hydrogeologic assessment of escalating groundwater exploitation in the indus basin, pakistan

Hydrogeologic assessment of escalating groundwater exploitationin the Indus Basin, Pakistan

S. Khan & T. Rana & H. F. Gabriel &Muhammad K. Ullah

Abstract Groundwater development has contributed sig-nificantly to food security and reduction in poverty inPakistan. Due to rapid population growth there has been adramatic increase in the intensity of groundwater exploi-tation leading to declining water tables and deterioratinggroundwater quality. In such prevailing conditions, thehydrogeological appraisal of escalating groundwater ex-ploitation has become of paramount importance. Keepingthis in view, a surface water–groundwater quantity andquality model was developed to assess future groundwatertrends in the Rechna Doab (RD), a sub-catchment of theIndus River Basin. Scenario analysis shows that if dryconditions persist, there will be an overall decline ingroundwater levels of around 10m for the whole of RDduring the next 25years. The lower parts of RD withlimited surface water supplies will undergo the highestdecline in groundwater levels (10 to 20m), which willmake groundwater pumping very expensive for farmers.There is a high risk of groundwater salinization due tovertical upconing and lateral movement of highly saline

groundwater into the fresh shallow aquifers in the upperparts of RD. If groundwater pumping is allowed toincrease at the current rate, there will be an overall declinein groundwater salinity for the lower and middle parts ofRD because of enhanced river leakage.

Keywords Groundwater/surface-water relations .Numerical modelling . Conjunctive water use .Salinization . Pakistan

Introduction

Rapid increase in the world population demands increasedcrop productivity per unit of surface water and ground-water consumed for irrigated agriculture. Environmentalsustainability of irrigated agriculture requires protection ofland and water resources whilst maintaining and enhanc-ing crop productivity. Irrigation activities need to remainprofitable and environmentally sustainable on the long-term basis, which is possible only through the conjunctivemanagement of surface water and groundwater.

Pakistan’s population increased from 87 to 134 millionduring 1982–1999 (Asian Development Bank 2002) andat present is over 144 million. During the same period,wheat and rice production increased at a similar rate,mainly because of the extra water available after theconstruction of Mangla and Tarbela dams and also rapidgrowth in the use of private tubewells in the Indus Basin.However, the live storage capacity of major storage dams inPakistan is decreasing due to sedimentation, e.g. the livecapacity of Mangla Dam—5.30 million acre feet (MAF),6.5×109 m3—decreased by around 20% from 1967 to2000 and the live capacity of Tarbela—9.30 MAF, 11.5×109 m3—decreased by over 40% from 1975 to 2000. Thepopulation trends show further increase in food and fibredemands and therefore escalating crop water demand. Amajor proportion of the increasing crop water requirementis being fulfilled through increased groundwater pumping.This has its limitations in terms of declining water qualityand increasing soil salinity and sodicity.

As Pakistan is a land abundant and water-short countrywith occasional water surplus periods during the monsoonseason, therefore, increasing and optimizing water pro-

Received: 22 September 2007 /Accepted: 10 June 2008Published online: 15 July 2008

* Springer-Verlag 2008

S. Khan ()) : T. Rana :H. F. GabrielCommonwealth Scientific and IndustrialResearch Organization (CSIRO),Locked Bag 588, Wagga Wagga, NSW 2678, Australiae-mail: [email protected].: +61-2-69332927Fax: +61-2-69332647

S. Khan : T. Rana :H. F. Gabriel :M. K. UllahInternational Centre of Water for Food Security,Charles Sturt University,Locked Bag 588, Wagga Wagga, NSW 2678, Australia

S. Khan : T. RanaCooperative Research Centre for Irrigation Futures,P. O. Box 56, Darling Heights, 4350, Australia

H. F. GabrielNational University of Sciences and Technology,Risalpur, 24080, Pakistan

Electronic supplementary material The online version of thisarticle (doi:10.1007/s10040-008-0336-8) contains supplementarymaterial, which is available to authorized users.

Hydrogeology Journal (2008) 16: 1635–1654 DOI 10.1007/s10040-008-0336-8

http://dx.doi.org/10.1007/s10040-008-0336-8

ductivity is a priority to meet the future food and fibrerequirements of its rapidly increasing population. Irrigatedagriculture in Pakistan is primarily practiced in the IndusBasin, which is the largest contiguous irrigation networkin the world and has been used for agriculture since theancient times. The annual average surface water supplythrough this network is about 180×109 m3 and thecommand area is 16.20 million (M)ha. In Pakistan, theclimate ranges from arid to semi arid. Surface watersupplies and rainfall are inadequate to meet the irrigationwater requirements at the basin level. This deficiency isbeing fulfilled by exploiting groundwater. Sustainable useof groundwater for agriculture, municipal and industrialusege is severely hampered by saline water intrusion inresponse to massive freshwater withdrawal. The mostprominent saline groundwater zones exist in the IndusBasin and in the mid portion of the doab (interfluvial areabetween two rivers). In doabs, fresh groundwater lensesoften overlie saline groundwater over large areas. Due tothe excessive withdrawals from aquifers, the saline waterlying under the fresh groundwater lenses and in the middleof doabs has been mobilized and is deteriorating the waterquality. The lateral salt-water intrusion into the freshgroundwater area and the vertical up-coning of the salineinterface is resulting in degradation of aquifers. This isalmost an irreversible process and needs special attentioneither to slow or stop it. Application of this salinegroundwater for agricultural production is resulting inhigher soil salinity and sodicity.

Conjunctive water management of surface and ground-water offers a potential solution to ensure environmentalsustainability and financial viability of irrigated agricul-ture. It refers to the management of water resources withina basin ensuring adequate amount of water of acceptablequality is made available in a timely manner for irrigation.Sources of water within a basin, which can be managedconjunctively, include supplies from dams and reservoirs,groundwater, agricultural drainage, sewage, industrialeffluent and rainwater. By meeting irrigation requirementsof various crops at varying growth stages, conjunctivewater management facilitates increased agricultural produc-tivity. A number of studies have been carried out by variousresearchers on conjunctive water management and its impacton crop productivity and related issues (; Bredehoeft andYoung 1983; Gorelick 1988; Lingen 1988; O’Mara 1988;Shah 1988; Siddiq 1994; O’Connell and Khan 1999;Brewer and Sharma 2000; Datta and Dayal 2000; Rajuand Brewer 2000; Sakhtivadivel and Chawala 2002;Jehangir et al. 2002; and Chaudhry and Shah 2003).Conjunctive management of multiple sources of water isgood in theory but has not worked well in practice so far(Vincent and Dempsey 1991). There are many difficultiesin actually carrying out effective and sustainable conjunc-tive water management. These include lack of institutionalarrangements and rules to control the use of surface waterand groundwater, and availability of resource data andmanagement tools for water professionals to managemultiple sources of water. Despite the significant advan-tages of conjunctive use, its potential has not been fully

developed and implemented in many real water systems(Pulido-Velázquez et al. 2006).

In a number of irrigation districts of Pakistan’s Punjabprovince, when surface water supplies are inadequate,irrigation with groundwater of marginal quality is acommon practice; this resulted in the depletion of aquifersand soil sodicity (Kijne 1996; Rust and van der Velde1994). Appropriate decision support tools can helpidentify useful technologies for allocating surface-watersupplies, derive methods for controlling waterlogging,rationalize groundwater abstraction, and help promoteartificial recharge of the aquifers. To address problems ofsustainable groundwater and surface water use, ground-water quantity and quality response to changes in rechargeand groundwater-pumping rates were determined by thedevelopment of a surface water and groundwater interac-tion model for the Rechna Doab as it offers tremendouswater-storage potential for agriculture in Pakistan. Thestorage capacity of the top 15 m of aquifers is more than10 times the current capacity of Pakistan’s largest damTarbela Dam (69×109 m3).

Hassan and Bhutta (1996) developed a regional lumpedwater balance model for the Rechna Doab. Their studyconcluded that the groundwater reservoir of Rechna Doabwas being depleted resulting in an average water table fallof about 2.30 m over the 1960–1990 period. Anotherstudy was carried out by Ahmad et al. (2002) in order tounderstand the impact of irrigation using groundwater onsoil moisture movement in Rechna Doab. The studyquantified the subsoil water fluxes in an environment withgroundwater irrigation by using the physically basednumerical model Soil-Water-Atmosphere-Plant (SWAP).The study area was limited to two sites, Pindi Bhattian andFaisalabad, only. Ahmad et al. (2005) investigated analternative methodology for computing the various waterbalance components of the unsaturated zone in RechnaDoab by using geo-information techniques. Using geo-graphic information systems (GIS) and a remote sensingtool, an aerial average net groundwater use of 82 mm/yearwas estimated. The results deviated 65% from the specificyield method and the deviation from estimates usingtubewell withdrawal related data was even higher.

ObjectivesSustainable conjunctive water management is constrainedby the seasonal surface-water flows, groundwater re-charge, quality and storage over a given period of time.For understanding the sustainability of the groundwaterresource in Rechna Doab, the specific objectives of thepresent study were as follows:

& Development and calibration of a flow and solutetransport model to describe the surface water–ground-water interactions in the Rechna Doab.

& Determination of spatial and temporal impact of futuresurface water and groundwater use scenarios in theRechna Doab with respect to the quantity and thequality.

1636


& Identification of the technical solutions for encourag-ing effective conjunctive water management in theRechna Doab.

Study area description

The Rechna Doab is the interfluvial sedimentary basin ofthe Chenab and Ravi rivers located in the heart of theIndus River basin irrigation system in Pakistan (Fig. 1). It

lies between longitude 71°48′ to 75°20′ E and latitude 30°31′ to 32°51′ N. It is a part of the alluvium-filled Indo-Gangetic Plain, comprising of 2.98 Mha of gross area, outof which 2.30 Mha are cultivated irrigated croplands. Thearea slopes in the south-west direction to a topographicrelief difference of 113 m. The average slope is 0.37 m/kmalong the 390-km length of the doab, which decreases byabout 25% in the lower reaches. The area falls in the rice-wheat and sugar cane-wheat agro-climatic zones of thePunjab province, with rice, cotton and forage cropsdominating the summer season (Kharif), and wheat and

Fig. 1 Geographic location of Rechna Doab in Pakistan

1637


forage dominating the winter season (Rabi). In some partsof the doab, sugar cane is also cultivated as an annual crop(Khan et al. 2006).

Administratively the Rechna Doab area is divided intoseven districts of Jhang, Toba Tek Singh, Faisalabad,Sheikupura, Hafizabad, Gujranwala and Sialkot. Thesedistricts also have 2–6 tehsils (administrative units).Rechna Doab is subdivided into 28 irrigation administra-tive subdivisions (Fig. 2). Of 2.30 Mha of total farm areain the Rechna Doab, 15.3% is located in the FaisalabadDistrict and the remainder is distributed as 8.9% in TobaTek Singh, 25.1% in Jhang, 17.5% in Gujranwala, 16.7%in Sialkot and 16.6% in the Sheikhupura Districts.

The physiography of the Rechna Doab consists of (1)active flood plains; (2) abandoned flood plains; (3) baruplands; and (4) Kirana Hills (longitudinal across the RechnaDoab). The groundwater quality in the Rechna Doab isdivided into three distinct zones: (1) fresh water zone (FWZ,with total dissolved solids (TDS) <1,000 mg/L) covering1.36 Mha; (2) mixing zone (MZ, TDS 1,000–3,000 mg/L)covering 1.42 Mha; and (3) saline water zone (SWZ, TDS>3,000 mg/L) covering 0.198 Mha (IWASRI 1988).

The Rechna Doab soils consist of alluvial depositstransported by the Indus River and its tributaries. The soilstextures are predominantly medium to moderately coarse,with favourable permeability characteristics and show asimilarity throughout the area (Ahmad et al. 2005). Themonotony of the bed of the alluvial plain is broken byscattered hills and bedrock outcrops near Chiniot (a tehsil

of Jhang District), Sangla Hill, and Shahkot subdivisions.The groundwater in the Rechna Doab occurs in two sub-basins at larger depths separated by the buried ridge.Groundwater of good quality is found in the upper parts ofthe Doab and in a 24–48-km-wide belt along the floodplains of the Chenab and Ravi rivers. Highly salinegroundwater is found in the lower and central parts of theDoab. The Upper Rechna Doab contains fresh water (TDSless than 500 mg/L), but in the central and lower portions,groundwater salt concentration varies from 3000 to18,000 mg/L. In the central and lower parts of the Doab,a majority of the tubewells are pumping marginal(500 mg/L) to poor quality groundwater, especially atthe tail ends of the canal irrigation system (Aslam 1997).

Due to scanty and erratic rainfall, successful agricultureis only possible in Rechna Doab through irrigation. Thecanal irrigation system was introduced in 1892 with theconstruction of the Lower Chenab Canal (LCC). Presently,almost two-thirds of Rechna Doab is fed by a perennialcanal system; i.e. the irrigation water flows constantly intoa secondary (distributary) and tertiary (watercourse) canalsystem as long as there is need for water and sufficientflow in the rivers (Ahmad et al. 2005).

The study area was subdivided into 20 zones based onhydrological and administrative considerations (Fig. 3).The 20 zones were Malhi, Sadhoke and Muridke, Shahdra,Gujranwala, Noushera and Sheikhupura, Sikhanwala,Mangtanwala, Chuharkana and Mohlan, Paccadalla, Nokhar,Sagar, Kot Khudayar andAminpur, Uqbana and Sangla Hill,

Fig. 2 Rechna Doab’s irrigation administrative units, headworks and canals

1638


Wer, Dhaular and Veryam, Buchiana, Tandlianwala, Bhaghatand Tarkhani, Kanya and Sultanpur, and Haveli.

Model description, calibration, and results

Model descriptionThe conceptual model of Rechna Doab is shown in theFig. 4. The model covers an area of 2.98 Mha betweenRiver Chenab and Ravi. This model was based on thehydrogeological studies carried out by Khan et al. (2003).For the model development, MODFLOW (Harbaugh andMcDonald 1996) and MT3D (Zheng 1996) packagesunder the PMWIN (Chiang and Kinzelbach 1998)environment were used.

The spatial domain represented in the model consistedof four layers (0–7, 7–30, 30–90 and 90 m to bedrock),106 rows and 132 columns (total of 13,992 cells in eachlayer with each cell being 2,500×2,500 m square size).This layer system was chosen to represent recharge,groundwater pumping and water quality aspects atdifferent depths within the doab. Stress periods of 183

and 182 days were used alternatively for Kharif (summer)and Rabi (winter) seasons with six computational timesteps in each stress period. The model was calibrated forthe period 1993–2000. The initial conditions used formodel simulation were specified as for premonsoon 1993,i.e. on 1 April 1993 (Khan et al. 2003).

In order to assess the impacts on groundwater qualitydue to changes in groundwater stresses, a numerical solutetransport model of the Rechna Doab was developed. Theconcentrations of salts in Rechna Doab vary aerially aswell as vertically. The present concentrations of salts inRechna Doab were used as initial conditions in the solutetransport models.

The regional solute transport model was linked withthe calibrated flow model, therefore, changes in ground-water quality at any location in Rechna Doab due tochanges in flow dynamics caused by changed rechargeand discharge rates can be readily assessed. According tothe field surveys and investigations carried out in the studyarea, sodium sulphate, sodium chloride, calcium bicar-bonate, magnesium bicarbonate, calcium sulphate, mag-nesium sulphate and potassium sulphate salts are present

Fig. 3 Division of Rechna Doab into hydrological zones (zone number in parenthesis)

1639


in the groundwater underlying the major part of the studyarea. Chemical composition of groundwater at differentdepths varies. The overall salinity content is given inTable 1 (from Ahmed and Chaudhry (1988)).

Parameters for deriving hydraulic propertiesThe lithological data of 390 bores were available forRechna Doab. In Rechna Doab, soil texture comprisesdifferent materials like clay, sand, silt and gravel. Thevalues of hydraulic parameters—hydraulic conductivity(K), specific yield (Sy,) and specific storage (Ss)—assumed

for different aquifer material are given in Table 2 and areused for the derivation of initial hydraulic parameters oftop two layers. The values of hydraulic parameters for thebottom two layers were derived from the past aquifer testdata. The hydraulic properties of these layers were lateradjusted during the calibration process.

The spatial distributions of horizontal hydraulic con-ductivity (Kh) for the top two layers were derived from thebore logs. The depth of different materials in each layerwas multiplied with the respective horizontal hydraulicconductivity and summed to estimate transmissivity ofeach layer. The transmissivity of each layer was then

Fig. 4 Schematic diagram of the conceptual model

1640


divided by the total depth of the layer (d) to get theaverage horizontal hydraulic conductivity (Eq. 1).

Kh ¼P

Khi diPdi

ð1Þ

The values of hydraulic conductivity for the bottomtwo layers were taken from aquifer test. The statisticalsummary of data and hydraulic properties for each layer isprovided in Table 3.

The spatial distributions of the vertical hydraulic conduc-tivity for the top two layers were derived from the bore logdata by assuming the vertical hydraulic conductivity (Kv) ofthe material is one-tenth of horizontal conductivity ofmaterials, given in Table 4 as the basis of previous studies(Greenman et al. 1967; Khan 1978; Aslam 1997; Rehmanet al. 1997). The general methodology used for calculatingthe vertical hydraulic conductivity is given in the following.The depth of all the materials in each layer was divided byrespective vertical hydraulic conductivity. The reciprocal ofthe sum of these values was multiplied with the total depthof the layer. This gives the average value of verticalhydraulic conductivity of the upper two layers (Eq. 2).

Kv ¼P

diP diKvi

ð2Þ

The vertical hydraulic conductivity for the bottom layerwas derived from the aquifer tests. Statistical summary ofvertical hydraulic conductivity for each aquifer is given inTable 4.

The specific yield values (Sy) were also derived fromdrill log data for the top two layers. The different depthsof all materials in each layer were multiplied withrespective specific yield. These values in each layer weresummed and divided by the depth of the correspondinglayer. This gave the average value of specific yield for theupper two layers (Eq. 3).

Sy ¼P

Syi diPdi

ð3Þ

Statistical summary of the specific yield for each aquifer isgiven in the Table 5.

Recharge to groundwater from canals depends upon anumber of parameters including geometry of canal,hydraulic conductivity of the bed and sides of the canal,and groundwater conditions underneath the bed of thecanal. It is possible to determine recharge from canals usinggroundwater flow equations if above parameters can beaccurately defined; however, unfortunately these parame-ters are not readily available for most of the canals. In thepast, a number of field studies were carried out by PakistanWater and Power Development Authority (WAPDA),Japan International Cooperation Agency (JICA) andInternational Sedimentation Research Institute (ISRIP).

Perkins and Jhonston (1963) attempted to statistically cor-relate canal seepage with the canal discharge using data fromsome of these studies. He developed relationships usingseepage studies on a number of canals in Rechna, Chaj andThal Doabs based on a set of 70 selected observations from300 canal seepage tests. Similar to Perkins’s original relation-ships, the following relationship was derived by the PunjabPrivate Sector Groundwater Development Project (PunjabPrivate Sector Groundwater Development Project Consultants1998) to estimate recharge from the canal to groundwater:

S ¼ 0:052 Q0:658 ð4ÞWhere S is seepage loss or recharge in ft3/s/mile and Q

is canal discharge in ft3/s. Using these recharge estimates,the conductance of the canal bed was calculated using thefollowing equation:

Conductance ¼ K lw

mð5Þ

Where K is the hydraulic conductivity; l is the length ofsegment in the groundwater model cell; w is the width of

Table 1 Chemical composition of Rechna Doab groundwater

Well number Welldepth (m)

Total dissolvedsalts (mg/L)

Ca ++(mEq/L)

Mg++(mEq/L)

Na+ andK+ (mEq/L)

CO3−2

(mEq/L)HCO3

−1

(mEq/L)Cl−1

(mEq/L)SO4

−2

(mEq/L)ECµS/cm

pH SARa

RTLA-7 57 1,050 1.25 2.23 8.69 0.9 8.82 0.85 1.60 1,591 8.4 6.6RTLB-18 110 580 1.6 2.91 1.99 0 5.0 0.2 0.70 879 7.8 1.3RTLE-13 76 3,100 0.8 3.88 27.79 0 10.0 11.4 10.25 4,697 7.8 20RTLG-15 91 5,000 1.5 6.9 42.40 0.4 0.4 32.0 12.40 7,576 7.9 21RTLH-10 98 10,600 2.46 5.31 103.46 0.2 5.40 50.6 65.0 16,061 8.2 52RTLH-13 91 15,000 25.45 36.05 133.06 0.6 2.50 140.8 50.56 22,727 8.3 24

aAdapted from Ahmad and Chaudhry (1988)Location of observation wells is shown in Fig. 5 *SodiumAdsorptionRatio SARð Þ ¼ Na mEq=Lð Þffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

Ca þ Mgð Þ mEq=Lð Þ2

p

Table 2 Aquifer properties values used for different lithologicalunits

Material type K (m/day) Ss (L/m) Sy

Clay 0.05 1.00E-03 0.05Clay silty 0.10 1.00E-03 0.07Clay silty with sand 1 1.00E-03 0.10Clay with interbed of sand 5 1.00E-03 0.10Clay with gravel 10 1.00E-04 0.15Sand 100 1.00E-06 0.25Sand fine 30 5.00E-06 0.15Sand medium 50 1.00E-06 0.20Sand 100 1.00E-06 0.25Sand with Silt 30 1.00E-05 0.20Sand coarse with gravel 150 1.00E-07 0.25Mixed sand and silt 100 1.00E-06 0.20Silt 1 1.00E-05 0.15Gravel 200 1.00E-07 0.25

1641


segment in the groundwater model cell; and m is the bankto bed height.

From field observations, the head (HRIV), and bottom(RBOT) of the channel were determined for input intoMODFLOW. Similar parameters were derived for theRavi and Chenab rivers from river geometry and hydraulicconductivity data. A top down approach was used tounderstand the spatial distribution of groundwater pump-ing to meet crop demand on a hydrological unit andadministrative area basis (Khan et al. 2003). Thisapproach subdivides the study area into a system ofchannel reaches and demand nodes linking these channelreaches, and, therefore, follows the topography of the area.This approach recognizes the data limitations e.g. avail-ability of groundwater pumping rates and therefore buildsthe desired complexity into the analysis only to answerspecific questions. The study area was divided into threenodal networks reflecting the direction of surface waterflow and connectivity of canals between the Chenab andRavi rivers as given below (Khan et al. 2003):

& Upper Chenab Canal and Marala-Ravi Link Canal(UCC-MR)

& Lower Chenab Canal and Qadirabad-Balloki LinkCanal (LCC-QB)

& Haveli and Trimu-Sidhnai Link Canal (Haveli-TS)

A lumped seasonal water balance was determined foreach of the demand nodes using the monthly canal

supplies, irrigation system loss estimates and net cropwater requirement. These results were used to provideinitial estimates of recharge and groundwater pumping inthe distributed surface-water–groundwater model. Thewater balance for the whole domain is presented inTable 6.

Model calibrationThe first step in model calibration was the identification ofthe calibration targets. Observed water levels in theRechna Doab were used for calibration purposes. Thesecond step consisted of determining the acceptable rangeof errors between simulated and measured calibrationtargets. As the third step, trial-and-error and inversesimulations were performed until simulated parameterswere within the acceptable range of errors. For this study,a combination of PEST and UCODE methods were used(Khan et al. 2003).

In the Rechna Doab Model, there were 5,178 activecells per model layer. There were four layers and themodel inputs that could be altered included leakagebetween layers, storage, hydraulic conductivity andconductance of channels. This equated to a possible62,136 input variables that can be altered to achieve thecalibration target. During the calibration, it is desirable tobase the comparison between the calculated and observedhead on original piezometer data rather than interpolatedpiezometer data because of the uncertainty involved in the

Table 3 Statistical summary of horizontal hydraulic conductivity,Kh (m/day). SD standard deviation

Statistics Layer 1(0–7 m)

Layer 2(7–30 m)

Layer 3 (30–90m)and Layer 4(>90 m, bedrock)

Mean 3.49E+01 7.79E+01 9.59E+01SD 4.09E+01 3.68E+01 4.84E+01Skewness 1.03E+00 −2.82E−01 1.43E+00Range 2.00E+02 2.00E+02 2.42E+02Minimum 5.00E−02 5.00E−02 2.3E+01Maximum 2.00E+02 2.00E+02 2.66E+02Count 390 390 40

Table 4 Statistical summary of vertical hydraulic conductivity, Kv(m/day). SD standard deviation

Statistics Layer 1(0–7 m)

Layer 2(7–30 m)

Layer 3 (30–90 m)and Layer 4(>90, bedrock)

Mean 1.63E+00 3.08E+00 9.59E+00SD 3.86E+00 4.84E+00 4.84E+00Skewness 2.14E+00 1.12E−00 1.43E+00Range 2.00E+01 2.00E+01 2.42E+01Minimum 5.00E−03 5.00E−03 2.37E+00Maximum 2.00E+01 2.00E+01 2.66E+01Count 390 390 40

Fig. 5 Location map of observation wells for Table 1

1642


interpolation process. However, interpolated piezometerdata have the advantage of being available in every modelcell making it easier to judge the success or failure ofevery model cell to replicate observation. A set of 190piezometer hydrographs was selected from the piezometerdatabase for dynamic history matching (Khan et al. 2003).

Calibrated aquifer parametersThe results of the calibrated model indicate that the Kh oflayer 1 ranges between 0.40 and 145 m/day. The 25% areaof this layer has Kh less than 22 m/day, and for another25%, the values of Kh are greater than 46 m/day. The Kh

values range between 22 and 46 m/day for the remaining50% of the layer while the average Kh of the wholeformation is 35 m/day. The Kh of the layer 2 rangesbetween 9 and 165 m/day. One fourth of the area of thislayer has Kh <77 m/day and the other one fourth hasvalues of Kh greater than 96 m/day. The remaining half ofthe layer contains Kh ranges between 77 and 96 m/day andthe average Kh of the whole formation is 86 m/day.

The calibrated model results show that the Kh of thelayers 3 and 4 range between 24 and 265 m/day. Thevalue of Kh is less than 78 m/day for around 25% area ofthe formation and values of Kh is greater than 111 m/dayfor another 25% of area. The remaining 50% of the layerhas Kh values between 78 and 111 m/day and the overallaverage Kh of layers 3 and 4 is 98 m/day.

The results of the calibrated model show that the Kv oflayer 1 ranges between 0.01 and 13.50 m/day. The values

of Kv is less than 0.30 m/day for 25% area of theformation and Kv values greater than 2.00 m/day representanother 25% of the layer. The Kv values range between78 and 111 m/day for the remaining 50% of the layer andthe average Kv of layer 1 is 1.60 m/day. The Kv of thelayers 2, 3, and 4 are between 0.02 and 15 m/day. The25% area of these layers has Kv <1.70 m/day and another25% area has Kv >5.20 m/day. The Kv grade 1.70–5.20 m/day represents the remaining 50% of the layer andthe average Kv of layers 2, 3 and 4 is 3.70 m/day.

Calibrated Ss in the first layer of the model rangesbetween 3.90E-05 m−1 and 1.00E-03 m−1. The 25% areaof this layer comprises of Ss values <0.45E-03 m−1 andanother 25% area has Ss values >0.70E-03 m−1. Ss rangebetween 0.45E-03 and 0.70E-03 m−1 for the remaining50% area of the layer while the average Ss of the wholeformation is 0.50E-03 m−1. The calibrated model param-eters show that the Ss of layers 2, 3 and 4 range between2.50E-06 m−1 and 1.00E-03 m−1. The Ss <9.70E-05 m−1

values cover 25% of the formation and Ss >0.26E-03represents another 25% of the layer. The remaining 50%of the layer has Ss values ranging between 9.70E-05 and0.26E-03 m−1 and the average Ss of the whole formationis 0.20E-03m−1.

The specific yield (Sy) for the first layer of the modelranges between 0.03 and 0.24. The Sy <0.10 values map25% of the total formation and another 25% of theformation has Sy >0.15. Sy ranges between 0.10 and 0.15for the remaining 50% of the layer while the average Syof the active formation area is 0.12. The Sy of the layer 2ranges between 0.07 and 0.24. Sy <0.19 covers 25% ofthe formation area and a range of Sy >0.23 representsanother 25% area. For the remaining 50% of the layer Syranges between 0.19 and 0.23 and the average Sy of theactive formation is 0.20.

HydrographsTo measure the performance of the model, calibratedwater levels were compared with the observed waterlevels for 190 observation bores. The historic data forthese wells, available from Kharif (summer) 1993 to Rabi

Table 6 Lumped seasonal water balance of the whole nodal network

Volume (MCM) Season (year)Kharif-1997 Rabi 1997–1998 Kharif-1998 Rabi 1998–1999 Kharif-1999 Rabi 1999–2000

Canal water inflow 25,219 15,305 28,800 15,307 31,525 14,772Canal water out flow 13,878 8,066 16,567 9,218 18,674 8,610Canal water diverted 11,341 7,240 12,233 6,088 12,851 6,162Escape from drains 117 113 57 150 81 45Main canal losses 4,196 2,898 4,601 3,004 4,885 2,861Unexplained loss/gain 1,665 1,463 1,550 −526 1,635 177Net distributed 5,364 2,766 6,025 3,459 6,250 3,078Conveyance and field losses 1,734 894 1,947 1,118 2,020 995Net available to crops 3,630 1,872 4,078 2,341 4,230 2,083Net crop water requirement 6,886 4,261 8,319 5,722 9,477 6,067Groundwater requirement 3,255 2,389 4,242 3,380 5,246 3,984Total recharge 5,615 2,064 4,220 2,075 3,896 1,993

Table 5 Statistical summary of specific yield, Sy. SD standarddeviation

Statistics Layer 1 (0–7 m) Layer 2 (7–30 m)

Mean 1.24E−01 1.94E−01SD 7.31E−02 6.15E−02Skewness 4.97E−01 −8.99E−01Range 2.00E−01 2.39E−01Minimum 5.00E−02 1.10E−02Maximum 2.50E−01 2.50E−01Count 390 390

1643


(winter) 1999–2000 period, were used for the calibrationpurposes. The hydraulic conductivity, specific yield,specific storage, vertical leakance and recharge anddischarge values were adjusted until reasonable matcheswere obtained between the observed and simulated waterlevels for all observed hydrographs.

The model successfully simulated the observed waterlevels in all the study area. As an example the twohydrographs for zone 1 (Malhi) of the study area areshown in Figs. 6 and 7. It can be observed from thesefigures that there is a close agreement between theobserved and simulated heads and the overall trend ofthe observed groundwater hydrograph is also followedwell by the modelled data.

Water level contoursFor further calibration performance measurements of themodel, subjective assessment of the goodness of fitbetween modelled and measured groundwater levelcontour plans was performed by comparing the consisten-cy of modeled contours in relation to spot heights ofmeasured groundwater levels. A comparison of the modelcalculated and observed spatial distribution of piezome-teric levels shows how well the model replicates thespatial variation of the interpolated observations. Thereare some local depressions depicted in the computedcontours due to higher spatial resolution of data and localrecharge and discharge processes. Water level contours forall 14 stress periods (April 1993–March 2000) showedthat the model very well replicates the whole study area. Acontour plan for the 14th stress period (October 1999–March 2000) showed that the calibrated model does agood job in reproducing the spatial distribution of waterlevels (Fig. 8). The model reproduced the interpreteddirection of the groundwater flow and closely approx-imates water levels in most of the study area and there was

no systematic over- or under-prediction of heads in mostpart of the model areas.

Statistical performance indicators for the regionalgroundwater modelQuantitative calibration performance measures generallyrelate to the calculation of piezometric head residuals (thedifference between measured and modelled heads) andassociated statistics at known monitoring locations. It isnot possible to draw absolute quantitative comparisons inregard to groundwater level contours, because contoursare the result of interpolations between data points, andare thus subjective, at least in part (subjective choices aremade even when selecting parameters or methods ofgenerating contours through software packages). UsingAustralian Groundwater Modelling Guidelines (Murray-Darling Basin Commission 2000; Middlemis et al. 2000,2001, 2002), quantitative measures of the average error of amodel are detailed and reported at the end (see Electronicsupplementary material).

However, these performance indicators provide lumpedmeasures of calibration that do not indicate the spatial ortemporal distribution of the error. In addition to thesemeasures, it is important to show that there is nosystematic error involved in the spatial distribution ofdifferences between modelled and measured heads. Thesimplest way to do this is to present a scatter diagram;therefore, a scattergram plot was produced with measuredheads on the horizontal axis, and modelled heads on thevertical axis, with one point plotted for each pair of data atselected monitoring sites (Fig. 9). All the points occurredwith a small degree of scatter about the line of perfect fit (a45° line through the origin representing an unattainableperfect calibration). It is also important that the plotted pointsin any area of the scattergram are not grouped consistentlyabove or below the 45° line in any segment of the plot, as this

Fig. 6 Calculated and observed hydrographs for piezometer L-51/16 in zone 1

1644


indicates a consistent over- or under-prediction, and a likelyfundamental flaw in the calibration. It can be clearly seenfrom the figure that there is no area of the scattergram wherethe points are grouped consistently above or below the 45°line in any segment of the plot. The coefficient ofdetermination (R2) was also calculated as 0.99, whichindicated a very high degree of correspondence between themodelled and interpolated observations.

The statistics in the Electronic supplementary materialare all based on head residuals. A systematic error inelevations will bias all of the statistics. There are cases,

however, when a simulated hydrograph might agree verywell with a measured hydrograph in pattern and ampli-tude, but differ in absolute magnitude, so that the twocurves run parallel to each other. Head-based statistics willsuggest a poor calibration, when in fact the calibrationmight be very good. Legitimate elevation residuals canresult from model discretization and interpolation of thelocations of measured and simulated sites, so that the realsites and model nodes are not at exactly the same place.To account for this effect, another technique is thestandard correlation function (r) between two time series

Fig. 7 Calculated and observed hydrographs for piezometer L-53/18 in zone 1

Fig. 8 Comparison of computed and observed water-level contours for the Stress Period 14 (Oct 1999–Apr 2000)

1645


(Zheng and Bennett 1995) which is mathematicallyexpressed as:

r ¼Pnt ¼ 1

ht � h� �

Ht � H� �

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiPnt ¼ 1

ht � h� �

2

s ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiPnt ¼ 1

Ht � Hð Þ2s ð6Þ

where h and H are the means of the modelled andmeasures heads respectively.

The standard correlation function (r) between two timeseries is calculated as 0.99, which is approaching unity,and is an indication for a good calibration.

Sensitivity analysisSensitivity analysis was carried for aquifer properties Kh,Kv and Ss. Results of the sensitivity analysis for shallowgroundwater (layer 3 of model) are given in Table 7. Thesensitivity analysis shows slight increase (less than0.30 m) in groundwater levels if Kh is decreased whilethe decrease in Kv results in decrease in hydraulic headsby more than 0.50 m. The model is least sensitive to Ss.

Comparison of advective and dispersive componentsof solute transport in the modelIn order to assess the relative significance of advective anddispersive solute transport in the model of the RechnaDoab Area, a Peclet number (NPE) is computed from thefollowing equation:

NPE ¼ Dc�L

¼ 2500

2:1ð7Þ

where Δc is cell length and αL is longitudinal dispersivity.The Peclet number comes out to be 1,190. For such

magnitude of Peclet numbers the solute transport isdominated by the advection, i.e. the transport of solutedue to velocity of the groundwater flow. Assuming adispersivity of 2.1 m and given cell length of 2,500 m(cell size 2,500×2,500 m), an advectively dominatedtransport was modelled.

Water balanceExternal stresses such as wells, areal recharge, evapora-tion, drains and streams were simulated to calculate thewater budget of each zone and the average values inmillion cubic meters (MCM)/year were determined for thewhole calibration period. Figure 10 shows the waterbalance results of zone 1 for the calibration period (April1993–March 2000). In the water balance diagrams storagechanges are referred as ΔS with plus or minus sign. Minussign refers to the water released from storage and plusrefers to the water added up in the storage. The waterbalance results for the Rechna Doab have always givendiscrepancies of less than 0.01%, which is generallyconsidered an acceptable error.

The horizontal inflow/outflow rates calculated by themodel refer to the flow rate of water horizontally into andout of the zone. These rates are generally an order ofmagnitude less than the vertical exchange rates. From this,it can be said that vertical flow is the most importantmechanism in the region. Vertical flow is generallydownwards with the rate of vertical flow in this directiondecreasing with the depth. All water balance results alsoindicate that the rate of vertical flow is decreasing with thedepth and so is the change in storage. In some zones, thehorizontal outflow is more than the inflow but in otherzones outflow is less than the inflow. The zones with highpumping, low recharge and less inflow than the outflowtend to have higher leakage from the rivers.

Fig. 9 Scattergram of measured versus modelled heads of set of 190 piezometer observations

1646


Model application–management scenarios

During the late 1990s, Government of Pakistan took aninitiative to privatize and replace public deep tubewells

with shallow private and community tubewells. A largenumber of tubewells have already been installed by thefarmers through their own resources. The public tubewellsinstalled in the hazardous groundwater areas were envis-aged to continue operating unless these were replaced withprivate shallow tubewells particularly in zones where freshgroundwater overlies the saline groundwater, e.g. middle ofdoab. One of the factors, which needs to be studied underfuture scenarios, was the change in spatial and temporalpatterns of groundwater pumping. For example, in thefuture, major stress on the aquifer will be in the shallowhorizons as the deep tubewells have been closed andrelatively shallow tubewells will be operating in the future.This will affect dynamics of flow and salt transport in theRechna Doab aquifer. Other factors which need to bestudied was impact of rapid rate of growth of privatetubewells and possibilities of shifting aquifer recharge anddischarge distribution to protect aquifer as a resource and tobetter manage shallow water tables and soil salinity.

Considering the premonsoon groundwater conditionsduring March 2000 as “initial conditions”, a number offuture scenarios were studied to simulate the futureresponse of the aquifer under the changed pumpingpattern. The regional groundwater model was used tosimulate the hydrodynamics of groundwater flow up to theyear 2025 under four scenarios. The outputs of ground-water flow model were used with the solute transportmodel to study the possible changes in the groundwatersalinity over the same period.

Table 7 Results of sensitivity analysis

Parameter Spatial groundwater level (m)Min 25% tile Median 75% tile Max

Kh1.75 Kh 132.40 170.90 191.90 211.90 240.201.50 Kh 132.30 170.90 191.90 211.90 240.301.25 Kh 132.10 170.80 191.90 211.90 240.30Kh 132.00 170.80 191.90 211.90 240.300.75 Kh 131.80 170.70 191.90 211.80 240.400.50 Kh 131.70 170.70 191.90 211.90 240.400.25 Kh 131.60 170.60 191.80 211.90 240.50Kv1.75 Kv 132.00 170.80 191.90 211.90 240.501.50 Kv 132.00 170.80 191.90 211.90 240.501.25 Kv 132.00 170.80 191.90 211.90 240.40Kv 132.00 170.80 191.90 211.90 240.300.75 Kv 132.00 170.70 191.90 211.90 240.300.50 Kv 132.00 170.70 191.90 211.90 240.200.25 Kv 131.90 170.60 191.80 211.90 239.70Ss1.75 Ss 132.00 170.80 191.90 211.90 240.301.50 Ss 132.00 170.80 191.90 211.90 240.301.25 Ss 132.00 170.80 191.90 211.90 240.30Ss 132.00 170.80 191.90 211.90 240.300.75 Ss 132.00 170.80 191.90 211.90 240.300.50 Ss 132.00 170.80 191.90 211.90 240.300.25 Ss 132.00 170.80 191.90 211.90 240.30

Fig. 10 Water budget (MCM/year) of zone 1 (Malhi) for calibration period (April 1993–March 2000)

1647


The following future management scenarios werestudied up to the time horizon of year 2025:

& Simulation of a dry future, year 2000. The rate ofgroundwater abstraction (13,750 MCM) and spatialrecharge due to rainfall and field level irrigation (6,500MCM) remains constant in the future. Groundwater ismainly pumped from layers 2 (7–30 m) and 3 (30–90 m).

& Spatially adjusted surface and groundwater-use patternsunder dry conditions. Surface water supplies are shiftedto the lower part of the doab and groundwater pumping isincreased in the upper part of the doab. The present rateof abstraction is increased by 30% and irrigation rechargeis decreased by 30% in the upper part of the doabwhile inthe lower part of the doab, abstraction is decreased by30% and irrigation recharge is increased by 30%.

& Increased groundwater following historic trends.Groundwater pumping is increased at the current oftubewell growth in the whole doab till 2010 and thenthis pumpage is maintained at the 2010 level till 2025

& Simulation of average conditions. Groundwater pum-page is maintained at the existing rate (13750 MCM)and the recharge of 1997 (10,000 MCM) is maintainedfor the simulation period of 2001 to 2025.

The following parameters were used while studying thefuture scenarios:

1. All the parameters/assumptions used on hydrogeolog-ical characteristics of the aquifer remain the same.

2. Groundwater conditions computed by the calibrated flowmodel for March, 2000 were used as “initial conditions”.Chemical quality of groundwater for selected piezometersalong the Doab is given in Table 1, which shows increasingsalinity levels fromupper to lower part of theDoab. Salinitylevels expressed as electrical conductivity (EC) units forindividual layers were incorporated using the piezometricsalinity levels for different locations and associated depths.Subsequent changes in the water quality of different layerswere computed by the model in accordance with the flowsystem simulated through the flow model and solutetransport simulated using MT3D model for variousscenarios and time periods up to the year 2025.

3. It was assumed that recharge from the irrigation systemwill remain unchanged during the next 25 years as nosignificant increase/decrease in the canal water suppliesis envisaged in the near future.

4. No dramatic changes in the hydro-climatic conditions orflows of Ravi and Chenab Rivers are being envisaged inthe future.

The model was run for 50 stress periods for the next25 years, i.e. Kharif (summer) and Rabi (winter) croppingseasons up to time horizon of year 2025.

Results of future scenarios

Model results for four scenarios are briefly describedbelow. The flow model was used to predict the behavior ofthe groundwater reservoir for the next 25 years, i.e. up tothe time horizon of year 2025. The flow model producedhydraulic heads and groundwater velocities which wereused for the future salinity scenarios. The MT3D modelwas used to study the impact of changes in the flowpatterns on the salinity of groundwater in the differentlayers of the aquifer. A summary of the spatial statistic ofthe groundwater level change and salinity change forfuture scenarios is given in Table 8.

Present dry conditions continue in the future(scenario 1): the rate of groundwater abstraction(13,750 MCM) and recharge (6,500 MCM) remainsconstant until 2025There is an imbalance of groundwater recharge anddischarge under conditions for the year 2000 (Fig. 11),when, net mining of groundwater resources was occur-ring. This is an extreme scenario, which was studied togain insights into possible groundwater depletion andcontamination.

Figures 12a) and 13a show spatial groundwater levelpatterns and change in the water elevations for the wholeof the doab and Fig. 14a shows the spatial distribution ofdepth to water table (DWT) for the whole doab. Thesegraphics show that if the present rates of recharge anddischarge continue, there will be an overall decline ingroundwater levels of the order of around 10 m for wholeof the doab. The lower parts of the doab, below theChiniot divide, will undergo the highest decline ingroundwater levels ranging from 10 to 20 m, which willmake groundwater pumping very expensive for farmers.This is depicted by the cost of the pumping curve withincreasing depths to groundwater levels (Fig. 15).

The difference between the March 2000 and year 2025model calculated salinity levels is presented in Fig. 16a,which shows that if the present rates of recharge anddischarge continue there will be an increase in groundwa-ter salinity levels for the upper part of the doab and atsome areas in the middle and lower part of the doab.

Table 8 Change in groundwater levels and salinity resulting from future scenarios

Scenario Groundwater level change (m) Salinity change (µS/cm)Min 25% tile Median 75% tile Max Min 25% tile Median 75% tile Max

1 −25.90 −19.40 −10.80 −2.20 1.30 −7,370 −332 −49 18 2,4322 −29.10 −8.50 −1.90 0 2.30 −6,380 −323 −52 15 1,3503 −29.50 −24.10 −22.10 −15.10 0.60 −6,450 −362 −59 22 8,6544 −23.00 −10.80 −1.50 −0.10 1.40 −8,436 −333 −38 30 1,403

1648


Fig. 11 Recharge and discharge for the calibration period: Rabi 1993–1994 to Rabi 1999–2000

Fig. 12 Water-level contours for a year 2025 (scenario 1), b year 2025 (scenario 2), c year 2025 (scenario 3), d year 2025 (scenario 4)

1649


Results of scenario 1 clearly show that present rates ofgroundwater pumping under drier conditions are notsustainable due to:

1. Decline in water-table levels which will make ground-water pumping expensive

2. Increased salinization of the aquifer due to highersalinity of recharge vertical up-coning and lateralmovement of highly saline groundwater.

Spatially adjusted surface water and groundwaterpatterns under dry conditions (scenario 2):groundwater abstraction is increased by 30%and irrigation recharge is decreased by 30%at the upper part of the doab, while at the lower partof the doab, abstraction is decreased by 30%and irrigation recharge is increased by 30%Spatial groundwater level patterns and change in thegroundwater elevations for the whole of the doab are

shown in Figs. 12b and 13b respectively. Figure 14bshows the spatial distribution of depth to water table(DWT) for the whole doab after 25 years. An analysis ofthese results shows that if the groundwater abstractions areincreased by 30% and irrigation recharge is decreased by30% in upper part of the doab, while for lower part of thedoab, abstraction is decreased by 30% and irrigationrecharge is increased by 30%, there will be an overalldecline in groundwater levels of the order of around 10–20 m for the upper part of the doab and a recovery ofaround 2–3 m for the lower part of the doab.

Figure 16b shows the difference between the originaland scenario 2 model calculated salinity levels in layer 3.Salinity trends for each zone were computed by the modelup to year 2025. The hydrographs and spatial change inthe salinity shows that if the groundwater abstraction isincreased by 30% and irrigation recharge is decreased by30% in the upper part of the doab and for the lower part ofthe doab, abstraction is decreased by 30% and irrigationrecharge is increased by 30%, there will be an overalldecline in groundwater salinity for the lower part of the

Fig. 13 Change in the pizometric level in a year 2025 (scenario 1), b year 2025 (scenario 2), c year 2025 (scenario 3), d year 2025(scenario 4)

1650


doab and an average increase of less than 500 μ S/cm(∼300 mg/L) for the upper part of the doab.

Results of scenario 2 indicate the need for a shift insurface water irrigation to the lower part of the doab to

reduce groundwater salinity problems and increasegroundwater pumping in the upper part of the doab toreduce waterlogging problems. An alternative conjunctivesurface water and groundwater irrigation policy betweenthe upper and lower doab can provide a means forsustainable water management for this region.

Increased groundwater following historic trends(scenario 3): pumping is increased with the sametrend as the growth of tubewells in the whole doabtill 2010, while the pumpage rate of the year 2010will be maintained till 2025Figures 12c and 13c show spatial groundwater levelpatterns and change in the water elevations for the wholeof the doab from the year 2000 to 2025. Figure 14c showsthe spatial distribution of depth to water table (DWT) forthe whole doab after 25 years. An analysis of these graphsshow that if pumping is increased with the same trend asthe growth of tubewells in the whole doab till 2010 andmaintained at this pumpage rate till 2025 (45% increase inthe pumpage from present levels), there will be an average

Fig. 14 Spatial distribution of depth to water table a scenario 1, b scenario 2, c scenario 3, d scenario 4

Fig. 15 Increase in the cost of groundwater pumping with declinein water-table depth

1651


decline in groundwater levels of the order of around 20 mfor the whole of the doab.

The difference between the year 2000 and scenario 3model calculated salinity levels is shown in Fig. 16c.Salinity trends for each zone were computed by the modelup to year 2025. The hydrographs and spatial change inthe salinity shows that if pumping is increased at thecurrent rate till 2010 and maintained at the rate of the year2010 till 2025 (45% increase in the present pumpage),there will be an overall decline in groundwater salinity forthe lower and middle part of the doab due to enhancedriver leakage and an overall increase for the upper part ofthe doab due to upconing from deeper saline groundwater.The decline in the lower and middle part of the doab ismainly due to a fresh-water movement from the waterbodies to these zones mainly as a result of excessivepumping and large-scale drawdown. This scenario clearlysuggests that an increase in the number of tubewells in thefuture is not sustainable. Groundwater regulation aimed atprotecting the quality and quantity of groundwaterresource is essential.

Simulation of average conditions (scenario 4):groundwater pumpage is maintained at the year2000 rate and the recharge of 1997 (10,000 MCM)is maintained for 2001–2025 periodScenarios 1 to 3 relate to relatively dry conditions whilescenario 4 relates to average conditions with an overallbalance between recharge and discharge. The analysis ofthe results of this scenario show the similarity to theresults of scenario 1, except lower drawdown levels due tothe higher recharge, which is about 55% more than thescenario 1. Hydrographs for each zone were computed bythe model up to year 2025. The hydrographs indicatesthere is an overall groundwater level drawdown in allzones with greater groundwater level decline in the lowerpart of the doab.

Figures 12d and 13d) show spatial groundwater levelpatterns and change in the water elevations for whole ofthe doab and Fig. 14d shows the spatial distribution ofdepth to water table (DWT) for the whole doab. Theanalysis of these graphics shows that if the present rates ofdischarge and recharge as that for 1997 exist for the next

Fig. 16 Change in the groundwater salinity for the Rechna Doab a scenario 1, b scenario 2, c scenario 3, d scenario 4

1652


25 years (stress periods 9 and 10), there will be an overalldecline in groundwater levels of the order of around 1.5 mfor whole of the doab, which is much lower than scenario1 (10 m). The lower parts of the doab below the Chiniotdivide will express decline in groundwater levels from 5to 10 m, which is 50% less than in the case of dryconditions, i.e. scenario 1.

The difference between the original and scenario 1model calculated salinity levels is presented in theFig. 16d, which shows that if year 2000 levels ofdischarge and year 1997 levels of recharge continue intothe future, there will be an increase in groundwatersalinity levels for the upper part of the doab, althoughthere will be an overall decline in salinity for the middleand lower part of the doab due to an increase in therecharge due to enhanced leakage from channels andrivers. Results of this scenario show that averageconditions with an overall balance between recharge anddischarge are sustainable with an acceptable decline inwater-table levels with an overall improvement in thesalinity levels.

Conclusions and recommendations

The consolidated exposed rocks near Chiniot, Sangla andShahkot subdivisions divide Rechna Doab into twoalluvial sub-basins. The aquifer characteristics show thatthe Rechna Doab aquifers are unconfined and have veryhigh water yields and transmission properties. The salinityconcentrations of groundwater increase with depth in mostcases. The salinity of groundwater ranges from 400 to600 mg/L up to a depth of 155 m in most cases.

Due to the strong dependence on groundwater forirrigation water supply, the lower parts of the doab havedeveloped semi-regional groundwater depressions withwater-table depths greater than 15 m deep in zone 17(Tandlianwala) and zone 19 (Sultanpur). These fallingwater tables have made groundwater pumping uneconom-ical for most farmers due to manifold increase in thecapital and operating costs. These water-table depressionsalso mobilize saline groundwater from adjacent regionsand from deeper groundwater resulting in salinization ofthe upper aquifers.

Scenario analysis has shown that if the present dryconditions continue, there will be an overall decline ingroundwater levels of around 10 m for the whole of thedoab during the next 25 years. The lower parts of the doabbelow the Chiniot divide will undergo the highest declinein groundwater levels (10–20 m), which will makegroundwater pumping very expensive for farmers. Thereis also a high risk of groundwater salinization due tovertical up-coning and lateral movement of highly salinegroundwater into the fresh shallow aquifers.

If groundwater abstractions are increased by 30% andirrigation recharge is decreased by 30% in the upper part ofthe doab, and abstraction is decreased by 30% andirrigation recharge is increased by 30% for lower part ofthe doab, there will be an overall decline in groundwater

levels of the order of around 10– 20 m for the upper part ofthe doab and a recovery of around 2–3 m for the lower partof the doab in the next 25 years. There will be an associatedoverall decline in groundwater salinity for the lower part ofthe doab and an average increase of less than 500 μS/cm(∼300 mg/L) for the upper part of the doab. If groundwaterpumping is allowed to increase at the current rate until 2010(a 45% increase on present pumping) and then remainssteady until 2025, there will be an overall decline ingroundwater salinity for the lower and middle part of thedoab due to enhanced river leakage and an overall increasein groundwater salinity for the upper part of the doab due toupconing from deeper saline groundwaters. This scenariodemonstrates any increase in the number of tubewells in thefuture will pollute groundwater resources as well as make itvery expensive for the farmers to use this vital resource dueto increased pumping costs. Groundwater regulation aimedat protecting the quality and quantity of groundwaterresource is essential.

It is recommended to:

& Promote a change in channel water usage patterns sothat around 30% less surface water is used in the upperpart of the doab, allowing a 30% increase in the lowerpart. At the same time there needs to be promotion ofgroundwater pumping in the upper part of the doaband discouragement of excessive pumping in the lowerpart. Increased groundwater pumping in the upper partwill substitute for the reduction in channel suppliesand also have the additional benefit of reducing croplosses due to waterlogging. Reducing pumping in thelower part will encourage a recovery of groundwaterlevels and slow or halt salinization of the shallowaquifers and soils.

& Develop a strategy to utilize the Rechna Doab aquifersas water storages. During periods of high flows in therivers, the shallow aquifers may be artificiallyrecharged to maintain the groundwater at depths of2–15 m. This can then be extracted by farmers forirrigation.

& Formulate a surface and groundwater managementplan for the Indus Basin using insights from thisRechna Doab study.

Acknowledgements Funding assistance from the Australian Centrefor International Agriculture Research (ACIAR) is appreciated.

References

Ahmad N, Chaudhry GR (1988) Irrigated agriculture of Pakistan.61-B/2, Gulberg-III, Lahore, Pakistan

Ahmad MD, Bastiaanssen WGM, Feddes RA (2002) Sustainable useof groundwater for irrigation: A numerical analysis of the subsoilwater fluxes. Irrigat Drain 51:227–241. doi:10.1002/ird.59

Ahmad MD, Bastiaanssen WGM, Feddes RA (2005) A newtechnique to estimate net groundwater use across large irrigatedareas by combining remote sensing and water balanceapproaches, Rechna Doab, Pakistan. Hydrogeol J (2005)13:653–664. doi:10.1007/s10040-004-0394-5

1653


http://dx.doi.org/10.1002/ird.59

http://dx.doi.org/10.1007/s10040-004-0394-5

Asian Development Bank (2002) Population and human resourcetrends and challenges. Asian Development Bank, Manila, Philip-pines.http://www.adb.org/Documents/Books/Key_Indicators/2002/default.asp. Cited 19 April 2008

Aslam M (1997) Predicting future tubewell salinity discharges,salinity management alternatives for the Rechna Doab, Punjab,Pakistan, vol V. Research Report Number R-21.5, PakistanNational Program, International Irrigation Management Institute(IIMI), Lahore, Pakistan

Bredehoeft JD, Young RA (1983) Conjunctive use of groundwaterand surface water for irrigated agriculture: risk aversion. WaterResour Res 19(5):1111–1121

Brewer JD, Sharma KR (2000) Conjunctive management in theHardinath Irrigation System, Nepal. Report No. R-94, Interna-tional Water Management Institute, Kathmandu, Nepal

Chaudhry A, Shah F (2003) Conjunctive use of surface andgroundwater resources under alternative institutional mecha-nisms. Department of Agriculture and Resource Economics,University of Connecticut, Storrs, CT, USA

Chiang W, Kinzelbach W (1998) Processing MODFLOW: asimulation system for modeling groundwater flow and pollu-tion. Hamburg, Germany, Zurich, Switzerland

Datta KK, Dayal B (2000) Irrigation with poor quality water: anempirical study of input use, economic loss, and copingstrategies. Ind J Agri Econ 55(1):26–37

Gorelick SM (1988) A review of groundwater management models.A World Bank Symposium on Efficiency in Irrigation, series 2,The World Bank, New York

Greenman DW, Swazenski WV, Bennett GD (1967) Groundwaterhydrology of the Punjab, West Pakistan with emphasis onproblems caused by canal irrigation, US Geol Surv Water SupplPap 1608-H

Harbaugh AW, McDonald MG (1996) User’s documentation forMODFLOW-96, an update to the US Geological Surveymodular finite-difference ground-water flow model, US GeolSurv Open-File Rep 96–485

Hassan GZ, Bhutta MN (1996) A water balance model to estimategroundwater recharge in Rechna Doab, Pakistan. Irrig DrainSyst 10:297–317

International Waterlogging and Salinity Research Institute (IWASRI)(1988) Data release on various aspects of waterlogging andsalinity in the Rechna Doab (Punjab). IWASRI Publication No.1,IWASRI, Lahore, Pakistan

Jehangir WA, Qureshi AS, Ali N (2002) Conjunctive watermanagement in the Rechna Doab: an overview of resourcesand issues. IWMI Working paper No. 48, International WaterManagement Institute (IWMI), Colombo, Sri Lanka

Khan MA (1978) Hydrogeological data: Rechna Doab, vol I.Project Planning Organisation, Water and Power DevelopmentAuthority (WAPDA), Lahore, Pakistan

Khan S, Rana T, Ullah K, Christen E, Nafees M (2003) Investigatingconjunctive water management options using a dynamic surface-groundwater modelling approach: a case study of RechnaDoab, CSIRO Technical Report 35/03, CSIRO, Clayton South,Australia

Khan S, Rana T, Yuanlai C, Blackwell J (2006) Can irrigation besustainable? Agric Water Manage 80(2006):87–99. doi:10.1016/j.agwat.2005.07.006

Kijne JW (1996) Water and salinity balances for irrigatedagriculture in Pakistan. Research Report no. 6, InternationalIrrigation Management Institute (IIMI), Colombo, ClaytonSouth, Australia

Lingen C (1988) Efficient conjunctive use of surface and ground-water in the People’s Victory Canal: a World Bank symposiumon efficiency in irrigation, series 2, The World Bank, New York

Middlemis H, Merrick NP, Ross JM, Rozlapa KL (2000) Thedevelopment of Australian guidelines for groundwater flowmodelling. Hydro2000 International Hydrology and Water

Resources Symposium, Perth, Western Australia, November2000

Middlemis H, Merrick NP, Ross JB, Rozlapa KL (2001) Ground-water modelling guidelines for Australia: an overview. Proceed-ings, MODSIM 2001 International Congress on Modelling andSimulation, Canberra, Australia, December 2001

Middlemis H, Merrick NP, Ross JB, Rozlapa KL (2002) Ground-water modelling guidelines for Australia: an overview of theneed for and use of the guidelines. International GroundwaterConference, Balancing the Groundwater Budget. Darwin, NT,Australia, 12–17 May 2002

Murray-Darling Basin Commission (2000) Groundwater flowmodelling guideline. Aquaterra Consulting, Project No. 125.final guideline, 16 January 2001, Murray-Darling Basin Com-mission, Perth, Australia

O’Connell N, Khan S (1999) Water use efficiency at a farm scale:SWAGMAN farm approach. In: Humphreys E (ed) Rice water useefficiency workshop proceedings, March 1999. Cooperative Re-search Centre (CRC) for Sustainable Rice Production, Rice CRCWorkshop Report P1-01/99, CSIRO, Clayton South, Australia; pp44–49. http://www.ricecrc.org/research/990301water.pdf. Cited on19 April 2008

O’Mara GT (1988) The efficient use of surface water andgroundwater in irrigation: an overview of the issues: a WorldBank symposium on efficiency in irrigation, series 2, The WorldBank, New York

Perkins TK, Jhonston OC (1963) A review of diffusion anddispersion in porous media. J Soc Petrol Eng 3:70–83

Pulido-Velázquez M, Andreu J, Sahuquillo A (2006) Economicoptimization of conjunctive use of surface water and ground-water at the basin scale. ASCE ASCE J Water Resour PlanManage 132(6):454–467

Punjab Private Sector Groundwater Development Consultants(1998) Canal seepage analysis for calculation of recharge togroundwater, Technical report no. 14, PPSGDP, PMU, P&DD,Government of Pakistan

Raju KV, Brewer JD (2000) Conjunctive management in the NorthBihar, India. Report No. R-95, International Water ManagementInstitute, Hyderbad, India

Rehman G, Jehangir WA, Rehman A, Aslam M, Skogerboe GV(1997) Principal findings and implications for sustainableirrigated agriculture: salinity management alternatives forRechna Doab, Punjab, Pakistan, vol I, Research report no. R-21.1, Pakistan National Programme, International IrrigationManagement Institute (IIMI), Lahore, Pakistan

Rust DHM, Velde EV (1994) Conjunctive use of canal andgroundwater in Punjab, Pakistan: management and policyoptions. Irrig Drain Syst 8(4):201–232

Sakthivadivel R, Chawala AS (2002) Innovations in conjunctivewater management: artificial recharge in Madhya Ganga CanalProject. International Water Management Institute, IWMI-TATAWater Policy Research Program, Annual Partners Meeting,Anand, India, February 2002

Shah T (1988) Managing conjunctive water use in canal commands:lessons from the analysis of the Mahi right bank canal, Gujrat.Research paper 3, Institute of Rural Management, Anand, India

Siddiq A (1994) Sustainability of Indus Basin: impact of tertiarysalinity on wheat productivity, damage assessment and futurepublic policy, PhD Thesis, University of Illinois, Urbana-Champaign, USA

Vincent L, Dempsey P (1991) Conjunctive water use for irrigation:good theory, poor practice, ODI Irrigation Management/Net-work paper 4, October 1991, ODI, London

Zheng C (1996) MT3D Version DoD_1.5: a modular three-dimensional transport model. The Hydrogeology Group, Tusca-loosa, USA

Zheng C, Bennett GD (1995) Applied contaminant transportmodelling. Van Nostrand Reinhold, New York, 440 pp

1654


http://www.adb.org/Documents/Books/Key_Indicators/2002/default.asp

http://www.adb.org/Documents/Books/Key_Indicators/2002/default.asp

http://dx.doi.org/10.1016/j.agwat.2005.07.006

http://dx.doi.org/10.1016/j.agwat.2005.07.006

http://www.ricecrc.org/research/990301water.pdf