Groundwater arsenic in the Red River delta, Vietnam ... - Fiva

More documents

Recommendations

Info

Table 2 Total As and stable isotope compositions in samples from the Pleistocene aquifer Sample date Depth (m) Astotal (lg/L) d 18 O d 2 H T1 12/04/2006 66 106 10.82 67.67 T2 12/04/2006 66 172 10.06 66.82 K11 11/27/2006 41 190 10.63 66.20 K50 12/01/2006 44 294 10.73 67.61 from the Holocene aquifer with d 18 O values between 11‰ and 10‰ and d 2 H values between 67‰ and 66‰ (Table 2). The higher depletion with respect to heavy O and H isotopes of the water in the deeper, Pleistocene aquifer most probably reflects the influence of recharge water precipitated at higher elevations in the mountains around the flood plain. 4.6. Groundwater flow modelling To quantify recharge and surface–groundwater interaction, a transient numerical groundwater model was set up for the model area depicted in Fig. 1, using the hydraulic data listed in Table 1. Towards the north, the model uses a general head condition as boundary condition, based on linearly interpolated daily measurements at the stations in Sontay and Hanoi. The eastern and western boundaries of the model area are considered as no flow boundaries. The seasonally changing flow conditions in the Holocene aquifer raises the question of the appropriate position of the southern flow boundary. This boundary should be placed far enough from the southern channel to maintain north-moving flow throughout the year. In the southern part of the model domain the saturated total thickness of the Pleistocene and Holocene aquifers is approximately 45 m (Fig. 4) and the boundary length is 5000 m (Fig. 1). Head measurements in boreholes C18, C19, C20 and C21, located approximately 500 m south of the southern channel (Fig. 1), show that during 11 months of the year, the water table has a slope towards the north with a hydraulic gradient between 0.5‰ and 1‰. Using an average hydraulic conductivity of 3.5 10 4 m/s for the aquifer sand, the estimated annual flux across the southern boundary is between 1.2 and 2.4 million m 3 . During high water stands in the Red River and the channels (middle of June to middle of July), the hydraulic gradient at boreholes C18–C21 is reversed and the groundwater flows towards the south. Southward groundwater flow during this reversed flow period amounts to approximately 5% of the northward groundwater flow during the remaining 11 months of the year. Based on a regional catchment model, Chambon (2007) estimated a groundwater flow of approximately the same magnitude along the south-southwest facing boundary. An analytical solution for a confined aquifer with an initial gradient of 1‰ towards the north, solved using the code of Barlow and Moench (1998), shows that the observed change in the water level of the southern channel should reverse the hydraulic gradient towards the south to a distance of up to 1500 m from the channel. Therefore in the numerical model the southern boundary is placed at 2500 m from the southern channel. At this F. Larsen et al. / Applied Geochemistry 23 (2008) 3099–3115 3109 boundary a variable flux of 5000 m 3 /d was used, except in August when the hydraulic gradient is lower and the flux was reduced to 1250 m 3 /d. The channels running through the field site were modeled using the MODFLOW River Package. Transient channel stages, required as model input, were compiled from (i) hand measurements in the channels, (ii) the Red River stage time series and (iii) daily pressure transducer recordings from the rainy season in 2007 in the northern channel. During the dry season, the channel stages are above the stage of the Red River (i.e., a perched water table), whilst in the wet season, the water from the Red River main channel flows through the channel distributaries and the stage variations become identical. This head dependent surface water–groundwater interaction is built into the model. As described in Section 4.5 recharge from percolation occurs mainly during the rainy season, and therefore in the model recharge could be confined to these months. To be able to reflect the variations in the rainfall during the monsoon, an input time series for recharge to the aquifer was calculated as a fraction of the daily precipitation record from the Sontay station. The recharge/precipitation fraction was calculated using a 10-a record, in order to comprise the influence of year-to-year variations in precipitation on the recharge rate. As the model aimed at establishing an overall water balance for the study site, the recharge rate time series was applied uniformly over the model domain. Conductance constants for the bottom sediment of the Red River and the channels were set to 4.0 and 0.5 m 2 /day, respectively. The higher conductance of the Red River reflects that the bottom sediments of the river are more coarse grained than the channel sediments (Fig. 4). The specific yield of the sand is 0.2 (Table 1) and is set to 0.015 in the confining clay. In order to simulate the dynamics in head and fluxes satisfactorily, short stress periods of one to two days were used along with the daily input data for recharge, Red River and channel stages in the rain season, whereas the dry season stress periods were up to 31 days. The model calibration was done using observation data from 15 May 2005 to 14 May 2006 and validated against observation data from 15 May 2006 to 31 December 2006 (i.e., the rain season of 2006). In the calibration process, the parameters, (i) flow across the southern boundary; (ii) the recharge rate; (iii) the conductance of Red River and channels; and (iv) the specific yield of the confining clay-rich layer, were changed manually within reasonable values to improve the match between observed and simulated values. Sensitivity analysis on these parameters was done by reducing and increasing the calibrated values by a factor of 2. The model was relatively insensitive to the value of the conductance of the Red River bed, with head off-sets of approximately ±0.05 m relative to the calibrated model. The value of the specific yield of the confining clay had a high impact on the simulation of the hydraulic head peaks during the surface water stage extremes. Poor fits were obtained for the rainy season if this value was outside the range 0.010 to 0.025. The highest sensitivities, with head off-sets on the order of ±1 m, were found for the recharge,
3110 F. Larsen et al. / Applied Geochemistry 23 (2008) 3099–3115 and the channel bed conductance, and not surprisingly the flow across the southern boundary. The latter model uncertainty was reduced by carefully sampling the head observations along this boundary and the uncertainty has therefore been confined to the estimated hydraulic conductivity of the sand, which seems well documented. The estimated conductance of the channel appears high and could very well be wrong by a factor of 5–10. Mean error (ME) values between simulated and observed hydraulic heads, using a uniform recharge rate of 60 mm/a, is 0.57 m in the calibrated model and 0.60 m in the validated model. Increasing the recharge rate gave larger errors and with a recharge rate of 100 mm/a, the ME values of the calibrated and validated model increased to 0.77 m and 0.80 m, respectively. The mean absolute errors (MAE) in the validated model with a recharge of 60 and 100 mm/a were 0.78 and 0.93 m, respectively. Given the above mentioned uncertainties in the flux across the southern boundary and the conductance of the channels, the uncertainty on the calibrated recharge rate alone from percolation could work equally well with a percolation of 100 mm/a. Increasing the recharge rate to much more than 100 mm/a is not possible with the calibrated flow model without simulating unrealistically high hydraulic heads. The observed and simulated hydraulic heads in boreholes C2, C5, C9 and C19 are depicted in Fig. 10 and show that when the water table rises, the simulated hydraulic heads match the observed values very closely, generally with differences smaller than 0.1 m. The larger deviations between observed and simulated heads during peaks are due to variations in the specific yield of the confining clay layer. When the water table falls, the model generally simulates the high head values correctly in the beginning but later the modeled water table becomes too low, with deviations up to 0.6 m. This mismatch is probably caused by a Fig. 10. (a–d) Observed and simulated groundwater heads in boreholes. (e) Observed and simulated hydraulic gradients between the K 50 and H 50 boreholes located within a distance of 750 m in the two transects.
Page 1 and 2:
Groundwater arsenic in the Red Rive
Page 3 and 4:
Søren Jessen Groundwater arsenic i
Page 5 and 6:
organic matter may prevent As mobil
Page 8 and 9:
Resumé Naturligt frigivet arsen (A
Page 10 and 11:
efterfølgende forsøgt at opstille
Page 12 and 13:
Preface This PhD thesis deals with
Page 14:
Table of contents 1 INTRODUCTION: T
Page 17 and 18:
The PhD research presented in this
Page 19 and 20:
educed form as As(III), which is al
Page 21 and 22:
A more complex and intimate relatio
Page 23 and 24:
sedimentary As content (Postma et a
Page 25 and 26:
(Anawar and Mihaljevi, 2009), and g
Page 28 and 29:
3 Regional As distribution in the R
Page 30 and 31:
in large part by changes in the eus
Page 32:
In the Red River delta, the Pleisto
Page 35 and 36:
the Red River delta, which are not
Page 38 and 39:
5 The mitigation potential of bank
Page 40:
Noteworthy is, that the ratio of Fe
Page 43 and 44: trend in the distribution of the RF
Page 45 and 46: 2008) or the Hach or Merck test kit
Page 47 and 48: Badloe C., Nguyen T. P. T. and Nguy
Page 49 and 50: Driehaus W., Seith R. and Jekel M.
Page 51 and 52: Islam F. S., Boothman C., Gault A.
Page 53 and 54: Mondal D. and Polya D. A. (2008) Ri
Page 55 and 56: A. Ali and Z. Adeel). ITN Centre, B
Page 57 and 58: Ahmed K. M. (2006) A transect of gr
Page 60: Arsenic in groundwater of the Red R
Page 63 and 64: In short, there is a consensus amon
Page 65 and 66: The local geology shows mainly sand
Page 67 and 68: ane concentration and P CO2 values
Page 69 and 70: Fig. 7. The distribution of arsenic
Page 71 and 72: H3AsO3 þ H2O $ H2AsO4 þ 3H þ þ
Page 73 and 74: Fig. 12. The relative composition o
Page 75 and 76: of the Fe-oxides in the upper 3 m c
Page 77 and 78: found that while As(V) coprecipitat
Page 79 and 80: desh: further evidence of decouplin
Page 82 and 83: Controlling geological and hydrogeo
Page 84 and 85: Here the results of the geological
Page 86 and 87: sediments. In general, the coarser
Page 88 and 89: Fig. 5. Thickness of the clay-rich
Page 90 and 91: a b c The groundwater head distribu
Page 94 and 95: delayed yield from the fine grained
Page 96 and 97: that the situation in Dan Phuong co
Page 98: Mathers, S., Zalasiewicz, J., 1999.
Page 102 and 103: 0883-2927/$ - see front matter © 2
Page 104 and 105: 3118 S. Jes sen et al. / Applied Ge
Page 112: 3126 S. Jes sen et al. / Applied Ge
Page 116 and 117: Postma et al., submitted to Geochim
Page 142 and 143:
Postma et al., submitted to Geochim
Page 144 and 145:
Page 146 and 147:
Page 148 and 149:
Page 150 and 151:
Page 152 and 153:
Page 154 and 155:
Page 156 and 157:
Page 158 and 159:
Page 160 and 161:
Page 162:
The mobility of arsenic in a Red Ri
Page 165 and 166:
Jessen et al., manuscript to be sub
Page 167 and 168:
Page 169 and 170:
Page 171 and 172:
Page 173 and 174:
Page 175 and 176:
Page 177 and 178:
Page 179 and 180:
Page 181 and 182:
Page 183 and 184:
Page 185 and 186:
Page 187 and 188:
Page 189 and 190:
Page 191 and 192:
Page 193 and 194:
Page 195 and 196:
Page 197 and 198:
Page 199 and 200:
Page 201 and 202:
Page 203 and 204:
Page 205 and 206:
Page 207 and 208:
Table 1
Page 209 and 210:
Fig. 1
Page 211 and 212:
Fig. 3
Page 213 and 214:
Fig. 5
Page 215 and 216:
Fig. 7
Page 217 and 218:
Fig. 9
Page 219 and 220:
Fig. 11
show all

Groundwater arsenic in the Red River delta, Vietnam ... - Fiva

Create successful ePaper yourself

Delete template?

Save as template?