Estimation and comparison of gabion weir oxygen mass transfer by ensemble learnings of bagging, boosting, and stacking algorithms

1

ABSTRACT

Gabion weir comprises porous materials packed with distinct shapes and sizes of gravel. The gabion weir is eco-friendlier than an impervious weir, as its opening enables water life and sediment materials to pass through it. Aeration is how natural processes or physical structures enhance the contact area and time between water and estranged air. This process improves the dissolved oxygen (D.O.) of water. The D.O. is one of the best determinants used for water quality measurement. This paper investigates the prediction of mass oxygen transfer over the gabion weir by ensemble models. The outputs of gabion weir oxygen mass transfer were estimated using bagging, boosting, and stacking by taking input parameters such as average size, porosity, the gabion weir height, discharge per unit width, and drop height. The dataset was taken by conducting experiments. By comparing these modeling ensembles, it was found that random forest-based bagging outperformed all proposed models. Nevertheless, all applied ensemble models were performing well, but published traditional equations were performing incredibly poorly. As sensitivity analysis suggests, the discharge per unit width was the most sensitive input. An uncertainty study was also carried out.

KEYWORDS:

• Gabion weir
• oxygen mass transfer
• bagging
• boosting and stacking ensemble learnings
• Sensitivity and uncertainty study

Disclosure statement

No potential conflict of interest was reported by the authors.