Download citation:
BibTeX | RIS | EndNote | Medlars | ProCite | Reference Manager | RefWorks
Send citation to:

---

When groundwater is contaminated, removal of contaminants and the restoration of quality may be slow and sometimes, impractical. It can be harmful for human health, the ecosystem and can result in water shortage. Thus, simulation of contaminant transport can be an important task in hydro-environmental studies and consequently, it is necessary to develop the robust models which can determine the temporal forecast of pollution. For temporal modeling groundwater level and contaminant concentration (GLCC), several computational methods, namely, finite difference method, finite volume method, finite element method and boundary element method have been applied for numerical solution of governing physical-based partial differential equation (PDE). Although the physical-based numerical technique are widely used for temporal and/or spatial modeling of systems, some real-world conditions such as anisotropy and heterogeneity can have meaningful impacts on GLCC and restrict the usefulness of such methods. As a result, these method may be replaced by other techniques. In situation where there is no sufficient field data and output accuracy is preferred over perception of phenomena, a data-driven or black box model can be proper subsided. The uncertainty and complexity of the groundwater process have caused data-driven models such as artificial neural networks (ANNs) and adaptive neuro-fuzzy inference system (ANFIS) are widely used by hydrogeologists. Several studies have been performed to examine the susceptibility of artificial intelligence (AI) models for GFCT modeling. Wavelet transform coherence (WTC) is a technique for examination the localized correlation coefficient and their phase lag between non-stationary time series as a function of both time-frequency spaces. Furthermore, the cross-wavelet power is indicated as high common power of two time series and is found in time-frequency space by cross wavelet transform (XWT). Specifically, XWT investigates the regions in time-frequency space with large common power about a consistent phase relationship, and accordingly suggestion for causality between and time series. On the other hand, the WTC explores the regions in time-frequency apace in which and time series co-vary, but not essentially with high power. So, while analyzing two time series for evaluating both causality and local co-variance, the WTC is more suitable. In order to examine the applicability of the proposed AI-meshless model in real world conditions, the contaminant transport problem in Miandoab plain located in the northwest of Iran was considered as the case study. Miandoab plain, is located in a delta region of Zarrineh and Simineh Rivers. Urmia Lake in north of Miandoab plain, the largest salt-water lake in the Middle East, has been experienced climate change in early 2 decades. The wavelet transform coherence used in this study can be considered as a novel method for spatial clustering of piezometers, for detecting the interaction of aquifers in the plain and relationship between water level of the lake and GLs and CCs of piezometers located near the lake shore witch can present helpful information in GL and CC modeling. The results showed that the efficiency of ANFIS model was more than ANN model up to 30%. Reliability of ANFIS model is more than ANN model in both calibration and verification stages duo to the efficiency of fuzzy concept to overcome the uncertainties of the phenomenon.

Keywords: Groundwater, Contaminant Transport, Artificial Intelligence, Wavelet Transform

Full-Text [PDF 1438 kb]   (3233 Downloads)    

Add your comments about this article : Your username or Email:

---

---