Predictive simulation of flow and solute transport for managing the coastal aquifer of the Dakshina Kannada district, Karnataka, India

Abstract

The present investigation is intended to simulate the response of an unconfined, shallow, tropical coastal aquifer to anticipated future stress scenarios due to developmental activities and climate change effect. The simulation of groundwater flow and solute transport are carried out using SEAWAT. The model is applied to the coastal basin of Dakshina Kannada district of Karnataka state, India having an areal extent of about 155 km2 in. The study area is divided into four sub-basins for the simulation considering the natural boundaries. They are the basins between Shambhavi river and Pavanje river (sub-basin 1), Pavanje river and Gurpur river (subbasin 2), Gurpur river and Netravathi river (sub-basin 3), Netravathi river and Talapady river (sub-basin 4). It is learnt from the field investigations that, the basin is predominantly an unconfined aquifer with depth ranging from 12m to 30m. The region is mainly covered by the lateritic formation below the top soil. The aquifer profiling was plotted based on the vertical electrical sounding. The aquifer parameters are estimated based on pumping test results. Nine aquifer hydraulic parameter zones are mapped for each basin based on pumping tests evaluations. The transmissivity values range from 10 to 1440 m 2/day. The numerical simulation of groundwater flow was carried out by building a MODFLOW model to the basin and the transport parameters are assigned to execute the MT3DMS model. Finally, the SEAWAT model which is a coupled version of MODFLOW and MT3DMS designed to simulate three-dimensional, variable density groundwater flow and multi-species transport is developed. The model of each subbasin has two dimensional grids in the horizontal plane with an approximate cell dimension of 100×100m with a single vertical layer. The digital elevation model (DEM) developed for the study area is interpolated to the top elevation of the model grid. The base of the model layer is set at -30m (with respect to mean sea level), which corresponds to the base of the shallow unconfined aquifer. The recharge is assigned on the upper-most active (wet) layer of the model during the monsoonii season (June to September). A total of 587, 730, 835 and 996 wells are introduced in the agricultural area of sub-basin 1, 2 , 3 and 4 respectively, based on the village wise installation of irrigation pump set data. The draft per well is assigned based on the water requirement of crops, i.e. evapo-transpiration in the absence of actual data. A constant concentration of TDS 35kg/m3 is specified to the model cells along the western boundary (Arabian sea). For rivers, TDS = 35 kg/m3and 17.5kg/m3 are considered during the non-monsoon (October to May) and monsoon (June to September) respectively considering the quantum of mixing of freshwater and seawater. The model is calibrated from September 2011 to August 2013 using observed groundwater heads and salinity data obtained from 29 observation wells. In the present study, PEST is used to calibrate the model. The total simulation period of two years has been divided into 24 stress periods. Daily time step has been considered for the transient simulation applying all the hydro-geologic conditions of the same period. The model is validated for the following year (2013-14). Both the flow and transport model performance during the monsoon (June to Sept) is not up to the mark, with all the three evaluation techniques (R2, RMSE and NSE) showing deviation from the desired levels. Overall, the model performance is satisfactory with NSE ≥ 0.5. The calibrated values of horizontal hydraulic conductivity and specific yield of the unconfined aquifer range from 1.85m/day to 49.50 m/day and 0.006 to 0.281. These values agree with the range established by the aquifer characterization studies carried out earlier. Also, recharge co-efficient of 20% of rainfall, porosity of 30% and river bed conductance of 10 m/day are obtained as appropriate parameters during the calibration process. The longitudinal dispersivity of 35m, transverse dispersivity of 3.5m and molecular diffusion co-efficient of 8.64 × 10-5m2/day are achieved. The spatial distribution map of groundwater table shows a gradually increasing trend from the coastline and the rivers towards the landward side (high elevated area). The water table rises to maximum elevation of about 43m (above msl) in sub-basin 1 and 3. The month of May is visibly drier than the month of August, with the lowest groundwater level contour moving towards inland by about 200m to 900m in comparison with that during the monsoon. Water balance study shows that, more thaniii 75% of available water is being discharged to the sea during the wet season compared to that during the dry season throughout the coastline. The river Gurupur contributes hugely to the aquifer throughout the calibration period. This is due to the fact that the area surrounding the river is a low lying marshy land. The management of freshwater aquifers within 1 km from the sea is of prime importance for sustainability against seawater intrusion. The salinity distribution across the study area for sub-basins 1, 2, 3 and 4 during the month of May 2013 shows a similar pattern in all the four sub-basins. An important outcome is that, the rivers that surround the system on the north and the south sides contribute equally as that of the sea in bringing in salinity into the aquifer. The TDS values are within 0.5 kg/m3 throughout the year, except that for well nos. 1, 15 and 25. It is essential to note that, all these wells are very close to the rivers (less than 300 m). The sensitivity analysis results clearly show that, the overall aquifer system is sensitive to hydraulic conductivity, groundwater draft and recharge rate. The model is sensitive to lower values of hydraulic conductivity (0.46 m/day to 12.40 m/day) and higher values of recharge rate (28 mm/day). The results also show that, the lateral movement of water from the river causes the adjoining area to respond differently to changes in the parameters than away from it. No significant influence of river bed conductance on the water table elevation was noticed over in the entire area except that in sub-basin 1 and zones adjacent to the river flow. The aquifer was found to be least sensitive to dispersivity, with the movement of salinity contour by just 10m for every increase in 25% of the dispersivity value. The area under consideration is recently experiencing exponential growth in terms of urbanization, industrialization and other developmental activities. Hence, in order to understand the response of the coastal unconfined aquifer to varied overdraft and recharge scenarios, the SEAWAT is used to simulate over a considerably longer period of 20 years (2014 to 2034). The fresh water drafts considered are symbolic in nature and in this work, only electrically operated pumping units are accounted since data on other wells are not available. Also, to account for soils with low permeability, and decrease in rainfall, effect of decrease in recharge rate is also investigated. Accordingly, five scenarios are planned for investigation. Scenario 1 representsiv existing abstraction rate, calibrated recharge rate and no sea level rise. Scenario 2 considers decrease in recharge rate by 50% with other parameters same as scenario 1. Scenario 3 simulates effect of varied freshwater draft of 50%, 100% and 150% of existing draft rate for the wells (case 1, 2 and 3 respectively). Scenario 4, is a combined case of scenarios 2 and 3. And finally, scenario 5 is scenario 4 with sea level rise of 1mm/year. When all the scenarios are compared, the water table is estimated to fall by 0.3m to 0.6 m compared to scenario 1. The study shows that, the decrease in recharge rate (scenario 2) alone can raise the TDS to 5kg/m3 in the first 8 years of simulation. However, with the present rate of groundwater utilization and recharge rate, the aquifer can be considered safe for the next 16 years with TDS < 1.5 kg/m3. Every 50% increase in groundwater utilization causes the salinity to increase steeply with every year of simulation till the end. Hence, except scenario 1 and scenario 3 (case 1), the remaining scenarios lead to the salinity above the drinking water standards (TDS > 1.5 kg/m3), by 6 years of operation. To study the spatial effect on water table and advancement of salinity into the aquifer from the coastline, well hydrographs and salinity at every 200m distance from the coastline are investigated. As per the analysis, the aquifer beyond 200m from the sea is safe (TDS<1.5kg/m3) against seawater intrusion for scenarios 1, 2 and 3. Only due to scenario 4 (case 2 and 3), the seawater intrudes beyond 600 m up to 1200m making the aquifer unsafe for utilization. The percentage area affected by seawater intrusion due to different anticipated scenarios are estimated. Scenario 4 (case 3) is considered to be the most unfavourable condition, with water quality becoming unfit for drinking purpose over more than 35% area (TDS>1.5kg/m3). However, with the present stress conditions continuing for the next 20 years, less than 10% of the total area is predicted to be with TDS > 1.5 kg/m3. But, overdraft by three times the present rate i.e. scenario 3 (case 3) may increase the salinity beyond 10% for sub-basins 1 and 2. In the case of sub-basin 3, the water table falls by about 1.5m for every 50% increase in the groundwater utilization rate, which is less than 1 m for the rest of the basins. The simulation results for this basin show that, the wells within 500m from the sea and rivers are highly saline with TDS>3kg/m3 which was also confirmed with fieldv observations. In addition, simulations are carried out to estimate the effect of climate change on seawater intrusion by considering anticipated sea level rise. The anticipated sea level rise of 1mm/year along the coastline has negligible influence on groundwater and salinity of the study area.