Characterization of Potential Direct Recharge in the Gold Mining District of South-Western Ghana Using the Hydrus-1 D Computer Code

The mean annual potential recharge rate to the ground water system in the Tarkwa gold mining district of South Western Ghana, has been estimated for the period 1977 to 2001 by analyzing precipitation recharge in response to rainfall pattern and distribution for the four main soil types; Huni, Tarkwa Phyllite, Banket and Kawere. Using the Hydrus-1D infiltration computer code model, simulation results of recharge-precipitation exhibited a linear relationship, and gave correlation coefficient values R2 ranging from 0.39 to 0.55 for the four soil types and 0.72 for the composite soil. Variability of annual recharge was assigned to various reasons, including the variation in the weather pattern (rainfall, temperature, evapotranspiration etc), topography and hydraulic properties of the aquifer system. The potential direct recharge rate was estimated to range from 269 mm/yr to 611 mm/yr, with an average value of 385 mm/yr. This is as a result of infiltration of about 18% to 36% (average 27%) of the mean annual precipitation. The high average value of 385 mm/yr is reasonable on the basis that it represents the potential maximum value of the actual recharge value of 299±72 mm/yr.


Introduction
Gold dominates the mining sector in Ghana, with prospective gold deposits localized in three regions namely; Western, Ashanti and Brong-Ahafo.The Western and Ashanti regions have varying years and indelible experiences of gold mining, whilst communities in the Brong-Ahafo region are yet to experience the full scale negative effects of surface mining.The Tarkwa area in the Wassa West District of South Western Ghana, where this study was conducted, is said to have the single largest concentration of mines and mining companies on the African continent, with about one-third of the total land area under concession for mining (Akabzaa & Darimani, 2001).Siegel (1997) has noted that, developing a mine-plan of operations, operational procedures, hydrological control structures and other best management practices to prevent negative environmental impacts require accurate knowledge of the variables associated with hydrological conditions at the mine.The most important being the proper characterization of baseline hydrogeological conditions, with groundwater recharge being paramount, so that the extent of impacts to hydrologic and other related resources can be minimized or avoided.Pressure on groundwater resources resulting from mine dewartering has been identified among others to be the single major mining impact in the study area and this has necessitated the need to provide more accurate recharge estimates for various tools that can assess the sustainability of long-term water use.
In this work, we have used the Hydrus-1D computer software package (Šimunek et al., 2005) to estimate the mean annual potential recharge rate by analyzing the variability of precipitation recharge for the various soil types (Huni, Tarkwa Phyllite, Banket and Kawere) in the unsaturated zone, in response to rainfall and evapotranspiration.This software was chosen because of its advantage of dealing with hydrological complexities and the derivation and estimation of water budgets and fluxes (Hernandez et al., 2003;Simunek & van Genuchten, 1999;Shah et al., 2007).Neuman et al. (1974) observed that most studies that use unsaturated zone models to estimate groundwater recharge normally assumed the unit gradient flow at the lower boundary, with the bottom water flux taken to be equal to the potential groundwater recharge (Nolan et al., 2003(Nolan et al., , 2007;;Small, 2005;& Keese et al., 2005).Potential recharge is defined as the likely recharge in cases where the water table is deep enough, such that it no longer has influence on the recharge rate.This is interpreted to mean that the recharge process can be estimated assuming that the flow in the unsaturated zone is freely draining.This rate is effectively a maximum rate, as in some cases the inability of groundwater to flow away fast enough will mean that recharge will be rejected.
Previous studies (Kuma & Younger, 2000;Kortatsi, 2004) contend that vertical groundwater recharge from precipitation and evapotranspiration are the main components of the water cycle in the study area, with the unsaturated zone playing an important role.Using the water budget method, Kuma (2007) estimated a recharge rate of (299±72) mm/yr for the study area.According to these researchers, the humid climate of Tarkwa district is characterized by a surplus of precipitation over soil evaporation and plant transpiration, with no distinct monsoon.As a result, precipitation is the primary source of recharge, whereas seepage from other surface bodies, watercourses, terrain depressions, fractures, and diversion from denser soil contribute indirectly with a trivial volume, especially in the environments of new surface mine development.Once infiltration is reduced by evapotranspiration, the rest of the moisture percolates down through the vadose zone to the water table as a potential recharge to the groundwater.The shallow areas, however, allow for some moisture to be driven back by a capillary rise in response to evapotranspiration demand

Hydrus-1D computer Code
The vertical movement of water in the unsaturated zone of HYDRUS-1D Computer code (Šimunek et al., 2005) is described by the Richard's equation whose one-dimensional mixed form is given in equation (1) as: This is subject to the initial and boundary conditions given in equation (2) as: where the parameters are defined by van Genuchten & Mualem, (1980).
The relationships between water retention and pressure head as expressed in equation ( 3) have five parameters that define the shape of the function; θs saturated porosity of the soil, θr residual moisture content, Ks saturated hydraulic conductivity, α inverse of the air entry value (or bubbling potential), and n pore size distribution index.
The required input data includes model parameters, geometry of the system, boundary conditions and initial conditions for transient flow simulation.The dimensions of the problem domain are defined via the geometry parameters while the physical properties of the system under consideration are described via the physical parameters.The unsaturated zone includes the soil water characteristic, θ(h) and hydraulic conductivity, K(h).

Study Area Description
Recharge is a region-specific property and therefore the study was concentrated at the Tarkwa mining district of South Western Ghana.The study area is located in the Ankobra Basin, which is bordered to the west by the Ankobra River that flows towards the south, and the north and south by the Huni and Bonsa Rivers, respectively.The area is highly dissected and has moderate relief with a general decrease in elevation southwards (Kuma, 2007).The main types of soil in the area are the forest oxysols in the south and the forest ochrosol-oxysol in the north (Asklund & Eldvall, 2005), and consists mostly of silty-sands with minor patches of laterite (mainly on the hilly areas) and underlain by the Banket Series and Tarkwa Phyllite rocks (Kuma & Younger, 2001;Kortatsi, 2004).

Numerical Model Design Using Hydrus-1D Computer Code
The annual potential recharge rate was forecasted through the modelling of the soil system.The basic approach was to perform numerical simulation on a conceptual one-dimensional column from the surface to the water table by using the following data.

Meteorological data
The meteorological data from 1977 to 2001 of South Western Ghana used for this study was obtained from the Ghana Meteorological Service weather station at Tarkwa.This was statistically processed and analyzed, and appropriate graphs presented as shown in Figure 1    Soil properties representing the four main soil types (Huni, Tarkwa Phyllite, Banket and Kawere) were determined for each of the profile components in describing the soil profile using the results of infiltration tests technique used by Kuma and Younger (2001).The soil hydraulic parameters, as presented in Table 1, were estimated using the pedo-transfer function model, Rosetta (Schaap et al., 2001).This model predicts the hydraulic parameters from soil texture and related data as implemented in Hydrus-1D (Šimunek et al., 2005).The van Genuchten parameters were estimated using Rosetta (Schaap et al., 2001), based on the soils data collected by Carsel and Parrish (1988).

Setting the Initial and Boundary Conditions
In this study the threshold value of pressure head, hcrit, was assumed to be minus 1000000 mm.However, because the surface soil remained relatively wet due to regular precipitation and remained above the hcrit threshold, the simulation results obtained were insensitive to this parameter value when specified in the range minus mm to minus 1000000 mm as observed in a previous study (Simunek et al., 2005).
The upper boundary condition was specified as an atmospheric boundary condition with surface runoff.Implementation of the atmospheric boundary condition requires specifying daily precipitation and evapotranspiration rates, therefore, a twenty-five year daily time steps series (01/01/1977 to 31/12/2001) of precipitation and potential evapotranspiration was defined in excel format, as presented for the first month of the first year in Table 2.This was repeated for the rest of the years and were input into the Hydrus-1D computer code (Šimunek et al., 2005).The time series data of precipitation and evaporation was introduced to the surface of the model grid system as a source term.When the top layer becomes saturated as a result of heavy downpour, precipitation rate exceeds infiltration rate.Precipitation then begins to accumulate in the surface and the ponded water is assumed to be immediately removed by runoff whiles the model automatically converts the pressure head of the top layer to zero.Then, the bottom water flux was assumed to be equal to the potential groundwater recharge which was calculated for the various soil types as indicated in Table 3.
In order to get the resulting solutions to produce the same initial cyclic pattern, the initial conditions of the model have to be generated by running a certain period of preliminary cyclic simulation earlier to the time of interest.A three-year preliminary simulations were therefore performed by a multiple-year precipitation and evapotranspiration sequence having a cyclic pattern of the first-year (Y1) meteo sequence such that Y2=Y3.Figure 2 shows the temporal changes of water fluxes at the bottom of the lower boundary showing the cyclic pattern of the peaks representing Y1, Y1, Y1, Y2, Y3, Y4, Y5.

Water Infiltrations in the Unsaturated Zone
The bottom water flux of the infiltration model was taken to be equal to the potential groundwater recharge.Figure 3 is a composite graph showing breakthrough plots of the elevation, pressure and the total head in order to check whether the system was behaving properly, and in particular, whether the flow had reached the base of the model.The first 180 days was taken by the system's initial conditions to adjust to the temporal changes.The system became stable by 635 days, i.e. within the three years, year 1 cycling, with infiltration reaching the water table.Table 3 shows the data summary of the entire results of the potential recharge estimates to the ground water system of the various soil types for the period 1977 to 2001.The annual percentage contribution of precipitation to recharge was also calculated.A preliminary investigation into the relationship between precipitation and recharge indicated that there was significant variation in the data and might have resulted from a number of reasons, most notably, different antecedent conditions.Depending on the soil types, the average potential annual recharge rates ranged from 355 mm/yr to 411 mm/yr, corresponding to 25.4% to 29.4% of the annual precipitation for the entire period of 25 years, respectively.Figures 5 and 6 are double y-axis graphs (combo graphs) that show the effect of rainfall pattern and distribution on the potential infiltration recharge over the period 1977 to 2001, for the entire study area and for the four main soil types, respectively.Although, there is variation in precipitation recharge to the groundwater system in the study area, generally, the plots look similar, which is an indication that there isn't much difference in precipitation recharge for the soil types.High recharge rates were observed from 1977 to 1984 and remained almost in a dynamic equilibrium from 1984 to 2001, with the exception of 1998, where a minimum rate was observed over the entire period.This is due to the corresponding variations in rainfall pattern and distribution in the area.Overall, the results suggest a strong relationship between the annual rainfall and observed recharge.To further investigate this relationship an attempt was made to relate the dependent parameter (recharge) to the independent parameter using a linear regression analysis expressed in equation ( 4) as: where R is the annual recharge rate (mm), P is the annual precipitation (mm), and m and C are calibration coefficients.
Figure 7 is a composite graph showing the linear variation between the annual recharge and precipitation for the four main soil types in the area, and that of the study area.The calculated R 2 values, which are the standard coefficient of determination or correlation, indicate the strength of the relationships.Results of the linear regression analysis are tabulated in Table 4 for the entire study area and the four soil types.
The results show R 2 value of 0.7245 for the study area, while that for the four soil types ranged from 0.3874 to 0.5484.The negative intercept value c indicates the existence of a threshold rainfall needed for recharge to occur.This threshold is quite large, ranging between -222 and -470, and therefore making %P less meaningless when taken into consideration the low correlation between recharge and precipitation.The high value of c may be that the relationship is not really linear, and therefore no need to extrapolate to P values below the lowest, as seen in the data set.

Conclusions
Using the Hydrus-1D Computer code infiltration model, the mean potential annual recharge to the groundwater system in the Tarkwa District of South Western Ghana has been estimated for the period 1977 to 2001.The recharge estimation was conducted by analyzing precipitation recharge for the four main soil types Huni, Tarkwa Phyllite, Banket and Kawere; in response to rainfall pattern and distribution within the study area.Simulation results establishing a recharge-precipitation relation, when a linear relationship was used, gave correlation coefficient values R 2 ranging from 0.39 and 0.55 for the four soil types and 0.72 for the entire study area.The variability of annual recharge in the area may be assigned to various reasons, including the variation in weather pattern (e.g.rainfall, temperature, and evapotranspiration), topography and hydraulic properties of the aquifer system.The potential direct recharge rate was estimated to range from 269 mm/yr to 611 mm/yr, with an average value of 385 mm/yr.This is as a result of infiltration of about 18% to 36% (average 27%) of the mean annual precipitation.The high average value of 385 mm/yr is reasonable on the basis that it represents the potential maximum value of the actual recharge value of 299±72 mm/yr as reported by Kuma (2007).
(a composite graph showing the daily and monthly variations in precipitation and evapotranspiration).The analyzed data was used as input for the Hydrus-1D Computer code (Šimunek et al., 2005) and the modelling of bottom fluxes in the estimation of the potential recharge.

Figure 1 .
Figure 1.Precipitation and Potential Evapotranspiration rates of South Western Ghana averaged over the period 1977-2001

Figure 2 .
Figure 2. Temporal changes of water fluxes at the bottom of the lower boundary

Figure 4
Figure 4 shows plots of temporal changes in actual and cumulative water fluxes and pressure heads across the upper and the lower boundaries of the infiltration model.The cumulative bottom flux was used in the calculations of the potential recharge estimates to the groundwater system for the period 1977 to 2001.

Figure 3 .
Figure 3. Water infiltrations in the unsaturated zone showing the elevation head, pressure head and the total head for the periods 180 days, 635 days and 1096 days

Figure 4 .
Figure 4. Graphs of temporal changes in actual and cumulative water fluxes and pressure heads across the upper and lower boundaries

Figure 5 .
Figure 5. Data summary of mean annual precipitation, percentage of precipitation and groundwater recharge in Tarkwa District of South Western Ghana

Figure 7 .
Figure 7. Variations of mean annual recharge with precipitation rate for the four main soil types and their composite, showing the R 2 values

Table 3 .
Predicted potential recharge values of Tarkwa District in South Western Ghana for a range of rainfall distribution spanning for the period

Table 4 .
Linear regression analysis of recharge-precipitation relationships of the study area for the period, 1977-