Monthly Groundwater Level Research Articles

Changes in groundwater drought associated with anthropogenic warming

Abstract. Here we present the first empirical evidence for changes in groundwater drought associated with anthropogenic warming in the absence of long-term changes in precipitation. Analysing standardised indices of monthly groundwater levels, precipitation and temperature, using two unique groundwater level data sets from the Chalk aquifer, UK, for the period 1891 to 2015, we show that precipitation deficits are the main control on groundwater drought formation and propagation. However, long-term changes in groundwater drought are shown to be associated with anthropogenic warming over the study period. These include increases in the frequency and intensity of individual groundwater drought months, and increases in the frequency, magnitude and intensity of episodes of groundwater drought, as well as an increasing tendency for both longer episodes of groundwater drought and for an increase in droughts of less than 1 year in duration. We also identify a transition from a coincidence of episodes of groundwater drought with precipitation droughts at the end of the 19th century, to an increasing coincidence with both precipitation droughts and with hot periods in the early 21st century. In the absence of long-term changes in precipitation deficits, we infer that the changing nature of groundwater droughts is due to changes in evapotranspiration (ET) associated with anthropogenic warming. We note that although the water tables are relatively deep at the two study sites, a thick capillary fringe of at least 30 m in the Chalk means that ET should not be limited by precipitation at either site. ET may be supported by groundwater through major episodes of groundwater drought and, hence, long-term changes in ET associated with anthropogenic warming may drive long-term changes in groundwater drought phenomena in the Chalk aquifer. Given the extent of shallow groundwater globally, anthropogenic warming may widely effect changes to groundwater drought characteristics in temperate environments.

A Novel Approach for Prediction of Monthly Ground Water Level Using a Hybrid Wavelet and Non-Tuned Self-Adaptive Machine Learning Model

In recent decades, due to groundwater withdrawal in the Kabodarahang region, Iran, Hamadan, hazardous events such as sinkholes, droughts, water scarcity, etc., have occurred. This study models groundwater level (GWL) of the Kabodarahang region using two novel techniques including Self-Adaptive Extreme Learning Machine (SAELM) and Wavelet-Self-Adaptive Extreme Learning Machine (WA-SAELM). Using the stepwise selection as different lags along with different input combinations, ten different SAELM and WA-SAELM models were developed. First, the best activation function is chosen for numerical models. After that, GWL values were normalized to equalize the values and enhance speed and accuracy of modeling. Then, an optimized mother wavelet is selected in order to simulate GWLs. Next, the best model was introduced as the superior model in which values of the correlation coefficient (R), Root Mean Squared Error (RMSE) and Nash-Sutcliffe efficiency coefficient (NSC) were obtained 0.969, 0.358 and 0.939, respectively. In addition, the results of the superior model are compared with classical neural network models such as Artificial Neural Network (ANN), Wavelet-Artificial Neural Network (WA-ANN), Support Vector Machine (SVM) and Wavelet-Support Vector Machine (WA-SVM). Among all models, WA-SAELM approximated GWLs with higher accuracy. Furthermore, based on the results obtained from an uncertainty analysis, the superior model was identified as a model with an underestimated performance. Additionally, an explicit and practical matrix was proposed for computing GWLs. Finally, the matrix was validated for another piezometer.

Markov chain Monte Carlo based on adaptive Metropolis algorithm applied in combined model to deal with the uncertainty of weights of single models

Abstract Reliability and validity of model prediction play a decisive role in water resource simulation and prediction. Among many prediction models, the combined model (CM) is widely used because it can combine the prediction results of multiple single models and make full use of the information provided by various methods. CM is an effective method to improve the predictive veracity but the weight of single model estimation is the key to the CM. Previous studies take errors as the objective function to calculate the weight, and the uncertainty of the weight of the individual model cannot be considered comprehensively. In order to consider the uncertainty of the weight and to improve universal applicability of the CM, in this paper, the authors intend the Markov chain Monte Carlo based on adaptive Metropolis algorithm (AM-MCMC) to solve the weight of a single model in the CM, and obtain the probability distribution of the weight and the joint probability density of all the weight. Finally, the optimal weight combination is obtained. In order to test the validity of the established model, the author put it into the prediction of monthly groundwater level. The two single models in the CM are time series analysis model (TSAM) and grey model (GM (1,1)), respectively. The case study showed that the uncertainty characteristic of the weight in the CM can be obtained by AM-MCMC. According to the study results, CM has obtained a least average root mean square error (RMSE) of 0.85, a mean absolute percentage error (MAPE) of 8.61, and a coefficient of determination (R2) value of 0.97 for the studied forecast period.

Trend Analysis of Groundwater Levels and Assessment of Regional Groundwater Drought: Ghataprabha River Basin, India

Groundwater drought is a relatively new concept, particularly in the Indian subcontinent, where groundwater levels are declining rapidly. The present study focuses on understanding the trends in groundwater levels and evaluates regional groundwater drought characteristics in the drought-prone Ghataprabha river basin, India. Cluster analysis was performed on the long-term monthly groundwater levels to classify the wells, and the Mann–Kendall test was accomplished to investigate the annual and seasonal groundwater-level trends. Standardized Groundwater level Index (SGI) was used to evaluate groundwater drought. Significant decreasing trends were observed in more than 61% of the wells in the study area with average decline of 0.21 m. Results of the SGI analysis showed that the wells of clusters 1 and 2 experienced recurrent droughts, which can be attributed to diminishing rainfall and over-exploitation of groundwater resources. The outcome of this study provides valuable information about the long-term behavior of regional groundwater levels which, in turn, helps to establish an operative groundwater management strategy for upcoming droughts.

The Influence of Climate Variability Effects on Groundwater Time Series in the Lower Central Plains of Thailand

This research studies the relationship between the climate index and the groundwater level of the lower Chao Phraya basin, in order to forecast the groundwater level in the studied area by using Autoregressive Integrated Moving Average with Explanatory (ARIMAX). The combination of 6 climate indices—Dipole Mode Index, Indian Summer Monsoon Index, Multivariate ENSO Index, Sea Surface Temperature NINO4, Southern Oscillation Index and the Western North Pacific Monsoon Index—were used, along with the groundwater level data from 14 stations during the period 1980–2011 to develop the forecast model and verify it with the data of 2012.The first step was correlation of the ARIMA model with Autocorrelation Function and Partial Autocorrelation Function. The possible model was then selected using BIC statistics. Diagnostic Checking was done to consider the white noise characteristic of estimated residuals by using the statistics of Box and Ljung (Q-statistic). If the selected models were found to be proper, then the Granger Causality Test of the leading parameters or the climate index would be performed as the next step. The results show that there is a relationship between the groundwater level and the climate index. The model could be used to forecast effectively the average RMSE value at 0.6. The last step was to develop the MODFLOW for a conceptual model and synthesize groundwater levels in the study area, which covers around 43,000 km2 and has 8 layers of groundwater, with Bangkok clay on the top. All other boundary values were set to be steady. The calibration was done using the data of 325 observed wells. The normalized RMS value was 9.705%. The results were verified by the data using ARIMAX over the same time periods. To conclude, the simulated results of the monthly groundwater level in 2012 of the wells have a confidence interval of around 95%, which is near the result from the ARIMAX model. The advantages of the ARIMAX model include high accuracy, no requirement for a large amount of data and inexpensive implementation. It is one of the effective tools for the groundwater prediction.

Economic impacts of urban flooding in South Florida: Potential consequences of managing groundwater to prevent salt water intrusion

High-value urban zones in coastal South Florida are considered particularly vulnerable to salt water intrusion into the groundwater-based, public water supplies caused by sea level rise (SLR) in combination with the low topography, existing high water table, and permeable karst substrate. Managers in the region closely regulate water depths in the extensive South Florida canal network to control closely coupled groundwater levels and thereby reduce the risk of saltwater intrusion into the karst aquifer. Potential SLR adaptation strategies developed by local managers suggest canal and groundwater levels may have to be increased over time to prevent the increased salt water intrusion risk to groundwater resources. However, higher canal and groundwater levels cause the loss of unsaturated zone storage and lead to an increased risk of inland flooding when the recharge from rainfall exceeds the capacity of the unsaturated zone to absorb it and the water table reaches the surface. Consequently, higher canal and groundwater levels are also associated with increased risk of economic losses, especially during the annual wet seasons. To help water managers and urban planners in this region better understand this trade-off, this study models the relationships between flood insurance claims and groundwater levels in Miami-Dade County. Via regression analyses, we relate the incurred number of monthly flood claims in 16 Miami-Dade County watersheds to monthly groundwater levels over the period from 1996 to 2010. We utilize these estimated statistical relationships to further illustrate various monthly flood loss scenarios that could plausibly result, thereby providing an economic quantification of a “too much water” trade-off. Importantly, this understanding is the first of its kind in South Florida and is exceedingly useful for regional-scale hydro-economic optimization models analyzing trade-offs associated with high water levels.

Prediction of Groundwater Level in Ardebil Plain Using Support Vector Regression and M5 Tree Model.

The Ardebil plain, which is located in northwest Iran, has been faced with a recent and severe decline in groundwater level caused by a decrease of precipitation, successive long-term droughts, and overexploitation of groundwater for irrigating the farmlands. Predictions of groundwater levels can help planners to deal with persistent water deficiencies. In this study, the support vector regression (SVR) and M5 decision tree models were used to predict the groundwater level in Ardebil plain. The monthly groundwater level data from 24 piezometers for a 17-year period (1997 to 2013) were used for training and test of models. The model inputs included the groundwater levels of previous months, the volume of entering precipitation into every cell, and the discharge of wells. The model output was the groundwater level in the current month. In order to evaluate the performance of models, the correlation coefficient (R) and the root-mean-square error criteria were used. The results indicated that both SVR and M5 decision tree models performed well for the prediction of groundwater level in the Ardebil plain. However, the results obtained from the M5 decision tree model are more straightforward, more easily applied, and simpler to interpret than those from the SVR. The highest accuracy was obtained using the SVR model to predict the groundwater level from the Ghareh Hasanloo and Khalifeloo piezometers with R = 0.996 and R = 0.983, respectively.

Spatiotemporal geostatistical modelling of groundwater level variations at basin scale: a case study at Crete's Mires Basin

Abstract Spatiotemporal geostatistical analysis of groundwater levels is a significant tool for groundwater resources management. This work presents a valid spatiotemporal geostatistical model for the groundwater level variations of an aquifer in Crete, Greece. The goal of this approach is to accurately map the aquifer level at variable time-steps using joint space–time information. The proposed model applies the space–time ordinary kriging (STOK) methodology using joint space–time covariance functions. A space–time experimental variogram is determined from the monthly groundwater level time-series between the hydrological years 2009 and 2014 at 11 sampling stations. The experimental spatiotemporal variogram is successfully fitted by the product–sum model using a Matérn spatial and temporal function. STOK was used to predict the monthly groundwater level at each sampling station from January to May 2015. Validation results show low prediction errors that range from 0.95 to 1.45 m, while the kriging variance accurately determines the variability of predictions. Maps of groundwater level predictions and uncertainty are developed for significant months of the validation period to assess the aquifer spatiotemporal variability. This work demonstrates that space–time geostatistics can successfully model the spatial dynamic behaviour of an aquifer when the space–time dependencies are appropriately modelled, even for a sparse dataset.

Assessing the suitability of extreme learning machines (ELM) for groundwater level prediction

Abstract Fluctuation of groundwater levels around the world is an important theme in hydrological research. Rising water demand, faulty irrigation practices, mismanagement of soil and uncontrolled exploitation of aquifers are some of the reasons why groundwater levels are fluctuating. In order to effectively manage groundwater resources, it is important to have accurate readings and forecasts of groundwater levels. Due to the uncertain and complex nature of groundwater systems, the development of soft computing techniques (data-driven models) in the field of hydrology has significant potential. This study employs two soft computing techniques, namely, extreme learning machine (ELM) and support vector machine (SVM) to forecast groundwater levels at two observation wells located in Canada. A monthly data set of eight years from 2006 to 2014 consisting of both hydrological and meteorological parameters (rainfall, temperature, evapotranspiration and groundwater level) was used for the comparative study of the models. These variables were used in various combinations for univariate and multivariate analysis of the models. The study demonstrates that the proposed ELM model has better forecasting ability compared to the SVM model for monthly groundwater level forecasting.

Examining the Relationship between Drought Indices and Groundwater Levels

Thorough characterization of the response of finite water resources to climatic factors is essential for water monitoring and management. In this study, groundwater level data from U.S. Geological Survey Ground-Water Climate Response Network wells were used to analyze the relationship between selected drought indices and groundwater level fluctuation. The drought episodes included in this study were selected using climate division level drought indices. Indices included the Palmer Drought Severity Index, Palmer Hydrological Drought Index, and Standardized Precipitation Index (SPI-6, 9, 12, 24). Precipitation and the average temperature were also used. SPI-24 was found to correlate best with groundwater levels during drought. For 17 out of 32 wells, SPI-24 showed the best correlation amongst all of the indices. For 12 out of 32 wells, SPI-24 showed correlation coefficients of −0.6 or stronger; and for other wells, reasonably good correlation was demonstrated. The statistical significance of SPI-24 in predicting groundwater level was also tested. The correlation of average monthly groundwater levels with SPI-24 does not change much throughout the timeframe, for all of the studied wells. The duration of drought also had a significant correlation with the decline of groundwater levels. This study illustrates how drought indices can be used for a rapid assessment of drought impact on groundwater level.

Simulation of groundwater level variations using wavelet combined with neural network, linear regression and support vector machine

Simulation of groundwater level (GWL) fluctuations is an important task in management of groundwater resources. In this study, the effect of wavelet analysis on the training of the artificial neural network (ANN), multi linear regression (MLR) and support vector regression (SVR) approaches was investigated, and the ANN, MLR and SVR along with the wavelet-ANN (WNN), wavelet-MLR (WLR) and wavelet-SVR (WSVR) models were compared in simulating one-month-ahead of GWL. The only variable used to develop the models was the monthly GWL data recorded over a period of 11years from two wells in the Qom plain, Iran. The results showed that decomposing GWL time series into several sub-time series, extremely improved the training of the models. For both wells 1 and 2, the Meyer and Db5 wavelets produced better results compared to the other wavelets; which indicated wavelet types had similar behavior in similar case studies. The optimal number of delays was 6months, which seems to be due to natural phenomena. The best WNN model, using Meyer mother wavelet with two decomposition levels, simulated one-month-ahead with RMSE values being equal to 0.069m and 0.154m for wells 1 and 2, respectively. The RMSE values for the WLR model were 0.058m and 0.111m, and for WSVR model were 0.136m and 0.060m for wells 1 and 2, respectively.

Wavelet analysis–artificial neural network conjunction models for multi-scale monthly groundwater level predicting in an arid inland river basin, northwestern China

Abstract In this study, the ability of a wavelet analysis–artificial neural network (WA-ANN) conjunction model for multi-scale monthly groundwater level forecasting was evaluated in an arid inland river basin, northwestern China. The WA-ANN models were obtained by combining discrete wavelet transformation with ANN. For WA-ANN model, three different input combinations were trialed in order to optimize the model performance: (1) ancient groundwater level only, (2) ancient climatic data, and (3) ancient groundwater level combined with climatic data to forecast the groundwater level for two wells in Zhangye basin. Based on the key statistical measures, the performance of the WA-ANN model was significantly better than ANN model. However, WA-ANN model with ancient groundwater level as its input yielded the best performance for 1-month groundwater forecasts. For 2- and 3-monthly forecasts, the performance of the WA-ANN model with integrated ancient groundwater level and climatic data as inputs was the most superior. Notwithstanding this, the WA-ANN model with only ancient climatic data as its inputs also exhibited accurate results for 1-, 2-, and 3-month groundwater forecasting. It is ascertained that the WA-ANN model is a useful tool for simulation of multi-scale groundwater forecasting in the current study region.

Spatiotemporal Characteristics of Groundwater Drought and Its Response to Meteorological Drought in Jiangsu Province, China

In this study, the temporal and spatial variations of groundwater drought using a Standardized Groundwater Level Index (SGI) were analyzed based on 40 monthly groundwater level observation wells from 1989 to 2012 in Jiangsu Province, China. Meteorological drought, calculated by the Standardized Precipitation Index (SPI), was also included to reveal its propagation and impact on the groundwater drought process. Results showed that the southern region of Jiangsu faced more frequent groundwater droughts and lower intensity, while the northern region faced less frequent groundwater drought with higher intensity. Furthermore, the cross-correlation between the spatial average of SGI and SPI for SPI accumulation periods of q = 1 to 12 was computed. The relationship between SGI and SPI varied in different regions. Detailed analysis of the characteristics of groundwater and meteorological drought for each region showed that meteorological droughts happened more frequently than groundwater drought in Jiangsu Province during the study period, while the mean duration and mean magnitude of groundwater droughts were longer and larger than those of meteorological droughts. It is expected that this study will provide useful information for drought monitoring and mitigation in Jiangsu and similar areas.

Application of RBFN network and GM (1, 1) for groundwater level simulation

Groundwater is a prominent resource of drinking and domestic water in the world. In this context, a feasible water resources management plan necessitates acceptable predictions of groundwater table depth fluctuations, which can help ensure the sustainable use of a watershed’s aquifers for urban and rural water supply. Due to the difficulties of identifying non-linear model structure and estimating the associated parameters, in this study radial basis function neural network (RBFNN) and GM (1, 1) models are used for the prediction of monthly groundwater level fluctuations in the city of Longyan, Fujian Province (South China). The monthly groundwater level data monitored from January 2003 to December 2011 are used in both models. The error criteria are estimated using the coefficient of determination (R2), mean absolute error (E) and root mean squared error (RMSE). The results show that both the models can forecast the groundwater level with fairly high accuracy, but the RBFN network model can be a promising tool to simulate and forecast groundwater level since it has a relatively smaller RMSE and MAE.

Groundwater Level Forecasting Using Artificial Neural Network for Devasugur Nala Watershed in Raichur District, Karnataka

Accurate forecasting of groundwater level is important for sustainable utilization and management of groundwater resources. The performances of Feed Forward Neural Network (FNN), Radial Basis Function Neural Network (RBF) and Elman or Fully Recurrent Neural Network (RNN) Artificial Neural Networks (ANN) in groundwater level forecasting were examined in order to identify an optimal ANN model for groundwater level forecast. Bayesian Regularization (BR), Levenberg-Marquardt (LM) and Gradient Descent with Momentum and Adaptive Learning Rate Back Propagation (GDX) training algorithms were used to train each ANN. Devasugurnala watershed located at northern part of Raichur district, Karnataka, under middle Krishna river basin was selected for the study. The results revealed that FFN-LM model with 3-10-1 architecture with least RMSE and highest correlation coefficient values was most efficient for monthly groundwater level forecasting for the study area, and was a promising tool for the forecasting of groundwater level.

Erratum to: Optimal design of BP algorithm by ACOR model for groundwater-level forecasting: A case study on Shabestar plain, Iran

Groundwater always has been considered as one of the major sources of drinking and agricultural water supply, especially in arid and semi-arid zones. Thus, there is a need to simulate (i.e., forecast) groundwater levels with an acceptable accuracy. In this paper, we present two applications of intelligent optimization algorithms for simulations of monthly groundwater levels in an unconfined coastal aquifer sited in the Shabestar plain, Iran. First, the backpropagation neural network (ANN-BP) with seven neurons in its hidden layer is utilized to reproduce groundwater-level variations using the external input variables including the following: rainfall, average discharge, temperature, evaporation, and annual time series. In the next application, ant colony optimization is used to optimize and find initial connection weights and biases of a BP algorithm during the training phase (ACOR-BP). The results were found to be acceptable in terms of accuracy and demonstrated that a hybrid ACOR-BP model is a much more rigorous fitting prediction tool for groundwater-level forecasting. This study has shown that such a hybrid network can be used as viable alternative to physical-based models for simulating the reactions of the aquifer under conceivable future scenarios. In addition, it may be useful for reconstructing long periods of missing historical observations of the influencing variables.

Reconstruction of multi‐decadal groundwater level time‐series using a lumped conceptual model

AbstractMulti‐decadal groundwater level records, which provide information about long‐term variability and trends, are relatively rare. Whilst a number of studies have sought to reconstruct river flow records, there have been few attempts to reconstruct groundwater level time‐series over a number of decades. Using long rainfall and temperature records, we developed and applied a methodology to do this using a lumped conceptual model. We applied the model to six sites in the UK, in four different aquifers: Chalk, limestone, sandstone and Greensand. Acceptable models of observed monthly groundwater levels were generated at four of the sites, with maximum Nash–Sutcliffe Efficiency scores of between 0.84 and 0.93 over the calibration and evaluation periods, respectively. These four models were then used to reconstruct the monthly groundwater level time‐series over approximately 60 years back to 1910. Uncertainty in the simulated levels associated with model parameters was assessed using the Generalized Likelihood Uncertainty Estimation method. Known historical droughts and wet period in the UK are clearly identifiable in the reconstructed levels, which were compared using the Standardized Groundwater Level Index. Such reconstructed records provide additional information with which to improve estimates of the frequency, severity and duration of groundwater level extremes and their spatial coherence, which for example is important for the assessment of the yield of boreholes during drought periods. Copyright © 2016 British Geological Survey. Hydrological Processes © 2016 John Wiley & Sons Ltd

Application of integrated ARIMA and RBF network for groundwater level forecasting

The combination model of autoregressive integrated moving average (ARIMA) and radial basis function network (RBFN) is used for the prediction of monthly groundwater level. The ARIMA model is used to estimate the linear principal of time series, and the RBFN model is used to predict the nonlinear residuals. The proposed hybrid model is applied to forecast the monthly groundwater level fluctuations for two observation wells in the city of Xi’an, China. The monthly groundwater level data from the year 1998 to 2008 are used for training the applied models and the data from the year 2009 to 2010 are reserved for testing. Predicted data from the hybrid model are compared with those from the ARIMA model and RBFN model using the accuracy measures. The result shows that the proposed hybrid model has both the good linear fitting ability of ARIMA model and the great nonlinear mapping ability of RBFN model. The prediction accuracy rate is higher than that of any single model. Therefore, the application of the combination model in the prediction of groundwater level is effective and feasible.

Stochastic simulation of groundwater dynamics based on grey theory and seasonal decomposition model in a coastal aquifer of South China

Accurate and reliable prediction of groundwater level is a critical component in coastal water management. It is important to find a suitable model with acceptable accuracy. This paper presents a comparative study with seasonal decomposition method of time series analysis and GM (1,1) method, which are applied to test the model accuracy by simulating groundwater levels of two representative wells located at a coastal aquifer in Fujian Province, South China. The average monthly groundwater level was monitored during 2006–2011, the data set from 2006–2010 was used for model establishment, and that of 2011 used for predicting the dynamic change of groundwater level. The results indicate that the multiplicative model fits better than the additive model for seasonal decomposition method. Finally, the effectiveness of the model and prediction accuracy was evaluated based on root mean square error (RMSE) and regression coefficient ( R2 ). From the obtained results, it is concluded that the GM (1,1) method can be a promising tool to simulate and forecast groundwater level and serve as an alternative physically based model.

Evaluation of precipitation effects on groundwater levels in a Mediterranean alluvial plain based on hydrogeological conceptualization

This paper presents the study of the impact of precipitation changes on alluvial aquifer systems in Xanthi Prefecture plain region, Greece. Two study areas were investigated, taking into account several hydrological and hydrogeological data for a time period of 23 years (1989–2011) measured and elaborated in the framework of the hydrogeological conceptual model assessment of the aquifer systems. It is considered that the evaluation and presentation of this kind of material is informative and quite necessary enforcing the reliability of the approach of the research presented in this paper. Graphs are designed showing precipitation monthly values and groundwater level fluctuation values recorded by three groundwater level loggers installed at the study areas. The conjunctive and synthetic assessment of the impact of precipitation on groundwater level fluctuation is carried out at three levels regarding precipitation moving average values fluctuation classified into thee time periods, high groundwater level values distinguished in three time periods, and correlation between precipitation moving average values and groundwater level values with temporal discrimination within specific hydrological years. It is concluded that the influence of precipitation on groundwater level fluctuation in the study area can almost responsibly be indicated based on original and reliable data contributing to better understanding and managing the functioning of the investigated aquifer system under several climate changes.