MODFLOW and More 2006: Managing Ground-Water Systems - Conference Proceedings, Poeler, Hill, & Zheng - www.mines.edu/igwmc/
Numerical modelling of groundwater flow in south Chennai coastal aquifer
c. Sivakumar1, L. Elango 1 and D. Ganasunda~ 1Department
of Geology, Anna University,
[email protected], Chennai, India 2Central Ground Water Board, Ahmedabad, India
ABSTRACT Numerical modelling has been widely used for understanding groundwater flow and thus helps in efficient management of aquifer systems. Proper management of aquifer systems is necessary as groundwater is the major source for domestic and industrial purposes especially in urban and semi urban areas. Coastal aquifer located south of the city of Chennai, India is under stress due to pumping of groundwater to meet the ever increasing water needs of the city. Further the groundwater is the major source of water for the people living in this area. Currently, about 4.25 MGD (million gallons per day) is being pumped for domestic supply, mainly for drinking purRoses, and for industries both by private and government agencies. This aquifer is bounded by the Bay of Bengal in the east, Muttukkadu estuary in the south, Adayar river in the north. All these water bodies as well as the Buckingham canal which pass through this area along north south direction comprise of saline water. Hence, large amount of pumping coincided with inadequate management may lead to adverse impact on this fresh water aquifer due to ingress of saline water from these surface water bodies. The present study was carried out with the objective of developing a numerical model for this area in order to understand the behaviour of the system with the changes in hydrological stresses. The conceptual model of the hydrogeologic system was derived from geology, borehole lithology and water level fluctuations in wells. Groundwater of the study area was found to occur in both alluvial formations and in the underlying weathered rocks was conceptualized as an unconfined single layered system. The finite difference computer code MODFLOW (Modular 3-d finite difference flow) with Groundwater Modelling System (GMS) as pre and post processor was used to simulate the groundwater flow in this study. Simulation was carried out from the year 2000 assuming the initial groundwater level measured from about 28 wells located in the area. Aquifer parameters were estimated from about five pumping tests carried out by the government agencies. The K and S values determined from these tests ranged from 25 to 75m/day and 0.17 to 0.23 respectively. The model developed was initially calibrated with steady state run and later by transient state. The model input parameters namely the K and S values were varied by about 10% during calibration. The simulated groundwater head values compare reasonably with the observed trends. The model was later used to simulate the effect of increase in pumping and the changes in the rainfall pattern. The model developed can be used as an effective tool for the management of this aquifer system. INTRODUCTION Groundwater is the major source of drinking water in both urban and rural region. Need for groundwater has increased due to urbanization and industrialization. The demand for water has increased over the year and this has led to water scarcity in many parts of the world. During the past two decades, the water level in several parts of the India has been falling rapidly due to an increase in extraction and also due to failure of monsoons. Proper management of aquifer systems is necessary as groundwater is the major source for domestic and industrial purposes especially in urban and semi urban areas. Numerical modelling has been widely used for understanding groundwater flow and thus helps in efficient management of aquifer systems. A number of groundwater modeling studies have been carried out around the world for effective groundwater management (Corbet and Bethke, 1992; Storm and Mallory, 1995; Gnanasundar and Elango, 2000; Senthil Kumar and Elango 2004). Such a study was carried out in the coastal aquifer located south of the city of Chennai, India, This aquifer is under stress due to pumping of groundwater to meet the ever increasing water needs of the city. The present study was carried out with the objective of developing a numerical model for this area in order to understand the behaviour of the system with the changes in hydrological stresses. The finite difference computer code MODFLOW
'to
-
-
_ ••••
---
_
•• - _ .
_
•• _ -
•• _
•• /
_ . _ _ •• _
•• - . _ • • • • _ - _
:.1
-7v~".'t
\'-J.v •....."
g~
t-'Iv c:;lIIU
tJ v ;;:)\
tJIV"'C~~VI
was used to simulate the groundwater flow in this study. DESCRIPTION OF THE STUDY AREA sOl
1630'
Adayar River
~N
The study area lies just south of Chennai city which extends from Adayar river in the north to Muttukkadu estuary in the south. It covers an area of 20km in length and 3km in width. Eastern side of this area is bounded by the Bay of Bengal was shown in Figure 1. Climate of this area is characterized by an oppressive hot summer, dampness in atmosphere nearly through out the year and good seasonal rainfall. The summer season is from April to June which is followed by the Southwest monsoon season from the June to the September. October to December constitute northeast monsoon season with the associated rains being confined November and December. Average total rainfall is about 1260 mm/year. GEOLOGY & HYDROGEOLOGY
The study area exhibits varied physiographic features with elevation ranging from 1.0 to 11.5 m above MSL. The ground surface slopes towards the east and west directions from the central ele.vated portion. The region which lies 12 50' towards the west of the canal has topographic elav'ation varying between 1.0 and 8.0 m above MSL. Western- boundary of the study area has the highest elevation of 8.0 m above MSL. Geologically, the area comprises of unconsolidated sandy formation of different depositional environment belonging to Quaternary age. The alluvial deposits are Figure 1. Study area with comprised of interlayerd clay, silt, sand, gravel location of monitoring II and pebble beds. These formations overlie the we S Charnockites of Archean age. Charnockites xcuring below this sandy formation function as impermeable strata or bed rock. The total depth of the aquifer system ranges among 3 and 23 m (Gnanasundar and Elango 2000). The hydraulic conductivity of ~e aquifer lying west of the Buckingham canal has values between 25 and 35 m/day while the sub basin along the east of the canal has high hydraulic conductivity with values ranges between 45 and 75 m Iday. The specific yield of the aquifer ran.ges from 0.17 to 0.23. GROUNDWATER MODELLING The groundwater flow in the aquifer of the study area was simulated using a finite difference approximation of the three dimensional partial differential equations, (Rushton and Redshaw. (1979». This equation describes the groundwater flow under non-equilibrium and anisotropic medium, provided the principal axes of hydraulic conductivity are aligned with x-y Cartesian coordinates axes. MODFLOW, a well established, three dimensional finite difference groundwater flow model was used to simulate groundwater flow of this study. The pre and post processor developed by the United States Department of Defense Groundwater Modelling System (GMS), was used to give input data and process the model
MODFLOW and More 2006: Managing Ground-Water Systems - Conference Proceedings, Poeter, Hill, & Zheng - www.mines.edu/igwmcl
output. B~ock centered finite difference approach was used to solve groundwater flow equation in this
model.
MODEL FORMULATION The conceptual model of the hydrogeologic system was derived from a detailed study of the geology, borehole lithology and water level fluctuations in welts.. Groundwater of the study area was found to occur in both alluvial formations and in the underlying weathered rocks.. The measured groundwater head in monitoring wells, screened in the alluvial and weathered formation, is about the same.. Based upon this information, the alluvial and the weathered fractured rocks can be considered as a single unconfined layered system.. GRID DESIGN AND BOUNDARY CONDITIONS Constant head boundary
+
• 1+
."~ ~
~ Constant
~
No Flow Bound ary.-
head
boundary
· -
.
~C onstant
head
boundary
~
The study area was discretised as shown in Figure 2.. The dimension of each grid is 400m x 600m. A part of the area was discretised into a relatively finer mesh size as they comprise of numerous pumping wells. The size of grid in this area is 200 m x 400 m. The Bay of Bengal, Muttukadu Estuary and the Adayar river were taken as constant head boundaries as the water levels in these water bodies are generally at mean sea level throughout the year. The Buckingham canal was taken as internal constant head boundary with the constant head at mean sea level. In the western boundary of the area is considered as a no flow boundary as shown in Figure 2.. INPUT PARAMETERS
Aquifer Characteristics
·· · ~ ~
·· ··
·
1+
1+ 1+ 1+
· ·
-l-
....
1+ 1+
~
Constant head boundary
Horizontal hydraulic conductivity of the sandy aquifer ranges between 25 and 75 m/day. Aquifer lying west of the Buckingham canal has values between 45 and 75 m/day. The low K value along the west of the canal is attributed to the presence of sandy clay and clay. The specific yield of the alluvial aquifer ranges from 0.. 17 to 0.23. The specific yield of the aquifer existing between the coast and the canal ranges between 0.. 21 and 0.23 while the other portion has values ranging between 0.17 and 0.19..
Groundwater Abstraction Figure 2. Model grid pattern and Boundary conditions
The groundwater of the study area is abstracted for domestic, industrial and some for irrigational purposes. The average annual pumping is about 4.25 MGD. This has been obtained from the government agencies such as Public Works Department and Metro water. Pumping rate by private agencies was estimated through field visits . The pumping rate of wells that exist within the "housing complexes in the northern part of the study area was also taken into account and pumping rates from these wells were estimated from well inventory survey by collecting data on pumping duration in a day. About 90% of the total pumping occurs within the regions lying between the Buckingham canal and the coast. The pumping activity between the other sides of the canal is limited only to domestic purposes mainly for drinking and other house hold purposes due to poor groundwater quality.
Groundwater Recharge The rate of rainfall recharge varies within the study area. Two recharge zones, A and B, have been demarcated based on the soil types, geomorphic pattern, land use and topography. The rainfall recharge rate in zone A is around 40 -480/0 and in zone B it is around 25 -350/0 which was shown in Table 1. High recharge occurs in Zone A where sand and sand dunes occur. Recharge rate· in the areas of zone B is quite less due to the presence of sandy clayl clayey soil.
Recharge 0/0 Zone .... 40-48% Zone A (sand) 25-350/0 Zone B ( Sandy Clayl Clayey Soil) Table 1. Recharge value incorporated in the model MODEL CALIBRATION After assigning all the required input parameters to the model, the calibration was carried out. The calibration strategy was initially vary the best known parameters as 4.0 RMS 0.961 little as possible, and vary the poorly known or 3.5 unknown values the most to achieve the best 3.0 overall agreement between simulated and 1.5 Computed observed datasets. Transient state model to calibration was carried out to minimize the 1.5 difference between the computed and field 1.0 water level conditions, with the water level data of January 2000 in 28 wells distributed 10 15 30 over the study area. Out of all the input Observed parameters, the hydraulic conductivity and specific yield are the poorly known. Based on Figure 3. Computed and observed the data, it was decided to vary hydraulic head in transient state calibration conductivity, specific yield values upto 10% to get a good match of the computed and observed heads (Figure 3). Table 2 shows the initial and calibrated hydraulic conductivity values of the simulated head. The root mean square error and the mean error were minimized through numerous trial runs. Transient - state simulation was carried out for a period of 3 years from January 2000 to December 2003, with monthly stress periods and 24 hr time steps. The trial and error process by which calibration of the transient model was achieved included several trials until a good match was achieved between computed and observed heads over space and time. SIMULATION RESULTS Computed Vs observed
Geology of the area
Hydraulic conductivity (m/day) Initial Calibrated 75 81 25 21
Specific yield
The calibrated model was used to simulate the regional groundwater head and it was Initial Calibrated compared with the observed data of 28 Sand 0.22 0.31 wells. The predicted regional head 0.23 Clayey 0.17 generally follows the observed regional sand groundwater head. The model was Table 2. Initial and calibrated hydraulic simulated in transient condition for a time Conductivity and Specific yield of the simulated interval of 3 years from January 2000 to head December 2003. The regional groundwater head for the month of January 2000 is shown in Figure 5. There is a fairly good agreement between the computed and observed heads. The result of the numerical simulation indicates that there is a very gradual decline in the groundwater level in the some of
509
MODFLOW and More 2006: Managing Ground-Water Systems - Conference Proceedings~ Poeter, Hill, & Zheng - www.mines.edu/igwmcl
the wells. A comparison between the observed and computed head values for the observation well no 8
and is shown in Figure 6. The computed head values mimic the observed head values. There is a very
very minimum decline in groundwater head over three years.
Well No: 8 3.5
u! 3 E ca 2.5
g
2 CD 1.5 ~ 1
-+- Observed Computed
-I
~ • •
Ie Ie
•
Ie
0.5
co 0 ;:
+-r-r-r-r-r.......-r-T......-r-r-r-r-1:-r-r-1,.,..,....,r-r-r-T""'T'""""T"""T--r-rT"'"T"""T"--r--T"T1I
1
4
7 10 13 16 19 22 25 28 31 34 Time in Months
Figure 5. Computed regional groundwater head (Jan 2000)
Figure 6. Computed and observed water head in well no 8.
CONCLUSION A regional flow model was developed for the south Chennai sandy aquifer wherein the groundwater head over space and time were simulated. The computed groundwater head over space and time matches reasonably with the observed groundwater head with a difference of 0.8 m. Sensitivity analysis revealed that the developed model is very sensitive to recharge and, to some extent, hydraulic conductivity and specific yield. The model simulation indicates that this aquifer can be sustainably managed with the present pumping rate of 4.25 million gallons per day. REFERENCES Corbet and Bethke., 1992. Disequilibrium fluid pressures and groundwater flow in western Canada sedimentary basin. J Geophys Res. 97(85), 7203-7217. Gnanasundar, D., and Elango, L., 2000. Groundwater flow modelling of a coastal aquifer near Chennai city, India. Journal of Indian water resources society. 20 (4),162-171. Rushton.K.R., and Redshaw.S.C., 1979. Seepage and groundwater flow. John Wiley and sons Ltd. NY
330. Storm, E.W., and Mallory, M.J., 1995. Hydrogeology and simulation of groundwater flow in the Eutaw Mcshan aquifer and in the Tuscaloosa aquifer system in northeastern Mississippi: U.S Geological Survey Water- Resources Investigations Report 94-4223, 83. Senthilkumar, M.. , and Elango, L., 2004. 'Three- dimensional mathematical model to simulate groundwater flow in the lower Palar River basin, southern India' Hydrogeology Journal, 12(4), 197-208.
