Estimation of stochastic colored noise signal driving DNAPLs degradation kinetics in the natural water environment

Abstract

Little attention has been devoted to the characterization of colored noise in stochastic modeling systems. This study aims to propose an integrated numerical-simulation and statistical-inference method for the characterization of individual and mixed colored noises driving the DNAPL degradation kinetics. Through the method, the properties of colored noise (i.e. intensity and time correlation) can be identified statistically. Results for the estimations indicate that the method is useful for identifying the colored noise since the regression equations have high R square values and most of the relative errors of the estimates are below 40%. It is also found that (i) the more the repeated experiments, the higher the estimation accuracy; (ii) the larger the true value, the higher the estimation accuracy; (iii) a large number of statistical samples does not always imply high estimation accuracy. Moreover, the increase in both number of samples and interval of sampling is not helpful in enhancing the estimation accuracy. Results from the estimation of the mixed multiplicative colored-noise also show that the estimates can be regarded to be acceptable. Future studies will be needed for estimation of other types of mixed additive and multiplicative colored noises.