Modelling and sensitivity analysis of varying roughness effect on dispersion coefficient: a laboratory study

摘要

Frequent disposal of effluent in rivers or streams in vicinity where the inhabitants rely majorly on it for domestic and agricultural purposes is dangerous, and that makes it relevant for precise modelling of dispersion coefficient or dispersion number. This will reduce under-estimation and over-estimation of this parameter as well as reduce frequent pollution assessment which have proven not to be sustainable. However, obtaining a model that will include all required parameter is still in process. This study considered the effect of varying roughness-which is the true nature of streams and rivers and its effects on the estimation of dispersion coefficient. It revealed that when varying roughness is increased by 1-unit, there will be an increase in dispersion number by 0.693 (t-statistics = 4.278; p < 0.05. In addition, an increase in 1-unit in the dispersion number value will require a decrease of both DO (t - ratio = -7.802; p < 0.05) and velocity (t - ratio = -4.992; p < 0.05) by 0.316 and 0.687 respectively. Sensitivity analysis further showed that roughness (K), dissolved oxygen (DO) are pertinent variables to be considered when dispersion co-efficient is to be modelled and have been previously left out. Furthermore, the ECM model generated, with R-2 = 0.98, prob(F-statistics) < 0.005 and Durbin-Watson t-statistics of 2.107 respectively shows statistical significance of the model. Hence, it is suggested that varying roughness or roughness and DO should be introduced in dispersion coefficient or dispersion number models to improve its accuracy.