Groundwater modeling of unconfined aquifer system of crystalline area - a case study in Lapsiya watershed, Hazaribagh, India

Ashok Kumar
Earth Resource Division
Remote Sensing Application Centre
IGSC- Planetarium, Patna - 800 001, India
Tele# +91-612-689001, Mobile# +91-9875036588
E-mail: Ashok_bcst@yahoo.com/ groundwater@indiatimes.com
Web: http://www/geocities.com/ashok_bcst

In the India considerably large geographical area comes under crystalline area. Groundwater occurrence and its management are the major task before the scientists and planner. These area experiences acute crisis of groundwater for drinking water and irrigation. In these areas due unconfined nature of aquifer system, the storage and retrieval of groundwater is major task before the scientists. The weathered materials are the principal aquifer system and ground water occurs under water table condition. Beneath the weathered horizon, fractures system within the basement surface is also supposed to be potential aquifer zone. But determination of fracture geometry is difficult task and these fractures zone have not been fully exploited. It has been established that aquifer geometry of the unconfined aquifer system is important parameters in understanding the groundwater storage, retrieval and recharge process in aquifer. The Digital Surface Terrain Modeling (DSTM) and Digital Basement Terrain Modeling (DBTM) exercise provides the upper and lower limit of the unconfined weathered aquifer system (Kumar et. al., 1997). This approach has been well tested in identifying the groundwater retrieval and storage sites in Chotanagpur region of India. But for complete understanding the complex mechanism of groundwater, this approach is not sufficient. The long term planning and management of groundwater needs understanding groundwater interaction with surface water, recharge, seepaze process, intake and rate of withdrawal in space and time and its long term effect on the aquifer system to achieve the sustainability. The entire exercise becomes complex process and it is outside preview of static modeling exercise such as DBTM approach. Several attempts have been made through computer modeling in alluvial plain of India but less stress has been made for the modeling of the aquifer in hard rock area.

In present study, modeling exercise has been attempted in Lapasiya watershed, Hazaribagh, India. It has helped in understanding the behavior of unconfined aquifer system with various varying input parameters. The outcome of the model helped in identifying suitable area for groundwater augmentation on the long term. The present model also helped in optimization of rate of new wells. The model has simulated up to a level to the near real field condition. The present modeling exercise and its results has given enough scope for taking up such types exercise in other parts of hard rock of India. There is still possibility for further refinement of various parameters. The present modeling exercise is a parts of UNDP training programme and it may not been treated as final.

Groundwater Modeling 
Modeling is an attempt to replicate the behaviors of natural groundwater or hydrologic system by defining the essential features of the system in some controlled physical or mathematical manner. Modeling plays an extremely important role in the management of hydrologic and groundwater system.

Objective of Modeling in Case Study 

1. The first objective of model was to simulate the condition similar to aquifer behaviors with time. The water table or equi-potential surface remains near to the surface after the monsoon; water table starts falling down from Nov. onwards and reaches maximum depth in the month of May/ June. After onset of monsoon, water table comes up.
   
   
2. To budget the groundwater resources
   
   
3. Find out the suitable area for bore well development and optimization of pumping rate and duration. In study area, 20 deep bore well have been identified through geo-hydrological and geophysical survey. But it sustainability could not be determined on long term basis.
   
   
4. To determine the sensitivity of the model the various input parameters i.e. recharge/ evapo-transpiration, hydraulic conductivity. So more stresses should be given in collection of field data.
   

Data required for the modeling and its source

Data Required by Model		Source of Data
System Geometry	Boundaries, elevations, thickness, surface drainage, bore location	Geological Map	Boundaries
Hydraulic Properties	Hydraulic conductivity, Transimissivity, Anisotropy, Leakge	Geophysical Surveys	Sections, thickness, bed rock, Digital Basement Terrain Model (DBTM)
Storage Properties	Specific yield, storage coefficient	Drilling Logs	Aquifers, Aquitards, Thickness, Bed rock
Sources and Sinks	Recharge, Pumpage, Leakage, Underflow, Baseflow, Evapotransipiration	Pump Tests	Transimissivity, Storage coeffecient, Leakage
Piezometric Heads	Water levels, Current and historical	Bore Records	Census, Location, Pumpage, Schedule, Hydrographs
Transport Properties	Porosity, Strengths, Constituents, Radioactivity	Surface Hydrology	Stream stage, Losses, Flood maps, Drainage, Baseflow, Channels
Concentration	Current and Historical	Meteorology	Rainfall, Evapotranspiration
Chemistry	Water analyses, Clay samples, Concentration maps	Water Use	Irrigation, Industrial, Urban, Efficiency, Waste, Backup source
Land Use	Soil map, Infiltration, Crop type	Piezometric Surfaces	Pre-pumping, Current, Short term drawdown, each aquifer, Hydraulic gradient

Ground-Water Flow Equation
The partial-differential equation of ground-water flow used in MODFLOW is (McDonald and Harbaugh,1988)



where 

K_xx , K _yy , and K _zz are values of hydraulic conductivity along the x, y, and z coordinate axes, which are assumed to be parallel to the major axes of hydraulic conductivity (L/T);

h is the potentiometric head (L);

W is a volumetric flux per unit volume representing sources and/or sinks of water, with W<0.0 for flow out of the ground-water system, andW>0.0 for flow in (T^-1);

S_S is the specific storage of the porous material (L^-1); and t is time (T).

Study Area
The Lapasiya watershed (AIS & LUS , 1988) is a part of Upper Hazaribagh plateau and forms the 500-600 (above m.s.l.) meters erosion surface. On the whole the plain is undulating with some minor ridges interrupting the level nature topography. The area may be termed as buried pediplain. The cover material is formed by coarse alluvium in the immediate valley of streams while rest of the pediplain has a gravely ferruginous soil. The porosity of soil does not permit wetting of the topsoil and the water rapidly percolates to the lower horizons. The present study area is a part of upper Hazribagh plateau. The watershed has total areal extent of 85 sq. km. Area on average receives 1322.41 mm of rainfall.

Surface Water Resource
Total 55 water bodies mostly ponds/ tanks have been identified in the watershed with the help of remotely sensed data. In which “Charwa” dam are the major water body and its areal extent are approximately 100 ha. The entire water bodies nearly harvest 8-10 % of the total annual rainfall (Kumar, 1997). 

Land Utilization
Kharif (paddy crops) including current fallow, water body, settlements etc covers 67.43 percent of watershed whereas rabi crop covers 07.43 per cent of the watershed area. The areal extent of rabi crops is indicator of utilization status of surface and ground water (Kumar, 1997). 

Aquifer System 
Thick weathered material serves as potential aquifers. In the valley portion water table generally cuts the topographic surface and groundwater get lost as seepage (spring). Water table in the valley portion ranges between 2.00m to 3.0m b.g.l. and generally deep on the upland in the range of 4 to 10m b.g.l. (Kumar, 1997). It has been observed that in case of maximum thickness of saturated weathered horizon of phreatic aquifer about 12m, yield of the dug wells range from 1.0m3 to 2.5m3 / day for a draw down of 0.5m to 3.00 m and well recuperates within 2 to 24 hr. Specific capacity of the aquifer varies from 1.39 to 5.61 lpm/m. draw down for the hilly areas having thin mantle of weathered material and 3.12 to 8.54 lpm/m draw down to low lying areas underlain by thick weathered material and soil covers. It has been observed that 70 per cent of total groundwater reserves get lost as base flow in river (Bhattacharya , 1990 ).



Basement Topography / Depth of Weathering
Based on depth of basement obtained from anylysis of VES data, sub-surface topographic model/ basement topographic model for Lapasiya (fig. 6) has been generated. Average depth of weathering is approximately 15-20 m. 




Conceptual Model 

1. As discussed in the previous sections, topography is undulating and pediplain has developed over granite gneiss’s with high drainage network. The channel of 4th order drainage remains wet throughout the year due to seepage of groundwater. Therefore, wet channel may be assumed as constant head boundary for present modeling exercise. Otherwise, it will be difficult to do the modeling of the area. We may also assume, wet channel as drain boundary condition. For this purpose, data on base flow in the channel is essential besides the drain conductivity. In the present exercise constant head boundary condition has been taken into consideration.
   
   
2. Although, aquifer system in hard rock consists of weathered and fractured system. The modeling of fractures is beyond the scope of present study because it is complex and detailed field data on fracture geometry and geo-hydrological characteristics is required. In hard rock area, the weathered material serves as principal aquifer. This aquifer is unconfined in nature and groundwater occurs under water table condition. Therefore, top layer excluding fractures has been taken for modeling. This is single layer case (Fig. 1.1).
   
   
3. Other basic assumption has been made in delimiting the area i.e. watershed. In practical purposes, the major water divides i.e. Lapasiya watershed outer boundary has been taken as no-flow boundary in modeling (Fig. 1.2).
   

Software Used - Visual MODFLOW 2.8
Visual MODFLOW is a computer program based on USGS MODLOW code with pre and post processor. It simulates three-dimensional ground-water flow through a porous medium by using a finite-difference method. Groundwater flow within the aquifer is simulated using a block-centered finite-difference approach. Flow associated with external stresses, such as wells, areal recharge, evapo-transpiration, drains, and streams, can also is simulated. The finite-difference equations can be solved using different solvers.

Input to the Model

Upper Boundary  The Upper boundary of the aquifer has been taken from the Digital Surface Terrain Model (Kumar, 1997). The upper surface of aquifer has taken from the topographic elevation value available in the Survey of India topographical sheets. The upper surface of the aquifer can be further improved if the contour values available in 1:25000 scale of Survey of India will be taken into consideration. The model success very much depends on the simulation of the upper topographic surface (Fig. 1a).	Constant Head Boundary   As discussed earlier the wet drainage channel of 4th order have been taken as constant head boundary. The large tanks have also been taken as constant head boundary. In the present study same extent of channel has been taken for constant head boundary for the entire period of simulation. Length aspects of constant head boundary can be improved with the help of remotely sensed data of different time period (Fig. 1.3).	Evapo-transpiration   Its estimation needs information on soil physical characteristics, land cover types, atmospheric condition etc. In the present case study, evapo-transpiration value has been approximated from the data available for same agro-climatic zone. There is scope for further refinement.
		Observation Wells   Sites used for initial head have been taken as observation wells. This is required for testing the simulated results (calculated) with observed head (Kumar, 1997), Fig. 1.7.
	Hydraulic Conductivity   The hydraulic conductivity of weathered material is very difficult to estimate. Normal pumping test has serious limitations in hard rock area and obtained results are highly variable. Based on available data on the different parts of Chhotanag-pur plateau, it has approximated between as 0.5 to 1.0 m/day (Athawale, 1984 & Karnath, 1994), Fig. 1.5.
Lower Boundary  The lower boundary of the aquifer has been inputted from the earlier Digital Basement Topographic data (Kumar, 1997). This is also very important parameter, which is required to inputted in detailed due to erratic behavior of basement topography (Fig. 1b).
		Pumping Wells   In the present study area, there is three deep bore wells. Ground water is being mostly augmented by dug well. In Initial phase, total drinking water requirement of village has been taken as one pumping well into the system. Similarly groundwater draft for the irrigation purposes has been taken as separate well. Besides that the deep bore well sites identified in the earlier NRDMS project have also been taken into consideration (Kumar, 1997), Fig. 1.9.
	Recharge  The total estimated recharge into the system have been assumed for each month depending upon the amount of rainfall during the month. It has been distributed in between 2 per cent to 40 per cent. The recharge from monsoon rainfall have assigned as 270, 136 and 40 mm for the upland, midland and lowland respectively. Recharge from tank has been assumed as 0.5 m /day (Athawale,1984 & Karnath, 1994 ), Fig. 1.4
Initial Head   The data collected in the earlier NRDMS project (Kumar, 1997) has been taken into consideration and it has been inputted into the modeling environment. The water table data of Jan 1994 has been taken as initial head in this model in the model (Fig. 1.6).

Model Simulation

Steady State Simulation
First the model has been simulated in steady state for period of one day (Fig. 1.10 & 1.14). All data, such as constant head, recharge, evapo-transpiration have been inputted month wise so that transient state run may carried out month wise. The grid cells representing hill in the watershed became dry in the steady state run. Some other area also became dry and it has been re-adjusted by re-defining the basement geometry at the particular point. It has been corrected some time by adjusting the hydraulic conductivity. Steady state run of model has been carried out by using the various solvers (Preconditioned Conjugate Gradient Package (PCG2), Slice Successive Over-relaxation Package (SOR), Strong Implicit Procedure Package (SIP), WHS Solver for Visual MODFLOW) available within the visual MODFLOW. Many time default solver WHS has not converged whereas PCG2 has given good results.

Transient State Simulation
After the successful run in the steady state, model was run for one-year period at the stress period (Fig. 1.15) of one month. Initially model was simulated without pumping well and simulated results were compared. The model acted like the field situation i.e. rise of water table in the monsoon period, decrease in water table after the monsoon. This indicates conceptual model and initial parameters were ok. Input parameters can be further refined i.e. spatial variation of recharge at different macro/micro-landform and soil types (topographic and soil maps used), spatial variation in evapo-transpiration in different land use units (land use map used), Variation in hydraulic conductivity on different landform and weathered material (aquifer hydro-geophysical property used). After refinement of the model input, model was finally calibrated for the actual field condition.

Thereafter model was simulated with the pumping wells (only drinking water wells and irrigation dug wells). Many of the pumping well dried up in one year (Fig. 1.13 & 1.16). This was due to cumulative pumping rate for the entire village was taken at one point. This can be further improved if it will be distributed in different location within the village area instead of putting cumulative value at one point. Similar results were obtained for the irrigation well. These error indicates that model is behaving correctly with the parameters. Due to very less hydraulic conductivity, radius of influence of wells in the weathered aquifer system is very limited even not more than 100m. Due to non-availability of spatial distribution of irrigation and drinking water wells, further improvement was not carried out. Few wells have not gone dry which are pumping less amount of groundwater for irrigation and drinking water purposes.



Thereafter, earlier identified deep bore well sites have been added into the system with constant pumping rate starting from 200 m3/day. These wells have been active for the period of one year. Most of them have gone dry at end of the one year. This indicates that we can not take the water at this rate. Model was thereafter model has been simulated with the reduced pumping rate. In this way different conditions have been generated and deep bore well pumping rates have been optimized. After running model with irrigation well, drinking well, deep bore well, more wells with less pumping rate was inputted into the system, this has helped in the determining the suitable area where we can observe the less draw down.

Model has been also simulated for the 10 years to generate the scenario for long tern planning of ground water of the aquifer system. 

Model Calibration
The observation well used in the model has been used for calibration of the model. The model calculated heads and observed heads have been analyzed. Majority of the heads falls in the 90 per cent confidence level (Fig. 1.18). The 95 per cent confidence level is supposed to be optimal. Therefore there is scope to refine the various parameters taking in-homogeneity in the aquifer system. Same exercise has been carried out in transient simulation. The calculated and observed heads have been plotted for the all the stress period. It has been found that heads are behaving with seasonal change in the water table.



Results
Inspection of model output has indicated that a place where basement depth is more, failure well is less. This means that well success is hard rock area depends on the thickness of the aquifer material. The largest water body in the watershed "charwa dam" effects on the surrounding ground water movement has been noticed. It has been observed that the up stream drainage area of the dam drains the groundwater to the dam. But much lateral control on groundwater movement has been noticed. The flow lines are coming to the dam area and it is moving towards down streamside.

The volumetric calculation of total available utilizable groundwater within aquifer has been made using output generated in the steady state. Total volume is 230.050x106 m3. This clearly indicates that availability of resource is not a problem. The model simulation has indicated that this type of aquifer can be pumped with slow rate (most appropriately at the rate of 100 m3/day) due to high draw down. Similarly, well can not be pumped for long duration at one stretch.

In the entire watershed putting huge number of dug wells can augment groundwater and shallow tube wells energized with 2 H.P. pumps. In middle portion and mid-north-east corner of the watershed, we can pump the water even at high rate i.e. up to 200 m3/day. Because simulation results are stable. This area gets ground water recharge from the upper reaches of watershed and recharge guided by the main river channel.

Another observation has been made regarding seepage loss of groundwater in drainage (presently it is a constant head boundary). It is decreasing with time due to continuous pumping. The seepage loss of groundwater can be optimized through the modeling simulation. 



Regional flow pattern of Groundwater
The flow direction and velocity vector obtained for different period indicates (Fig. 1.11) that majority of the flow direction is in NE direction. This is shortest route of the groundwater movement from the upper reaches to lower reaches. It has been also observed that micro water divides are also controlling the flow patterns. Few heads of the observation sites located near the constant head boundary i.e. drainage channel has not shown any change with time. This is because no seasonal variation has been taken into account in assigning constant head boundary for the whole simulation period.

Flow Budget from the model output
Results of flow budget (Fig. 1.17) indicate that an amount of 9712.80 m3 per day has been pumped on the 1st Jan. from the 18702-m3 available effective storage of the aquifer. After end of 31st Jan., all the pumping wells are not able to pump more than 4361.5 m3 per day. This indicates that some of wells have gone dry. Total storage available in the system also comes down to 7843.50 m3. The result indicates decrease in pumping volume till the month of June-July. The in out to the system is also decreases till the month of June-July. After start of monsoon i.e. June- July, situation reversed after increase in recharge to the system. Inspection of draw down of the individual pumping wells indicated that radius of influence wells are very limited and rarely interfering the other wells. Further wells are going dry only where depth of basement is shallow and pumping rate is high. It has been found that 50 m3/day upping rate is optimum. Even in some places, groundwater may be pumped with the rate of 100 m3/day - 200 m3/day

Conclusions
The modeling exercise of unconfined aquifer system of hard rock area in Indian condition is possible and model can be simulated to near real field condition. Based on present modeling exercise following points emerged out 

1. Model accuracy very much dependent aquifer geometry.
   
   
2. Groundwater reserve estimation of the entire aquifer system can be determined from the modeling. 
   
   
3. Model can be further improved if more and more spatial data on input parameter i.e. hydraulic conductivity, recharge, base-flow in the river, are to collected and inputted into the model for better control.
   
   
4. Modeling is a complex exercise; lot of discussion with experts and consultation is required.
   

Groundwater modeling of unconfined aquifer system can provide solution for estimating the available groundwater resource, optimizing the pumping rate and identifying suitable locations/ area where there will less adverse effects on the aquifer system in long duration pumping. The pumping rate of pumps can be optimised in the upper reaches to check groundwater seepage in the drainage channel. The modeling exercise has given better understanding of the aquifer behavior with change in different input parameter. 

Acknowledgement

Groundwater modeling of Lapasiya watershed, Siwane sub-basin, Hazaribagh, India was part of UNDP-DST training programme on GIS based Groundwater Modeling at Centre for Groundwater Studies, CSIRO, Wembley, Western Australia. Author is thankful to Dr. Chris Barber, Director, CGS, Western Australia, Dr. Kumar A. Narayan, Principal Research Officer; Dr. Ramsis Salama, Research Group Leader; Mr. Tonny Barr and Dr. Raiyast Ali, Scientists, Land and Water, CSIRO, Wembley, Western Australia, and Dr. Prabhakar Clement, Centre for Water Research, University of Western Australia, Perth, Australia for providing the training in the Visual MODFLOW and GMS package of groundwater modeling. 

References

• AIS & LUS ( 1988 ). Watershed Atlas of India, All India Soil and Land Use Survey, New Delhi.
  
  
• Athawale R. N. ( 1984 ). Nuclear tracer techniques for measurement of natural recharge in hard rock terrains. Proc. Int. Workshop on Rural Hydrogeology and Hydraulic in Fissured Basement Zones held at University of Roorkee, pp 71-80.
  
  
• Bhattacharya B. B. ( 1990 ). Hydrogeology and Groundwater Resources of Hazaribagh District, Bihar. Unpublished Report, CGWB, Eastern Region, Calcutta.
  
  
• Karnath K. R. ( 1994 ). Groundwater assessement, development and management, Tata McGraw Hill Publishing Company Limited, New Delhi.
  
  
• Kumar Ashok ( 1997 ). Natural Resource Management for Sustainable Utilisation and Management of Water Resources in Siwane sub-basin, Hazaribagh, Bihar, DST Project Report ( ES/011/212/95 ), BCST, Patna.
  
  
• Kumar Ashok, Sinha Ranjan and Prasad B. B. ( 1997 ). Digital Basement Terrain Modeling ( DBTM ) – A tool for sustainable utilisation and management of groundwater in hard rock area. National conference on emerging trends in development of sustainable groundwater sources held at Hyderabad from Aug. 17-28. JNTU.
  
  
• McDonald, M.G., and Harbaugh, A.W., 1988, A modular three-dimensional finite-difference ground-water flow model: U.S. Geological Survey Techniques of Water-Resources Investigations, book 6, chap. A1, 586 p.