Importance analysis for Hudson River PCB transport and fate model parameters using robust sensitivity studies
- ETI, Seattle, WA (United States)
- Foster Wheeler Environmental, Bellevue, WA (United States)
The authors have performed robust sensitivity studies of the physico-chemical Hudson River PCB model PCHEPM to identify the parameters and process uncertainties contributing the most to uncertainty in predictions of water column and sediment PCB concentrations, over the time period 1977--1991 in one segment of the lower Hudson River. The term ``robust sensitivity studies`` refers to the use of several sensitivity analysis techniques to obtain a more accurate depiction of the relative importance of different sources of uncertainty. Local sensitivity analysis provided data on the sensitivity of PCB concentration estimates to small perturbations in nominal parameter values. Range sensitivity analysis provided information about the magnitude of prediction uncertainty associated with each input uncertainty. Rank correlation analysis indicated which parameters had the most dominant influence on model predictions. Factorial analysis identified important interactions among model parameters. Finally, term analysis looked at the aggregate influence of combinations of parameters representing physico-chemical processes. The authors scored the results of the local and range sensitivity and rank correlation analyses. The authors considered parameters that scored high on two of the three analyses to be important contributors to PCB concentration prediction uncertainty, and treated them probabilistically in simulations. They also treated probabilistically parameters identified in the factorial analysis as interacting with important parameters. The authors used the term analysis to better understand how uncertain parameters were influencing the PCB concentration predictions. The importance analysis allowed us to reduce the number of parameters to be modeled probabilistically from 16 to 5. This reduced the computational complexity of Monte Carlo simulations, and more importantly, provided a more lucid depiction of prediction uncertainty and its causes.
- OSTI ID:
- 242374
- Report Number(s):
- CONF-9511137-; ISBN 1-880611-03-1; TRN: IM9626%%202
- Resource Relation:
- Conference: 2. Society of Environmental Toxicology and Chemistry (SETAC) world conference, Vancouver (Canada), 5-9 Nov 1995; Other Information: PBD: 1995; Related Information: Is Part Of Second SETAC world congress (16. annual meeting): Abstract book. Global environmental protection: Science, politics, and common sense; PB: 378 p.
- Country of Publication:
- United States
- Language:
- English
Similar Records
Development of a probabilistic PCB-bioaccumulation model for six fish species in the Hudson River
MicroExposure Monte Carlo analysis modeling PCB exposures through fish ingestion from the Upper Hudson River