Mahfuzur R. Khan, Clifford I. Voss, Winston Yu, and Holly A. Michael*

Water Resources Management in the Ganges Basin: A Comparison of Three Strategies for Conjunctive Use of Groundwater and Surface Water

Mahfuzur R. Khan, Clifford I. Voss, Winston Yu, and Holly A. Michael*

Journal: Water Resources Management

(*) Corresponding Author:

Holly A. Michael
Department of Geological Sciences

Department of Civil and Environmental Engineering
University of Delaware
255 Academy Street, Newark, DE 19716, USA
Email:

A-1 Introduction

This document contains supplementary materials for the article, “Water Resources Management in the Ganges Basin: A Comparison of Three Strategies for Conjunctive Use of Groundwater and Surface Water”. Section A-2 presents the synthesis of hydrological data collected from various sources. In section A-3, delineation of different type areas for the GWM (Ganges Water Machine) strategy is presented. The temporal development of hydraulic head at different location from the infiltrating river/canal for the base case, and the effects of vertical anisotropy on the storage for the sensitivity simulations are discussed in section A-4. In section A-5, GWM simulation results for each type area are presented in a table. Additionally, the temporal development of actual groundwater recharge from rainfall, irrigation return flow, and existing canal seepage for GWM is shown in a figure (Figure ESM-6). A hydrological budget in UP is presented in Section A-6. Temporal development of infiltration rates for the GWM and PAC (Pumping Along Canals) strategies are presented in Section A-7.

A-2 Hydrogeological Parameters

The average horizontal (Kh) and vertical (Kv) hydraulic conductivity were determined by simulating flow through a hypothetical heterogeneous aquifer system (Figure ESM-1) constructed based on published lithological cross-sections and fence-diagrams (Chaturvedi and Ramakrishna 1983; Umar et al. 2001) and reported values of Kh of the aquifer sediments in different areas within UP (e.g. Ala Eldin et al. 2000; Umar et al. 2001; Umar 2006; Rao et al. 2007; Umar et al. 2008; Raut 2009; Singh et al. 2009; Umar and Ahmed 2009). The simulated Kh and Kv were 15 m/d and 0.15 m/d, respectively, through the heterogeneous field, indicating that the coefficient of anisotropy (Kh/ Kv) is approximately 100 on the vertical scale of 300 m. The basin-wide average values of specific yield (0.16) reported in Government of India (1997) were used.

Fig. ESM-1 Postulated aquifer system simulated in this study to determine the effective horizontal and vertical hydraulic conductivity in UP

The UP aquifers receive recharge from rainfall, irrigation return flow, and canal seepage. Groundwater recharge from each of these sources was estimated using criteria established by the Government of India (1997). A forty-year record of monthly average rainfall data (World Bank unpublished) was used to estimate the recharge from rainfall. Estimated recharge rates during the monsoon season vary between 11 cm and 25 cm from west to east, with a spatial average of 15 cm. Dry season recharge from rainfall is much less and occurs relatively uniformly (2 cm to 4 cm during the dry season) over UP. Recharge from irrigation return flow was estimated using the data on irrigation areas existing during 1996-97 (Irrigation Department 1996-1997). There are two main cropping seasons in UP, namely the Kharif (wet season crop) and the Rabi (dry season crop). It was assumed that the irrigation activity remains constant for both cropping seasons. Recharge from irrigation during each cropping season varies from less than 5 cm in the north and southeast to over 10 cm in the west. These two sources provide a combined monsoon season recharge that ranges from 20 cm to over 30 cm over the basin, with a spatial average of 25 cm per season. Similarly, the combined non-monsoon recharge, which is mainly from irrigation, varies from less than 5 cm to more than 12 cm with a spatial average of 10 cm. Groundwater recharge from canals was estimated using GIS data of major canals (World Bank unpublished) and assuming a 35 m average width of the major canals. Based on canal density, UP was subdivided into a number of sub-basins; then groundwater recharge from canals per unit area of each sub-basin was estimated. The estimated canal recharge in canal irrigation areas varies from 2 cm/year to 10 cm/year depending on the canal density. When applied to the entire State, the estimated annual canal recharge is 4.5 billion m3. This recharge is added to the rainfall recharge in the model. This estimated total potential recharge for the 4-month wet season and 8-month dry season were simulated in the model using the MODFLOW Recharge package. In the model, each year was divided into two stress periods corresponding to the seasons.

The Ganges and most of its UP tributaries are perennial and closely connected to the underlying aquifers (National Institute of Hydrology 1999). Assuming that 80% of annual flow occurs during dry season, the average annual monsoonal and dry season flow at the UP-Bihar boundary (excluding the Gandak River) would be about 240 BCM and 60 BCM, respectively (World Bank unpublished). This dry season flow is from three sources: (1) snowmelt, surface water, and groundwater in the headwater region in the Himalaya in the north (Andermann et al. 2012), (2) surface water and groundwater in the hills and highlands in the south, and (3) baseflow from groundwater within UP. There is no primary estimate of baseflow in UP. A numerical modeling estimate (World Bank unpublished) suggests that shallow groundwater in UP contributes about 13 billion m3 of the dry season river flow. This amount is considered here as the total baseflow for UP rivers. This amount is required to return to the river downstream of the pumping section to maintain the dry season flow downstream. For GWM, this amount is required in addition to the upstream flow already routed from the upstream section of the well fields to the downstream section to maintain the dry season flow downstream.

River widths and the spacing between adjacent rivers were estimated using GIS data of the Ganges and its tributaries. River widths in the basin vary widely, from a few hundred meters to more than 1000 m. In general, however, stream widths increase downstream as new tributaries join the main channels. The distance between major tributaries of the Ganges ranges from 30 km to more than 60 km.

A-3 Type Area Delineation for the Ganges Water Machine strategy

Different type areas for the GWM strategy were determined based on the spatial distribution of the aquifer width, river width, and recharge from existing canals that vary considerably over UP (Table ESM-1). Eight distinct type areas were delineated (Figure ESM-2).

Fig. ESM-2 Type areas for GWM in UP. Total type-area length is approximately 3000 km

Table ESM-1 Attributes of type areas

A-4 Temporal Head Development and the Effect of Anisotropy for the Sensitivity Analysis

Figure ESM-3 shows the temporal development of hydraulic heads at different locations in the aquifer relative to the infiltrating water body i.e. river/canal for the base case simulation. Figure ESM-3 indicates that the hydraulic heads reach a dynamic steady-state after a few decades of pumping, with seasonal head fluctuations decreasing with distance from the river. Water table configuration at different time interval during the monsoon season is shown in Figure ESM-4 for the same simulation at dynamic steady-state. Changes in the pre-monsoon head and storage volume at the dynamic steady-state in response to the change in different aquifer parameters are summarized in Table ESM-2.

Fig. ESM-3 Hydraulic head vs. time at locations relative to the river. The head data is obtained at the middle of the pumping depth (100 m)

Fig. ESM-4 Water table elevations at different time intervals after the onset of monsoon infiltration.

Figure ESM-5 shows the temporal development of seasonal groundwater storage and the development of a new dynamic steady-state for different values of aquifer vertical anisotropy. The time required to reach a new dynamic steady-state increases with increasing anisotropy.

Fig. ESM-5 Annual storage volume with time for three values of anisotropy

Table ESM-2 Sensitivity analysis results. Effects of parameter variations on monsoonal storage volume and steady-state water table depth at the depth and location of the pumping wells. All values correspond to dynamic steady-state conditions and storage volume is per km length of river. Pre-monsoon head is the simulated maximum head at a distance of 3 km from the river and at a depth of 100 m below surface. Changes (Δ) are given as % of the base case estimates, negative indicates a decrease

*million m3 per year

**NA: Dynamic steady-state was not achieved within the 50 year simulation period.

A-5 GWM Simulation Results for Different Type Areas

The storage and pumping rate per kilometer river length for three different design pre-monsoon heads were simulated for the GWM scheme. The total annual storage for each type area was estimated by multiplying that by the total well-field length and by two (to include both sides of each river) (Table ESM-3).

Table ESM-3 Pumping rates and resulting storage per kilometer length of river for individual type areas and the total for the basin for GWM and the hydrogeologic model described in Section A-2.

Figure ESM-6 shows the temporal development of groundwater recharge for different design head scenarios of GWM. The vertical axis presents the actual groundwater recharge from rainfall, irrigation return flow, and existing canal seepage as a percentage of the potential recharge applied in the model top boundary using MODFLOW Recharge package.

Fig. ESM-6 Actual groundwater recharge with time as percentage of potential recharge applied in the GWM model for different design head scenarios

A-6 Water Budgets and Allocation

In UP, most of the domestic water supply is from groundwater. Currently, the total annual domestic and industrial water demand in UP is about 10 billion m3 (Planning Commission 2007). Considering 50% consumptive use (Planning Commission 2007), the net requirement is 5 billion m3.

Water loss due to evaporation and canal seepage occurs during water transportation via canals. The mean annual evaporation rate in the UP region ranges from 200 cm/year to 250 cm/year (Jain et al. 2007). When the existing canal networks are considered (see section 2.2.4), there are approximately 500 m2 of canal surface area per km2 of irrigated land. If the entire 12 Mha of irrigated land is brought under such canal networks, the canal surface area would be about 600 million m2. The total loss due to evaporation during the dry season would be less than 2 billion m3. Because DPR (Distributed Pumping and Recharge) involves pumping at the point of use, there would be little or no transport of the pumped water during the dry season. Therefore, the dry season evaporation loss from canals is assumed to be zero for this strategy. This study assumes lined canals for water transportation, therefore the water loss due to canal seepage is considered zero for all the strategies except GWM, for which seepage from existing canals is included in the model as groundwater recharge. Therefore, in addition to evaporative loss from canals, there is also canal seepage loss for GWM. This is added to GWM models as 4.5 billion m3 of groundwater recharge.