Analysis of environmental dispersion in wetland flows with floating vegetation islands

Wang, Huilin; Cong, Peitong; Zhu, Zhengtao; Zhang, Wei; Ai, Yidan; Huai, Wenxin

doi:10.1016/j.jhydrol.2021.127359

Cited by 7 publications

(1 citation statement)

References 56 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In most developing countries, due to the lack of minimum required information about rivers, there are no comprehensive and exible methods to determine the environmental ow (EF) (Sedighkia and Datta 2023; Aazami et al 2022). Hence, managing and maintaining the ecosystem of such rivers seem very di cult (Wang et al 2022;Shang et al 2021). EF is the water regime provided for a river, wetland or coastal area in order to protect ecosystems and their bene ts (Espinoza et al 2021; Zheng et al 2020).…”

Section: Introductionmentioning

confidence: 99%

Reducing the computational cost and time of environmental flow estimation based on machine learning approaches

Haghgoo,

Amanollahi,

Kamangar

2024

Preprint

View full text Add to dashboard Cite

In recent decades, the reckless exploitation of rivers has caused significant changes in their ecosystems and upstream flow. It is imperative to understand that preservation of river ecosystems solely relies on maintaining the environmental flow (EF). Estimating the EF requires filed sampling, which are both time-consuming and costly. Thus, the purpose of this research is to estimate EF for a river and generalize its result to adjacent rivers using the modelling. To determine the EF, the physical habitat simulation (PHABSIM) model was used. Habitat suitability indexes (HSI) were created based on the filed survey for water velocity, flow depth, channel index and water temperature in a river. To predict the EF for other rivers, the linear regression model (LR) and two different types of neural network models, including Adaptive Neuro-Fuzzy Inference System (ANFIS) and multi-layer perceptron (MLP) were utilized. In this study, 80% and 20% of the data were used for training and testing phases, respectively. Among the models, in the ANFIS model, the date obtained for both training phase and testing phase were as follows respectively. R2 = 0.98, RMSE = 0.0248 and MAE = 0.0006 as well as R2 = 0.97, RMSE = 0.0295 and MAE = 0.0008. The accuracy of them were higher compared to MLP and LR models in predicting EF. Therefore, the ANFIS hybrid model can be a suitable alternative method for estimating the EF.

show abstract