03085nam a22003257 4500002160449MiAaHDL20250726000000.0m d cr bn ---auauahd|afb|||baca900809c19871987miu|||| |b||| 00|00|eng d(MiU)990021604490106381sdr-miu.990021604490106381(MiU)MIU01000000000000002160449-goog(OCoLC)68298297(UMI)08712097(MiU)Aleph002160449UMIUMIMiUUMI diss--UMDILKS, DAVID WAYNEANALYSIS OF WATER QUALITY MODEL UNCERTAINTY USING A BAYESIAN MONTE CARLO METHOD (WISCONSIN, MICHIGAN).1987, c1987.187 p.DISSERTATION (PH.D.)--THE UNIVERSITY OF MICHIGANDissertation Abstracts International,48-02B, p. 381, AAF8712097rate studies. An interesting finding from these results is that ignoring cross-correlation between input parameters can increase model output uncertainty by almost a factor of three. Based upon these results, the Bayesian Monte Carlo technique was found to be an important addition to the method available for predicting water quality model uncertainty.The Bayesian Monte Carlo technique was applied to two different water quality models, a chloride/total phosphorus model of Green Bay, Lake Michigan, and a dissolved oxygen model of the Grand River near Grand Rapids, Michigan. The technique successfully predicted uncertainty distributions both on the model input parameters and on the model output. The Bayesian Monte Carlo technique was also used to determine the relative decrease in model uncertainty associated with improved field data and/or laboratoryThis conceptual technique described in the literature was critically reviewed and modified to allow its use for general application. The resultant technique, developed in this research, uses the statistical likelihood function to determine the ability of a model simulation to describe the observed data. Bayesian analysis is applied over n parameter dimensions by Monte Carlo sampling of input parameters from their prior distributions, and weighting those values by the likelihood function.aforementioned problems. This procedure combines Monte Carlo Analysis with Bayesian Inference to provide an uncertainty analysis technique that automatically determines the uncertainty in both inputs and results, based upon data observations.This research focused upon finding a technique that can be widely and easily applied for quantifying errors in water quality models. Four techniques were reviewed for use as a general error analysis tool: (1) Sensitivity Analysis, (2) First-Order Error Analysis, (3) Monte Carlo Analysis, and (4) Extended Kalman Filtering. Each procedure was found to have some value, but none was totally acceptable in its present form. A new method has been proposed in the literature which addresses many of theMode of access: Internet.ENVIRONMENTAL SCIENCES39015025178529198700216044919900809000000.020250726000000.020260224140649.02026-02-25T09:02:46Z2024-10-05T17:30:02ZSDR-MIUmiuALMAprepare.pl-004-009BKsdr-miu.990021604490106381MiUSDRMIU002160449mdp.39015025178529MIU1987990021604490106381MIUMIU20260225googlemdp.3901502517852919871987icbibUS bib date1 >= 1931