# A Nonlinear Autoregressive Modeling Approach for Forecasting Groundwater Level Fluctuation in Urban Aquifers

## Abstract

**:**

^{2}) values ranging between 0.762 and 0.994 in the validation period. Comparison with other statistical models applied to the same study area shows that NARX models presented here reduced the mean absolute error (MAE) of groundwater levels forecasts by 50%. The findings of this paper are promising and provide a valuable tool for the urban city planner to assist in controlling the problem of shallow water tables for similar climatic and aquifer systems.

## 1. Introduction

## 2. Study Area and Materials

^{2}. The land use in Kuwait City is primarily residential with commercial areas. Water resources are scarce with no surface water bodies existing in the entire country. Seawater desalination is the only domestic water supply resource available. Several desalination plants were constructed to secure the increasing water demand. Due to the rapid economic development the country has witnessed during the last few decades, the population has grown rapidly to 4.5 million placing the country under a serious water shortage threat. Most of the population is concentrated within the urbanized area of Kuwait City, representing less than 2% of the total land area for the country and its suburbs near the coast.

## 3. Methodology

#### 3.1. Non-Linear Autoregressive Networks with Exogenous Input

#### 3.1.1. Network Architecture

_{y}and n

_{x}are the delay of the target and the exogenous time series respectively, and f is a nonlinear function which is generally unknown (the black box function).

#### 3.1.2. Training Algorithms

**,**$\lambda $ is a learning parameter, $I$ is the identity matrix and $e$ represents the vector of the network error.

#### 3.2. Data Preprocessing

#### 3.3. Autocorrelation and Cross-Correlation Analysis

#### 3.4. Model Validation and Performance Assessment

^{2}), the mean absolute error (MAE), and the Nash–Sutcliffe coefficient (NASH). The coefficient of determination indicates the strength of association between the observed and simulated targets, and ranges between zero and one, where zero indicates no statistical association, and one indicates a perfect match between the observed and simulated targets. R

^{2}is defined as the square of the well-known Pearson correlation coefficient. In this case, R

^{2}does not directly measure how good the predictions are (unlike how Pearson correlation does), but rather it assesses the quality of a predictor that might be constructed from the model. The following equation calculates R

^{2}for the simulated and observed groundwater depths values:

## 4. Results and Discussion

#### 4.1. Modeling Results

#### 4.2. Comparisons with Previous Studies

^{2}values obtained from the current study for all examined wells ranged between 0.762 to 0.994 which are comparable to similar studies which used similar neural network approach (e.g., Guzman et al. [14] obtained R

^{2}values ranging between 0.83 and 0.92 when using ANNs for modeling daily groundwater levels fluctuation). In addition, the performance of statistical metrics for the current study indicate that the developed modeling approach efficiency is even comparable with other data driven approaches such as support vector machine (SVM), which usually gives more accurate results [45].

## 5. Summary and Conclusions

^{2}values are increasing significantly for three of the examined wells (NZ-1A well from 0.762 to 0.966, JB-1A well from 0.887 to 0.987, and HL-1A well from 0.765 to 0.973). Comparisons with other groundwater level fluctuation modeling techniques applied to the study site showed the superiority of the NARX-based approach (MAE was reduced by 50% in comparison with periodical models). In particular, the proposed NARX model has been found to represent the variability of groundwater levels while other periodic models have failed to do so. Unlike physically based models, the proposed NARX model is simple to use and does not require detailed site investigations.

## Funding

## Conflicts of Interest

## References

- Bear, J.; Cheng, A.H.-D. Modeling Groundwater Flow and Contaminant Transport; Springer Science & Business Media: Berlin, Germany, 2010; Volume 23, ISBN 1402066821. [Google Scholar]
- Anderson, M.P.; Woessner, W.W.; Hunt, R.J. Applied Groundwater Modeling: Simulation of Flow and Advective Transport; Academic Press: Cambridge, MA, USA, 2015; ISBN 0080916384. [Google Scholar]
- Rajaee, T.; Boroumand, A. Forecasting of chlorophyll-a concentrations in South San Francisco Bay using five different models. Appl. Ocean Res.
**2015**, 53, 208–217. [Google Scholar] [CrossRef] - McCulloch, W.S.; Pitts, W. A logical calculus of the ideas immanent in nervous activity. Bull. Math. Biophys.
**1943**, 5, 115–133. [Google Scholar] [CrossRef] - Linares-Rodriguez, A.; Lara-Fanego, V.; Pozo-Vazquez, D.; Tovar-Pescador, J. One-day-ahead streamflow forecasting using artificial neural networks and a meteorological mesoscale model. J. Hydrol. Eng.
**2015**, 20, 5015001. [Google Scholar] [CrossRef] - Liong, S.-Y.; Lim, W.-H.; Paudyal, G.N. River stage forecasting in Bangladesh: Neural network approach. J. Comput. Civ. Eng.
**2000**, 14, 1–8. [Google Scholar] [CrossRef] - Maier, H.R.; Jain, A.; Dandy, G.C.; Sudheer, K.P. Methods used for the development of neural networks for the prediction of water resource variables in river systems: Current status and future directions. Environ. Model. Softw.
**2010**, 25, 891–909. [Google Scholar] [CrossRef] - Raheli, B.; Aalami, M.T.; El-Shafie, A.; Ghorbani, M.A.; Deo, R.C. Uncertainty assessment of the multilayer perceptron (MLP) neural network model with implementation of the novel hybrid MLP-FFA method for prediction of biochemical oxygen demand and dissolved oxygen: A case study of Langat River. Environ. Earth Sci.
**2017**, 76, 503. [Google Scholar] [CrossRef] - Sudheer, K.P.; Gosain, A.K.; Ramasastri, K.S. A data-driven algorithm for constructing artificial neural network rainfall-runoff models. Hydrol. Process.
**2002**, 16, 1325–1330. [Google Scholar] [CrossRef] - Jha, M.K.; Sahoo, S. Efficacy of neural network and genetic algorithm techniques in simulating spatio-temporal fluctuations of groundwater. Hydrol. Process.
**2015**, 29, 671–691. [Google Scholar] [CrossRef] - Izady, A.; Davary, K.; Alizadeh, A.; Nia, A.M.; Ziaei, A.N.; Hasheminia, S.M. Application of NN-ARX model to predict groundwater levels in the Neishaboor Plain, Iran. Water Resour. Manag.
**2013**, 27, 4773–4794. [Google Scholar] [CrossRef] - Chang, F.-J.; Chang, L.-C.; Huang, C.-W.; Kao, I.-F. Prediction of monthly regional groundwater levels through hybrid soft-computing techniques. J. Hydrol.
**2016**, 541, 965–976. [Google Scholar] [CrossRef] - Wunsch, A.; Liesch, T.; Broda, S. Forecasting groundwater levels using nonlinear autoregressive networks with exogenous input (NARX). J. Hydrol.
**2018**, 567, 743–758. [Google Scholar] [CrossRef] - Guzman, S.M.; Paz, J.O.; Tagert, M.L.M. The use of NARX neural networks to forecast daily groundwater levels. Water Resour. Manag.
**2017**, 31, 1591–1603. [Google Scholar] [CrossRef] - Al-Sanad, H.A.; Shaqour, F.M. Geotechnical implications of subsurface water rise in Kuwait. Eng. Geol.
**1991**, 31, 59–69. [Google Scholar] [CrossRef] - Hamdan, L.; Mukhopadhyay, A. Numerical Simulation of Subsurface-Water Rise in Kuwait City a. Groundwater
**1991**, 29, 93–104. [Google Scholar] [CrossRef] - Mossaad, M.; Gutler, J.; Ghoneim, H.; Senay, Y.; El-Nahhs, F. Implementation of a geotechnical monitoring program in a pilot drainage project in Kuwait. In Proceedings of the Second International Conference on Groundwater Level in Urbanized Areas, Mansoura, Egypt, 7–8 November 1999. [Google Scholar]
- Székely, F. Numerical modeling of drainage schemes in layered formations: A Kuwait City case study. Groundwater
**1999**, 37, 879–883. [Google Scholar] [CrossRef] - Senay, Y. Water Level Fluctuation Analysis in the Residential Areas of Kuwait City during August 1990 - July 1991; Ministry of Electricity and Water: Kuwait City, Kuwait, 1991; Unpublished Report.
- Senay, Y.; Hamdan, L.; Yacoubi, A. Study of the Subsurface Water Rise in the Residential Areas of Kuwait, Vol 3: Hydrogeology; Kuwait Institute for Scientific Research Report No. KISR2227; Kuwait Institute for Scientific Research: Kuwait City, Kuwait, 1987. [Google Scholar]
- Sayed, S.A.S.; Saeedy, H.S.; Székely, F. Hydraulic parameters of a multilayered aquifer system in Kuwait City. J. Hydrol.
**1992**, 130, 49–70. [Google Scholar] [CrossRef] - Almedeij, J.; Al-Ruwaih, F. Periodic behavior of groundwater level fluctuations in residential areas. J. Hydrol.
**2006**, 328, 677–684. [Google Scholar] [CrossRef] - ASCE Task Committee on Application of Artificial Neural Networks in Hydrology. Artificial neural networks in hydrology. I: Preliminary concepts. J. Hydrol. Eng.
**2000**, 5, 115–123. [Google Scholar] [CrossRef] - Abdulkadir, S.J.; Yong, S.-P. Scaled UKF–NARX hybrid model for multi-step-ahead forecasting of chaotic time series data. Soft Comput.
**2015**, 19, 3479–3496. [Google Scholar] [CrossRef] - Lee, W.-K.; Tuan Resdi, T.A. Simultaneous hydrological prediction at multiple gauging stations using the NARX network for Kemaman catchment, Terengganu, Malaysia. Hydrol. Sci. J.
**2016**, 61, 2930–2945. [Google Scholar] [CrossRef] [Green Version] - Khaki, M.; Yusoff, I.; Islami, N. Application of the Artificial Neural Network and Neuro-fuzzy System for Assessment of Groundwater Quality. CLEAN–Soil Air Water
**2015**, 43, 551–560. [Google Scholar] [CrossRef] - Samarasinghe, S. Neural Networks for Applied Sciences and Engineering: From Fundamentals to Complex Pattern Recognition; Auerbach Publications: New York, NY, USA, 2016; ISBN 0429115784. [Google Scholar]
- Bishop, C.M. Neural Networks for Pattern Recognition; Oxford University Press: Oxford, UK, 1995; ISBN 0198538642. [Google Scholar]
- Yu, H.; Wilamowski, B.M. Levenberg-marquardt training. Ind. Electron. Handb.
**2011**, 5, 1. [Google Scholar] - Aldrich, J. Correlations genuine and spurious in Pearson and Yule. Stat. Sci.
**1995**, 10, 364–376. [Google Scholar] [CrossRef] - Granger, C.W.J.; Newbold, P. Spurious regressions in econometrics. J. Econom.
**1974**, 2, 111–120. [Google Scholar] [CrossRef] [Green Version] - Cao, G.; Zheng, C. Signals of short-term climatic periodicities detected in the groundwater of North China Plain. Hydrol. Process.
**2016**, 30, 515–533. [Google Scholar] [CrossRef] - Ghanbari, R.N.; Bravo, H.R. Coherence among climate signals, precipitation, and groundwater. Groundwater
**2011**, 49, 476–490. [Google Scholar] [CrossRef] - Bergmeir, C.; Benítez, J.M. On the use of cross-validation for time series predictor evaluation. Inf. Sci.
**2012**, 191, 192–213. [Google Scholar] [CrossRef] - Nash, J.E.; Sutcliffe, J. V River flow forecasting through conceptual models part I—A discussion of principles. J. Hydrol.
**1970**, 10, 282–290. [Google Scholar] [CrossRef] - Krause, P.; Boyle, D.P.; Bäse, F. Comparison of different efficiency criteria for hydrological model assessment. Adv. Geosci.
**2005**, 5, 89–97. [Google Scholar] [CrossRef] [Green Version] - He, Z.; Zhang, Y.; Guo, Q.; Zhao, X. Comparative study of artificial neural networks and wavelet artificial neural networks for groundwater depth data forecasting with various curve fractal dimensions. Water Resour. Manag.
**2014**, 28, 5297–5317. [Google Scholar] [CrossRef] - Nourani, V.; Mousavi, S. Spatiotemporal groundwater level modeling using hybrid artificial intelligence-meshless method. J. Hydrol.
**2016**, 536, 10–25. [Google Scholar] [CrossRef] - Yang, Q.; Hou, Z.; Wang, Y.; Zhao, Y.; Delgado, J. A comparative study of shallow groundwater level simulation with WA–ANN and ITS model in a coastal island of south China. Arab. J. Geosci.
**2015**, 8, 6583–6593. [Google Scholar] [CrossRef] - Emamgholizadeh, S.; Moslemi, K.; Karami, G. Prediction the groundwater level of bastam plain (Iran) by artificial neural network (ANN) and adaptive neuro-fuzzy inference system (ANFIS). Water Resour. Manag.
**2014**, 28, 5433–5446. [Google Scholar] [CrossRef] - Kouziokas, G.N.; Chatzigeorgiou, A.; Perakis, K. Multilayer feed forward models in groundwater level forecasting using meteorological data in public management. Water Resour. Manag.
**2018**, 32, 5041–5052. [Google Scholar] [CrossRef] - Asefa, T.; Wanakule, N.; Adams, A. Field-Scale Application of Three Types of Neural Networks to Predict Ground-Water Levels 1. JAWRA J. Am. Water Resour. Assoc.
**2007**, 43, 1245–1256. [Google Scholar] [CrossRef] - Khalil, B.; Broda, S.; Adamowski, J.; Ozga-Zielinski, B.; Donohoe, A. Short-term forecasting of groundwater levels under conditions of mine-tailings recharge using wavelet ensemble neural network models. Hydrogeol. J.
**2015**, 23, 121–141. [Google Scholar] [CrossRef] - Jeong, J.; Park, E. Comparative applications of data-driven models representing water table fluctuations. J. Hydrol.
**2019**, 572, 261–273. [Google Scholar] [CrossRef] - Gong, Y.; Zhang, Y.; Lan, S.; Wang, H. A comparative study of artificial neural networks, support vector machines and adaptive neuro fuzzy inference system for forecasting groundwater levels near Lake Okeechobee, Florida. Water Resour. Manag.
**2016**, 30, 375–391. [Google Scholar] [CrossRef] - Rajaee, T.; Ebrahimi, H.; Nourani, V. A review of the artificial intelligence methods in groundwater level modeling. J. Hydrol.
**2019**, 572, 336–351. [Google Scholar] [CrossRef]

**Figure 4.**The sequential procedure for the nonlinear autoregressive modeling approach with exogenous input, (NARX)-based approach, for modeling groundwater level fluctuation in urban aquifers.

**Figure 5.**(

**A**) Groundwater levels detrending procedure for well BN-1A. (

**B**) Autocorrelation function (correlogram) applied for groundwater levels remainder at well BN-1A. Upper and lower dashed lines designate significant correlation limits. (

**C**) Cross correlation between groundwater depth remainder and monthly average temperature at well BN-1A.

**Figure 7.**Modeled versus observed groundwater levels for all examined wells with respect to the perfect 1:1 match line.

ID | Area | Easting (m) | Northing (m) | Data Period | Ground Level Above m.s.l. (m) | Water Table Depth Range (m) |
---|---|---|---|---|---|---|

BN-1A | Bayan | 503,750.6 | 243,098.7 | 1992–2001 | 27.74 | 8.42–12.23 |

NZ-1A | Nuzha | 499,402.2 | 247,349.0 | 1992–2000 | 16.45 | 2.68–4.07 |

JB-1A | Jabriya | 502,472.2 | 244,781.0 | 1993–2002 | 23.00 | 4.52–6.08 |

HL-1A | Hawally | 500,673.1 | 247,010.4 | 1993–2002 | 20.06 | 3.76–5.60 |

Auto Correlation for Remainder Data | Cross Correlation with Temperature | |||||||
---|---|---|---|---|---|---|---|---|

Well | 1st lag | 2nd lag | 3rd lag | 4th lag | 5th lag | 6th lag | Magnitude | Lag |

BN-1A | 0.92 | 0.80 | 0.65 | 0.5 | - | - | −0.48 | 3rd lag |

NZ-1A | 0.89 | 0.75 | 0.57 | - | - | - | −0.37 | 2nd lag |

JB-1A | 0.93 | 0.85 | 0.75 | 0.65 | 0.56 | 0.49 | −0.33 | 3rd lag |

HL-1A | 0.91 | 0.77 | 0.62 | 0.48 | - | - | −0.35 | 4th lag |

**Table 3.**NARX models performance at each validation round (R1: round 1, R2: round 2, and R3: round 3).

Well | Validation Round | R^{2} | MAE (m) | NASH ^{a} | ID ^{b} | FD ^{c} | HL ^{d} |
---|---|---|---|---|---|---|---|

BN-1A | R1 | 0.971 | 0.063 | 0.964 | 1:3 | 1:4 | 30 |

R2 | 0.992 | 0.026 | 0.992 | ||||

R3 | 0.994 | 0.049 | 0.993 | ||||

NZ-1A | R1 | 0.762 | 0.072 | 0.735 | 1:2 | 1:3 | 30 |

R2 | 0.967 | 0.042 | 0.966 | ||||

R3 | 0.966 | 0.048 | 0.966 | ||||

JB-1A | R1 | 0.887 | 0.020 | 0.823 | 1:3 | 1:6 | 70 |

R2 | 0.987 | 0.010 | 0.986 | ||||

R3 | 0.987 | 0.032 | 0.985 | ||||

HL-1A | R1 | 0.765 | 0.052 | 0.643 | 1:4 | 1:4 | 70 |

R2 | 0.953 | 0.050 | 0.949 | ||||

R3 | 0.973 | 0.065 | 0.964 |

^{a}NASH = Nash-Sutcliffe coefficient;

^{b}Input delays;

^{c}Feedback delays;

^{d}Hidden layers; MAE = mean absolute error, R

^{2}= coefficient of determination.

© 2020 by the author. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## Share and Cite

**MDPI and ACS Style**

Alsumaiei, A.A.
A Nonlinear Autoregressive Modeling Approach for Forecasting Groundwater Level Fluctuation in Urban Aquifers. *Water* **2020**, *12*, 820.
https://doi.org/10.3390/w12030820

**AMA Style**

Alsumaiei AA.
A Nonlinear Autoregressive Modeling Approach for Forecasting Groundwater Level Fluctuation in Urban Aquifers. *Water*. 2020; 12(3):820.
https://doi.org/10.3390/w12030820

**Chicago/Turabian Style**

Alsumaiei, Abdullah A.
2020. "A Nonlinear Autoregressive Modeling Approach for Forecasting Groundwater Level Fluctuation in Urban Aquifers" *Water* 12, no. 3: 820.
https://doi.org/10.3390/w12030820