangold, S., Maurer, M., Scheidegger, A., Singer, H., Booij, K. and Ort, C. (2019)
Passive samplers to quantify micropollutants in sewer overflows: accumulation behaviour and field validation for short
pollution events. Water Research.
https://doi.org/10.1016/j.watres.2019.04.012
*****************************************************************************************************************
Data:
Calibration Experiment I and II (passive sampler mass and water concentration)
"Experiment_I_II_Concentration.Data.txt"
Software:
R
Required Folder:
Create the folder \R Functions (if not available), move all R files except R-script "Run_Bays_Parameter...R" to this
folder
Steps to run for existing data:
1. Open the file: Run_Bays_Parameter_Estimation_Mixed_Rate_Model.R
2. Change workspace (line 24) to location of folder
3. Choose to run for first-order model or mixed rate control model (line 39ff);
In case of mixed rate control model: select if to run for Experiment I AND II or soley Experiment II (line 46-48)
4. Adopt number of substances to run for in line 88 (if all substances are run, part A takes 3-4hours)
5. Check Output & Graph folder to see results
6. Part C, Line 110ff gives parameter estimates (median)
--> select C.1 for first-order and C.2 for mixed rate control model
Steps to run for new data sets:
1. Follow step 1-4 above
2. Load your data in line 69ff (compare to provided data and change your data to same format)
3. Change data files (and outliers) lines ca. 35-80 in the files: R Functions/Bayesian_Inference_MCMC.r and R Functions/Plot_Bayesian_Inference.r
4. Adopt number of substances to run for in line 88
5. Run part A --> check plots of Chains in Graphs/Check Regression and adopt cutoff for part B if necessary
6. Check Output & Graph folder to see results
7. Part C, Line 110ff gives parameter estimates (median)
Comments:
A. Mixed rate control model
------------------------------------------
Can be run for Experiment I AND II or soley for Experiment II (line 46-48)
Water concentration is assumed constant
B. First-order model
------------------------------------------
The model integrates the water concentration, thus the variability of the concentration is taken into account
The first order model systematically underestimates the initial accumulation.
The resulting parameter values predict a early equillibrium, which could
also be interpreted as increasing resistance of the sorbent. Thus, the parameter
values calculated here have to interpreted with care. Nevertheless, the provided code
could be applied to different data sets if a first order model is appropriate.