Groundwater Modelling to Predict Management Options for Vavuniya Region Aquifer

The problems in sustainable groundwater management are of major and vital importance in the areas that comprise shallowly weathered and rarely fractured rocks with thin soil mantle, as recharge and yield are relatively low in these areas due to low porosity and permeability. In order to find out better ways and means to evaluate, develop and manage groundwater resources in these regions, an attempt Was made to understand the groundwater system of Vavuniya region aquifer through numerical modelling and to examine the behaviour of the aquifer under various operating conditions. The results reveal that the groundwater usage has already reached its optimum level in this region and immediate action is required not only to control further expansion of groundwater exploitation but also to regulate groundwater withdrawal especially in low rainfall years. Further, the influences of the river and subsurface dams in groundwater system were examined and the results show that the non-perennial river has less influences and the subsurface dams certainly have an impact on groundwater system but this has to be studied further in detail, in order to minimize the negative impact and utilize the merits.


Introduction
The Vavuniya region aquifer is an unconfined overburden aquifer.This aquifer comprises a thin soil mantle of an average thickness of 3m and a regolith of 10m over a rarely fractured rock base.Due to the low porosity and permeability, the hydraulic conductivity and specific yield of this aquifer are relatively low.Therefore the problems in sustainable groundwater management are of major and vital importance in this area.
The four-year observation well records since 1999 reveal that there is a substantial decline in the groundwater table in this region.Figure 1 clearly illustrates that the groundwater table did not reach its previous year maximum level during this period.

MiylW MayMX) MryMOl MiyXXO
Figure 1 Maximum Groundwater level at the end of recharging period.
This may be due to the excessive exploitation of ground water or due to the reduction in recharge of aquifer or the combination of both.
The intention of this research is to examine the behaviour of groundwater flow in this aquifer and come out with an appropriate strategy to ensure sustainability in groundwater management for this region.

The specific objectives are,
• to achieve an understanding of the basic mechanisms that govern the flow in the aquifer through numerical modelling.
• to prepare a water balance of the territory.
• to examine the influence of river and Subsurface Dams in the groundwater system.

Root Constant
The degree to which potential and actual evaporation rates diverge is a function of soil and vegetation characteristics and is explained by Penman (1949) in terms of a "root constant"; this is a measure of the amount of water readily available within the root range, expressed as an equivalent rainfall (Rushton and Ward, 1979).
As each zone has agricultural, forest and wetlands in different proportions, the weighted root constant for each zone was computed separately and a sensitivity analysis was carried out to examine the sensitivity to the root constant.The sensitivity analysis showed that the root constant is not that sensitive in these studies and it would not affect the recharge value for its selected range.

Rainfall recharge
According to the Penman -Grindley model (Penman 1950;Grindley 1967Grindley , 1969) ) the evapotranspiration takes place at its potential rate only if the soil moisture deficit is within the root constant.If the soil moisture deficit exceeds the root constant but within the upper limit of root constant, then only 10% of the potential evapotranspiration takes place.Beyond that, no evapotranspiration takes place.The recharge takes place only if there is no soil moisture deficit.This concept is used in the calculation of daily actual evapotranspiration and daily net rainfall recharge.

Irrigation Recharge
The Irrigation schemes contribute to recharge of ground water through, (i) Storage losses from reservoirs (ii) Irrigation losses from channels (iii) Irrigation losses from fields Though there are many methods available for more accurate prediction of these losses, they demand more field details, which are not available in the field in most cases.In such circumstances the researchers are compelled to assume more data which may not be correct and that leads to inaccurate estimations.
Therefore, one of the intentions of this study is to find appropriate methods, which require the data that is normally available, or is easy to collect from the field.
The operation study which deals with the reservoir water balance is a regular practice that is being carried out by the Irrigation Department to asses the water availability on a monthly basis.The estimation of percolation losses from the reservoir and irrigation losses from Channels and fields are some of the important steps involved in operation study.
The estimation of these losses using Ponrajah (1984) method is applied to many irrigation schemes in the region by the Irrigation Department for several years and the results are proven to be very satisfactory.
Therefore, the methods described by Ponrajah (1984) are used in this study.The basic concepts of these methods are as follows.

2.8.1
Recharge due to storage losses from reservoirs.
According to Ponrajah (1984), the seepage losses from reservoirs are directly proportional to the volume of water stored in the reservoir.Therefore, the recharge from tank storage per day can be expressed by the following equation.

Rts = CjXV/N (1)
where, Rts is the recharge from tank storage per day, V is the average monthly tank storage, N is the number of days in the calendar month and Cj is the Storage Recharge Factor.The storage recharge factor C 3 was found as a by-product from the calibration of runoff against reservoir daily water balance.The value of C 3 found from calibration is 0.015.

River Recharge
The tributary of Parankiaru, which passes through the domain, is a typical rain fed stream and has two branches.All 7 medium Irrigation Schemes in the domain are sited in these branches.Water flows in these streams, only during the-recharging period.MODFLOW (Prudic, 1988), which computes river leakage using Darcy's Law is employed in this calculation.

Groundwater withdrawals
The seasonal basic details required for the computation of Groundwater withdrawal are collected for each Grama Niladhari Division.Using this basic data, Groundwater withdrawal for each season was calculated separately for each zone.Based on these assumptions, the model calculates coefficients, called conductance, that are multiplied by head difference to determine flow between cells.(Reilly and Harbaugh, 1993)

Zero flow boundary condition
According to Boonstra and Ridder (1981), a groundwater divide, by definition, is a zero flow boundary as no flow occurs across the streamline running over the top of the divide.Accordingly the ridge of the groundwater head equipotential contours was considered as noflow boundary.

River Boundary Condition
MODFLOW's River Package (RIV) allows incorporating surface water boundary conditions into a groundwater flow model.Rivers and streams contribute to or drain from the groundwater system depending on the head gradient between the surface water body and the groundwater regime.

Constant Head Boundary Condition
For the scope of this study, Constant Head Boundary Condition was used only for the steady state simulations.

Recharge Boundary Condition
Groundwater recharge is accommodated by MODFLOW through the Recharge Package (RCH).The recharge package is designed to simulate aerial distributed recharge to the groundwater system.
The values of seasonal recharge for each zone by various recharge components are computed and converted into aerial distributed recharge and assigned to the respective zones.The effect of evaporation and evapotranspiration has already been already taken into account while computing recharge values and hence there is no need to input them into the model separately.

Ground water withdrawal
The values of seasonal withdrawals for each zone by various withdrawal components are computed and converted into pumping rates and assigned to the pumps of the respective zones.

Time Steps
The Duration of all transient state groundwater flow covers 7 Seasons starting from 1st of June 1999.Out of these data sets available for all seven seasons, the first five were used for calibration and the rest two were used for verification.The duration is not relevant for the steady state simulations.The details of stress periods covered by the transient state groundwater flow simulation runs are illustrated in Table 1.

Model predictions
The following simulation runs were carried out.

Rl -Steady state flow simulation
Hypothetical Constant Heads boundary conditions were imposed in the periphery of the domain and the river boundary conditions were imposed within the domain.These conditions were adjusted by trial and error so that the head equipotential contours generated by the simulation run matches with the observed head equipotential contours of the 1st season.Neither recharge nor withdrawal of groundwater was considered in this simulation.

R2 -Steady state flow simulation
The purpose of this simulation is to improve the heads generated by the simulation Rl at the observation well locations to match almost exactly with the observed values.The error in initial heads at the points of Observation Wells will affect the results of the transient simulation very much.Therefore, to generate heads at observation well locations almost equal to the observed values, constant head boundary conditions were imposed at well points.The other conditions remain the same as the simulation Rl.The matching between the observed and predicted head equipotential contours is found to be fairly satisfactory.No verification was done for Rl and R2, as none of the variable parameters was calibrated in these simulations.The groundwater flow pattern of these two simulations was examined by assigning some forward tracking particles in the domain.These values are slightly higher than the pumping test results.This may be due to the degree of accuracy of the method adopted to analyze the pumping test data or the degree to which the subsurface conditions complied with the assumptions or the combination of both.

Figure 3A: Comparison beticeen Model Run R3 & Model Run R4
coefficient were optimized while keeping the other inputs the same as R3. Figure 3 shows the comparison between the observed and predicted head equipotential counters of the domain at the end of 1095 days for the R4.
Though the actual values of CI (Channel recharge coefficient) and C2 (Field recharge coefficient) differ from zone to zone and season to season, an average value of 0.53 for CI and 0.7 for C2 showed satisfactory results in this simulation.
The purpose of this simulation run is to examine the sensitivity of the river boundary condition in this study.Except for the exclusion of river boundary condition, all other initial and boundary conditions were the same as R4.The optimized values of data obtained from R3 & R4 were used in this simulation.
The wells (OW3, OW36 and OW5) showed slightly higher predicted values in R5 run compared to R4 run.This implies that the stretches of river in this vicinity function as a sink and gain from the groundwater recharge.The wells (OWU, OW23, OW25 and OW26) showed slightly lower predicted values in R5 run compared to R4 run.This implies that the stretches of river in this vicinity function as a source and contributes to the groundwater table.
The two observation wells (OW22 and OW 14) showed slightly lower predicted values for the first two recharging periods and showed slightly higher predicted values for the third recharging period in R5 run compared to R4 run.This implies that the stretches of river in this vicinity function as a sink or source depending on the head difference between river and water table during the particular season.
In general, the changes in observed heads in the observation wells are small and this implies that the wells are less sensitive to the river boundary condition.This may be because the quantum of river recharge and river withdrawal are very small (less than 5%) compared to the total recharge and total withdrawal.In this simulation, the wells located closer to the downstream of a subsurface dam showed a fall in water level and the wells located closer to the upstream of a subsurface dam showed a rise in water level.The wells situated in the middle portion of a strip showed less sensitivity to Subsurface Dams.This information reveals that the sub surface dam helps to raise the water level in its upstream area at the expense of its downstream area.Further, the cumulative mass balance of run R6 showed that the difference in storage for the entire domain, caused by the sub surface dams, is very small.

Water Balance
After finding out the actual recharge into the aquifer and discharge from the aquifer of the study area during recharging and discharging periods, using the cumulative mass balance resulting from the model simulations, available In an ideal situation, the total recharge to an aquifer shall be equal to the total withdrawal from the aquifer for a period of one year (covering recharging and a discharging periods), in order to optimize the use of groundwater potential in a sustainable manner.Such water balance for the study area, derived from the average values of 8 seasons is given in Table 2.The above analysis shows that the groundwater withdrawal is not proportional to the recharge received during the year and this reveals that the groundwater is not managed in a sustainable manner in this region.If low rainfall is recorded continuously for two or more years, the groundwater situation will not only affect the agriculture but also will cause problems in meeting the domestic needs.This cautions the urgent need for a monitoring and regulating system to ensure the sustainability in groundwater management.
Further, it is noted that the % of total recharge is not directly proportional to the rainfall.This is because the quantum of recharge not only depends on the magnitude of the rainfall, but also depends on its distribution.For example, if 35 mm of rainfall is recorded in an evenly distributed manner for a seven day period during which 5 mm daily evapotranspiration takes place, no recharge will occur.But if this 35 mm of rainfall is recorded in a day, 30 mm recharge will occur.

Conclusions
Based on the results of the studies the following conclusions can be made.
(1) Since the groundwater is not managed in a sustainable manner in this region, the problems related to groundwater are inevitable especially during low rainfall years.Therefore, action is required to regulate the groundwater usage and withdrawal especially in low rainfall years.
(2) The non-perennial river passing through this region causes a lower effect in the 51 ENGINEER «|jJj groundwater system.Some stretches of this river function as a source and contributes to the groundwater table.Some other stretches of this river function as a sink and withdraws from the groundwater table.But, volumes of recharge and withdrawal are almost equal and very small.Therefore, the net recharge or net withdrawal is almost negligible.
(3) The analysis with subsurface walls reveals that the wall boundary helps to raise the water level within a strip in the lower area at the expense of the upper area.But the effect is comparatively very low.The low hydraulic conductivity and low specific yield of this aquifer may be the reason for this less response.Anyhow, it can be concluded that the wall boundary condition certainly has an impact on the groundwater system and this has to be studied in detail in order to minimize the negative impact and utilize the merits of this condition.

4 )
is the crop water requirement, P is the precipitation, Ea is the application efficiency, Ec is the conveyance efficiency and Kc is the crop factor.Average values of 60% and 65% were assumed for Ea and Ec considering the field conditions and records available in the Irrigation department.The values of Kc recommended for different growth stages of paddy were adopted appropriately.Conveyance losses are due to Seepage, Deep percolation and Evaporation.The deep percolation losses contribute to groundwater recharge.The magnitude of deep percolation from channels is equal to conveyance losses multiplied by a factor C r where, C, is the channel losses recharge coefficient (0 < C, < 1).2.8.3 Recharge due to irrigation losses from fields According to Ponrajah (1984), Farm losses can be expressed by the following equation.F.L=(l-Ea)x(ETc-PxEa)/Ea (evapotranspiration, ETc is the crop water requirement, P is the precipitation, Ea is the application efficiency, Ec is the conveyance efficiency and Kc is the crop factor.Average values of 60% and 65% were assumed for Ea and Ec considering the field conditions and records available in the Irrigation department.The values of Kc recommended for different growth stages of paddy were adopted appropriately.Farm losses are due to Surface runoff, Deep percolation and Leakage.Deep percolation losses contribute to groundwater recharge.Deep percolation from fields is equal to farm losses multiplied by a factor C 2 , where, Cj is the farm losses recharge coefficient (0 < C 2 < 1).
dimensional, groundwater flow model, which was introduced in 1984 as a versatile simulator of groundwater flow within an aquifer(Prudic, 1988), was selected for this study.MODFLOW can simulate steady state or transient flow; unconfined, confined, and leakyconfined conditions; spatial variations in hydraulic parameters; discharge to or from streams; pumping or injection from wells; and specified-head, specified-flux, and headdependent -flux boundary conditions.The block -cantered formation places a point, called a node, at the center of the cell, where the hydraulic conductivities are assumed to be uniform over the extent of a cell.

4. 3
R3 -Transient state flow simulation No constant head boundary conditions were imposed.Only river boundary conditions were imposed.Initial heads were imported from model Run R2.In this simulation the values of hydraulic conductivity and specific yield were optimized.The modelling results reveal that the Hydraulic Conductivity and Specific Yield values for 8 zones are in the ranges of 4 -7 m/ day and 5 -7% respectively.

4. 6 R6
-Transient state flow simulation Except for the inclusion of Subsurface Dams (refer Figure 4), the initial and boundary conditions were the same as R4.The optimized values of data obtained from R3 & R4 were used in this simulation.The purpose of this simulation run is to examine the effect of Subsurface Dams in this study area.

Figure 4 -
Figure 4 -Position of subsurface daws and observation wells in different zones

Table 1 -
The details of stress periods

Table 3 -The Water Balance Comparison
* (