An Investigation of Recharging Groundwater Levels through River Ponding: New Strategy for Water Management in Sutlej River

Abstract

Groundwater is an essential water resource in the current era, and studying its sustainability and management is highly necessary nowadays. In the current area of research interest, the reduced mean annual Sutlej River flow, the increase in the population/built-up areas, and enhanced groundwater abstractions have reduced groundwater recharge. To address this issue, groundwater recharge modeling through ponding of the Sutlej River was carried out using a modular three-dimensional finite-difference groundwater flow model (MODFLOW) in a 400 km² area adjacent to Sutlej River. The mean historical water table decline rate in the study area is 139 mm/year. The population and urbanization rates have increased by 2.23 and 1.62% per year in the last 8 years. Domestic and agricultural groundwater abstraction are increasing by 1.15–1.30% per year. Abstraction from wells and recharge from the river, the Fordwah Canal, and rainfall were modeled in MODFLOW, which was calibrated and validated using observed data for 3 years. The model results show that the study area’s average water table depletion rate will be 201 mm/year for 20 years. The model was re-run for this scenario, providing river ponding levels of 148–151 m. The model results depict that the water table adjacent to the river will rise by 3–5 m, and average water table depletion is expected to be reduced to 151 to 95 mm/year. The model results reveal that for ponding levels of 148–151 m, storage capacity varies from 26.5–153 Mm³, contributing a recharge of 7.91–12.50 million gallons per day (MGD), and benefiting a 27,650–32,100-acre area; this means that for areas benefitted by dam recharge, the groundwater abstraction rate will remain sustainable for more than 50 years, and for the overall study area, it will remain sustainable for 7–12.3 years. Considering the current water balance, a recharging mechanism, i.e., ponding in the river through the dam, is recommended for sustainable groundwater abstraction.

1. Introduction

Groundwater is a vital water source worldwide because of its availability and good quality. It is also the largest source of freshwater storage for human consumption [1,2]. In recent decades, groundwater as a safe water resource has been taken for granted [3,4]. However, due to several circumstances including climate change and the boosted human population, groundwater is seriously vulnerable to depletion in several countries around the globe. Such a threat necessitates understanding the processes that make groundwater available for use [5].

The populations of numerous countries, e.g., India, Pakistan, China, and Egypt, depend largely on surface- or groundwater to fulfill their agricultural or human needs [6]. The issue of groundwater decline is increasing day by day, and appreciable work on groundwater recharge mechanisms is warranted in Pakistan’s water-stressed areas [7,8]. Various researchers have researched groundwater decline and possible recharge mechanisms [9,10]. The impact of the Rubber Dam in the Mahananda River, Bangladesh was studied at the groundwater level [11]. The modeling results revealed that the groundwater level decrease rate of 147–221 mm/year (Figure 1) over the last 15 years was reduced to 50–96 mm/year with Rubber Dam recharge. Analyses of the groundwater depletion rate and quality were conducted in Lahore using a modular three-dimensional finite-difference groundwater flow model (MODFLOW) [12]. The results indicated that due to overexploitation, the water level decline rate is 1.07 m/year, and water quality is deteriorating. Due to excessive depletion in several areas of the Indus Basin Irrigation System and waterlogging in some other areas of the same basin, there is a dire need for integrated water resource management [13]. Research was conducted on annual recharge computation for Rubber Dams in the Louhe River in China [14]. The results showed that the groundwater rose by more than 4 m in the areas adjacent to the river (Figure 2). Several other studies also established the same manner where groundwater level fluctuation based on the influence of hydraulic structure or surface water [15,16,17].

The alarming groundwater depletion situation is believed to be a constraint in the economic growth of countries with heavy dependence on surface- and groundwater usage [18]. Several researchers have tried to quantify the volume of water being used in excess of the volume recharged, termed non-renewable water [19,20]. As reported by [1], the values ranged from 200–1200 km³/year. An attempt has been made to mitigate the emergent situation of reduced water supply to the communities through the construction of numerous large and small reservoirs in affected areas, which serve the purpose of groundwater recharge [21]. There is an immense need to establish a balance between groundwater extraction and recharge or to return the groundwater levels to above the critical levels for survival, economic usage for agricultural demands, and human needs.

With the development of groundwater investigations, it is important to understand the development of comprehensive conceptual models and the analytical solutions or numerical methods of groundwater modeling. Modeling and simulation are now popular instruments to manage groundwater resources and their sustainable use. Groundwater models evaluate changes in the water balance of an aquifer caused by pumping, land-use changes, climate, etc., and how these changes affect groundwater storage, stream flow, lake levels, and other environmental variables.

Groundwater models are based on two equations: Darcy’s equation and the groundwater balance equation. The combination of both creates a partial differential equation. Usually, the finite difference and finite element methods are used for solving these equations, which require dividing the area into finite intervals [22]. The areas formed by these intervals are nodal. These nodal areas are also called cells. It is possible to add algebraic equations in place of partial differential equations. Accordingly, the model area is discretized into cells of suitable size depending upon the size of the model area and the quality of data available for the model. Groundwater depletion is believed to be a localized issue; hence, the flow models usually provide an accurate understanding of local groundwater recharge. Researchers have used groundwater flow models to analyze local depletion issues [23,24]. However, most of the studies are limited to the estimation of the reduction in groundwater extraction for maintaining balance in the system [25,26,27].

Although several pieces of research have been reported in the literature on groundwater level recharging, a research gap still exists in determining the balance between water abstraction and recharge and evaluating the sustainable groundwater recharge mechanism using numerical models, particularly in shallow unconfined aquifers, and the possibility of recharge through the incorporation of ponding action. Hence, the current research motivation was inspired by identifying the local groundwater recharge potential in the Indus Basin Irrigation System. The research presents a quantitative analysis of groundwater recharge via ponding in the Sutlej River combined with other components of the groundwater system using the groundwater model "MODFLOW." Thus, a 3D groundwater flow numerical modeling environment is proposed for practical applications and contaminant transport simulations.

2. Case Study and Data Description

Bahawalnagar is a city in Punjab Province in Pakistan with a total area of 1800 km² and is bounded by the Sutlej River. According to the 1981 Census of Pakistan, the city’s population was 367,367, and according to the 2017 Census of Pakistan, it was 815,143 (Punjab Development Statistics, 2017). This growth has been much more prominent during the last decade. The use of water by the city consists of domestic and agricultural activity, which has been enhanced by the increase in the population and cropping intensities. The main hydrological features in the area are the Sutlej River and Fordwah Canal.

The study area selected for groundwater modeling is 20 km × 20 km (400 km²), bounded by Sutlej River on one side and Fordwah Canal on the other, as shown in Figure 3. The area falls in a semi-arid to sub-humid subtropical continental region. The average annual high and low temperatures are 32.2 °C and 18 °C, respectively. The hottest month is June, with the highest temperature being 50.1 °C and an average high temperature of 42 °C, whereas the coldest month is January, with the highest temperature being 29.3 °C and an average high temperature of 20.6 °C. The average annual precipitation recorded in Bahawalnagar is 285 mm.

The study area has a general subsurface profile comprising silty clay/clayey silt up to a maximum depth of 30 ft (9.15 m) and silty sand/fine sand underneath silty clay/clayey silt up to a 100 ft depth and below. The general topographic elevation varies from 150 to 159 m. The water table elevation in the areas adjacent to the river varies from 142 to 143 m, and that in the areas adjacent to the canal is 151 m.

Data on factors including groundwater elevations, river inflow and stage, canal flow, full supply level, rainfall and evaporation, population and Landsat images, pumping and observation wells, aquifer properties, and the survey were collected. The historical water table data (2008–2015) reveal that the mean water table decline rate in the study area is 139 mm/year (Figure 4). The river flow data show that the mean annual inflow is 1766 cumecs. The canal data show that the canal’s monthly average flow varies from 20 to 45 cumecs. The data reveal that the population is increasing by 2.23% per year, whereas abstraction is increasing from 1.15% to 1.30% per year.

3. Adopted Methodology

3.1. Urbanization Rate

Landsat images from 1996, 2000, 2010, 2015, and 2018 were obtained with minimum cloud cover criteria. Layer stacking using a band combination of 4, 3, and 2 was carried out to obtain a multispectral image. A pan-sharpening tool was used to increase the spatial resolution of the images. The images were imported in Environment for Visualizing Images (ENVI) software, where supervised classification was performed by selecting training samples. Based on the training samples, the collected results were generated by assigning color ramps to particular feature classes. Then, the urbanized area was estimated to calculate the urbanization rate. The urban area maps for the year 1996 and the year 2018 are shown in Figure 5.

3.2. Backwater Effect

The Hydrological Engineering Center River Analysis System (HEC-RAS) model was developed to estimate the backwater effect of the lake. The corresponding stage value against monthly discharge values for the years 2014 to 2016 were calculated to finalize the dam’s height and compute the average stage for the groundwater model. The normal depth boundary condition (steady state) was assigned to the model using an average energy slope of 0.00012 (ft/ft). Considering the cross-section depth, a dam height of 4 m (EL of 148 m) was assigned, and a model was run to find out the backwater effect of ponding. Lake evaporation was also estimated using pan evaporation data.

3.3. Setting up of MODFLOW Model

A groundwater model for the study area of 20 km × 20 km was set up in Visual MODFLOW Premium software to conduct a quantitative analysis of groundwater recharge (a) in the current situation, with no ponding, and (b) by ponding in the river up to elevations of 148, 149, 150, and 151 m. MODFLOW solves a system of equations describing the major flow and related processes in the hydrological system using finite difference methods and is extensively used worldwide to research groundwater resource management (Hughes & Langevin, 2017). Topographic elevations were assigned to the model using a contour interval of 1 m. Run type, i.e., the transient flow, was assigned to the model for a simulation period of 20 years. The stress period was taken as 1 year, divided into 10-time steps.

A grid size of 200 m × 200 m and two layers were assigned to the model. Each layer was assigned flow properties, including hydraulic conductivity, specific yield, effective porosity, and specific storage, as shown in

Table 1. Flow Properties.

Description	Layer 1	Layer 2
Depth (m)	10	140
Hydraulic Conductivity, K (m/s)	1.95 × 10−7–6.8 × 10−6	1.1 × 10−5–1.35 × 10−5
Specific Yield, Sy	0.10	0.22
Effective Porosity	0.45	0.35
Specific Storage, Ss (1/m)	1 × 10−4

. Initial heads using the base year data were assigned to the model. The abstraction from wells and recharge from the river, canal, and rainfall were modeled in MODFLOW.

3.3.1. Model Calibration and Validation

Calibration of the model is an important step in model development as this exercise makes the model equivalent to the actual underlying aquifer conditions. In comparison, static levels never represent the true regional water table. The acceptable range of errors between the calculated and measured water table depends on the accuracy required, as no guidelines exist for calibrating the groundwater model. The trial-and-error method was used until the calibrated results matched the observed one.

Sensitivity analysis was used for the calibration of the model. The model was calibrated at the initial condition before the proposed dam conditions. Separate simulations were carried out to determine the effect of each parameter, and various parameters were calibrated during the calibration process. Hydraulic conductivity was varied in different ranges along different areas. In the model, there are 10,000 cells. The MODFLOW head observation well package was used to assign piezometric head values at four (04) piezometers for the years 2014 to 2016. The model was calibrated using the observed head values at four (04) piezometers for 2014 by varying the hydraulic conductivity in ranges.

Model validation is a process by which the calibrated model is shown to be capable of reproducing a set of field observations or predicting future conditions without further adjustment to the calibrated parameters. Several researchers have argued that, at least philosophically, a groundwater model cannot be validated in the absolute sense, like any scientific hypothesis. Thus, the term "model validation" should be avoided [28]. However, the model was validated for the same four (04) piezometers using the observed head values for 2015 and 2016. The scatter graph between the observed and calculated head values shows that all the calculated values of the model are well within the 95% confidence interval, as shown in Figure 6. Figure 7 shows the head-versus-time curves for the calibrated wells. The calibration result shows that the hydraulic conductivity (K) and specific yield (S_y) of layer 1 are 4.5 × 10⁻⁶ m/s and 0.10, respectively, and those of layer 2 are 1.1 × 10⁻⁵ m/s and 0.22, respectively.

3.3.2. Model Application (Scenario Modeling)

The period during which discharging and recharging components remain constant within the model is called the stress period. Twenty (20) stress periods are made in the model. Each stress period can be subdivided into several time steps. Trial simulations of the model area were made at 10-day time steps and 30-day time steps. It was found that no appreciable difference in the results was observed; therefore, the 30-day time step was selected for further simulations as the convergence of results was achieved and the computational time was less. Two scenarios were proposed for modeling. Scenario 1 was proposed to calibrate and validate the model based upon the observation data of previous years; however, predictions were also made for 20 years to compare the result with Scenario 2, which was proposed to estimate the quantity and lateral extent of recharge via ponding in the Sutlej River up to elevations of 148, 149, 150, and 151 m for the simulation period of 20 years using the calibrated model.

4. Application Results and Analysis

4.1. Water Table Decline Rate

The historical water table data of the study area displayed that the mean water table elevation in 2008 was 144.86 m and in 2015 was 143.89 m (Figure 4), thus showing a drawdown of 0.97 m in a total of seven (07) years; this indicates that the average decline rate in the study area during the period from 2008 to 2015 was 139 mm/year. This decline rate can be explained by several factors, including excessive water pumping during any planned groundwater-level operation; the noticeable increase in domestic needs as per the boost in built-up areas, as visualized in Figure 5b; and the region’s climate change, where rainfall events have changed dramatically. Though the decline rate is not excessive, it still represents the alarming situation of the study area, where the per capita groundwater resources are reducing, and furthermore, the population rate is increasing.

4.2. Urbanization Rate

The results of urbanization classification showed that the urban area in 1996 was 46 km², and in 2018, was 72 km² (Figure 8), thus showing an increase of 26 km² in 22 years; this indicates that the urbanization rate in the study area has been increasing at a rate of 1.1% per year over the last 22 years. In contrast, this increase has been more pronounced, i.e., 1.62% per year for the last 8 years. Hence, the above factors reinforce each other, i.e., the increasing population and urbanization, and are contributing to the groundwater table decline.

4.3. Backwater Effect

The results of the HEC-RAS model explain that for ponding up to an elevation of 148 m (a 4 m high dam), the backwater effect is up to about 25 km upstream with a lake area of 26.87 Km². The backwater effect can be seen in the longitudinal section of the reservoir obtained using HEC-RAS, shown in Figure 9. The comparison of lake evaporation with river inflows shows that lake evaporation varies from 0.37 to 2.35 cumecs, whereas the river inflows vary from 20 to 255 cumecs most of the time. The river inflow data display that the mean annual inflow of the river is 1766 cumecs. In contrast, the mean annual lake evaporation is 14 cumecs only, which indicates that mean annual lake evaporation is about 1% of the mean annual river inflow.

4.4. Results of Scenario 1 (Current Situation)

For scenario 1, the modeling results after 1 year and 20 years in the form of groundwater contour maps are shown in Figure 10 and Figure 11, respectively. The periods of 1 and 20 years were evaluated in a manner that enabled us to visualize the behavior of the catchment response toward the groundwater level for short-term and long-term assessment. Figure 10 and Figure 11 show that the groundwater level is declining because the groundwater abstraction is increasing compared to the recharge. The mean groundwater elevation computed from the groundwater contour maps was plotted with respect to time, as shown in Figure 12. The results of groundwater model studies indicate/predict the following:

• The water table is expected to decline at an average rate of 201 mm per year.
• The groundwater abstraction in 2014, 27.45 million gallons per day (MGD) against a recharge of 22.45 MGD, will increase to 33 MGD in 2034.

4.5. Results of Scenario 2 (Ponding up to Elevations of 148, 149, 150, and 151 m)

For Scenario 2, 148, 149, 150, and 151 m ponding levels were assigned to the model, and the simulations were adopted. The observed modeling results are as follows (

Table 2. Elevation Area Capacity ~ Recharge Volume.

Elevation (m)	Height (m)	Lake Area (km2)	Lake Volume (Mm3)	Recharge Volume (MGD)
148	4	26.8	26.52	7.91
149	5	32.5	60.07	9.45
150	6	34.8	93.71	11.00
151	7	37.6	153.32	12.50

):

i.

The elevation area capacity curve for the proposed ponding through the dam is shown in Figure 13. The contribution/volume of groundwater recharge with different dam ponding elevations (148–151 m) is presented in Figure 14. The groundwater model results show that upon ponding the river from elevations of 148 to 151 m, the lake area varies from 26.8 to 37.6 km², and its capacity varies from 26.5 to 153 Mm³, contributing a base-year recharge of 7.91 to 12.5 MGD, respectively.

ii.

The results of groundwater modeling for a dam ponding elevation of 148 m, in the form of groundwater contour maps after 1 year and 20 years of ponding/the simulation period, are shown in Figure 15 and Figure 16, respectively. Meanwhile, the results of groundwater modeling for a dam ponding elevation of 151 m, in the form of groundwater contour maps after 1 year and 20 years of ponding/the simulation period, are shown in Figure 17 and Figure 18.

iii.

The results of groundwater modeling show that the lateral extent of recharge via ponding will benefit an area of 27,650 to 32,100 acres, and the maximum water table will rise by 2 m and 5 m adjacent to the river in the case of ponding up to elevations 148 m and 151 m, respectively.

iv.

The results of the model (Figure 19) depict that the average water table decline in the study area is expected to be reduced from 201 to 151 mm/year (ponding at EL of 148 m) and 95 mm/year (ponding at EL of 151 m).

v.

The comparison of cumulative recharge and abstraction, as plotted in Figure 20 and Figure 21, revealed that for areas that benefitted from dam recharge, the groundwater abstraction rate will remain sustainable for more than 50 years for all the cases of ponding and for the overall study area; moreover, the cumulative abstraction will remain sustainable for 7, 8.3, 10.5, and 12.3 years in the case of ponding in the river up to elevations of 148, 149, 150, and 151 m, respectively. After that, some other recharging mechanism would be required.

4.6. Discussion

The indicated hypothesis for the research aims presented in the introduction section is clearly reflected in the numerical analysis’ modeling results. The ponding recharge adopted in the current investigation explained the importance of conducting such a hydrological practice as "river ponding" where the groundwater level can be maintained for different watershed purposes, including domestic water usage, agriculture/irrigation, and, most significantly, for soil moisture content. Although the apparent reason for the groundwater level decline was massive urbanization, several other factors such as climate change, human interaction, geoscience, and high rates of evaporation would be worthy of further investigation in the future for better insight into the actual reasons for dropping groundwater levels. Throughout modeling Scenario 1, it is clear that a serious operation plan is needed for groundwater abstraction for the studied case study. Otherwise, random abstraction can negatively impact groundwater levels, and thus, can be misleading for groundwater management and sustainability.

5. Conclusions

The groundwater simulation model (MODFLOW) was successfully set up, calibrated, and validated for a particular case study of the Sutlej River. The model was calibrated and validated for 3 years, from 2014 to 2016, and simulated for the next 20 years. The research focused on addressing the gap in determining the balance point between water abstraction and recharge and evaluating the sustainable groundwater recharge mechanism using numerical models, particularly in shallow unconfined aquifers. In addition, considering future prospects, the possibility of recharge through ponding action in the dam was analyzed. Hence, the current research motivation was inspired by identifying local groundwater recharge potential in the Indus Basin Irrigation System.

The main conclusions are as follows:

• The historical water table data of the study area revealed that the mean water table elevation had declined 0.97 m in the previous seven (07) years, which is equivalent to 139 mm/year. Though the decline rate is not particularly excessive, it still represents the alarming situation of the study area, where the per capita groundwater resources are reducing. Furthermore, the population, and thus, demand is increasing.
• The results of urbanization classification showed that the urban area has been increasing at a rate of 1.1 % per year for the last 22 years, whereas this increase has been more pronounced, i.e., 1.62% per year, in the last 8 years. Hence, the above factors reinforce each other, i.e., the increasing population and urbanization contribute to less recharge and more abstraction and ultimately result in an enhanced decline rate of the groundwater table.
• The results of the HEC-RAS model show that for ponding up to an elevation of 148 m (4 m high dam), the backwater effect goes up to about 25 km upstream of the dam with a lake area of 26.87 km². Based upon the discharge and stage data, it is observed that the average lake elevation remains below a 146 m elevation most of the time, which implies that about a 23 km² area will be inundated.
• The results of the groundwater model depict that the lateral extent of recharge via ponding will benefit an area of 27,650 to 32,100 acres, and the maximum water table will rise by 2 m and 5 m adjacent to the river in the case of ponding up to elevations of 148 m and 151 m, respectively.
• The results of the model scenarios depict that the decline rate in the overall study area is expected to be reduced from 201 mm/year (no ponding) to 151 and 95 mm/year (ponding at elevations of 148 and 151m, respectively), which shows a promising contribution of ponding action to groundwater recharge.
• The mass balance results of groundwater modeling show that for areas that benefitted from the dam, the groundwater abstraction rate will remain sustainable for more than 50 years, promoting environmental balance in the study area.

Considering the prevailing abstraction rate and recharge rate due to rainfall/the canal/the river in the study area, a recharging mechanism, i.e., ponding in the river through the dam, is recommended for sustainable groundwater abstraction. Together with this, an effective groundwater monitoring mechanism is also recommended to monitor and provide continuous information regarding the temporal sustainability of groundwater extraction and recharge.