Analysis of the effect of inputs uncertainty on riverine water temperature predictions with a Markov chain Monte Carlo (MCMC) algorithm.

Water temperature is a key characteristic defining chemical, physical, and biologic conditions in riverine systems. Models of riverine water quality require many inputs, which are commonly beset by uncertainty. This study presents an uncertainty analysis of inputs to the stream-temperature simulation model HFLUX. This paper's assessment relies on a Markov chain Monte Carlo (MCMC) analysis with the DREAM algorithm, which has fast convergence rate and good accuracy. The inputs herein considered are the river width and depth, percent shade, view to sky, streamflow, and the minimum and maximum values of inputs required for uncertainty analysis. The results are presented as histograms for each input specifying the input's uncertainty. A comparison of the observational data with the DREAM algorithm estimates yielded a maximum error equal to 7.5%, which indicates excellent performance of the DREAM algorithm in ascertaining the effect of uncertainty in riverine water quality assessment.