OR/14/070 Field data collected

From Earthwise
Jump to: navigation, search
Taylor, R G1, Burgess, W G1, Shamsudduha, M1, Zahid, A2, Lapworth, D J3, Ahmed, K4, Mukheriee, A5 and Nowreen, S6. 2014. Deep groundwater in the Bengal Mega-Delta: new evidence of aquifer hydraulics and the influence of intensive abstraction. British Geological Survey Internal Report, OR/14/070.

1  University College London (UCL), UK;     2  Bangladesh Water Development Board, Bangladesh;     3  British Geological Survey, UK;     4  Dhaka University, Bangladesh;     5  Indian Institute of Technology (Kharagpur), India;     6  Bangladesh University of Engineering & Technology, Dhaka, Bangladesh

New observational field data include:

Groundwater chemistry: samples from multiple depths at 8 locations (Table 1) analysed for inorganic constituents including arsenic (As), environmental isotopes (O and H stable isotopes), and residence-time indicators (CFCs, SF6, 14C — planned).

Groundwater head: hourly groundwater (and barometric) pressure records from multiple depths (except Digha, West Bengal) at 8 locations (Table 1).

Groundwater salinity: large-scale survey conducted by project collaborators (BWDB) provides a new regional delineation of shallow groundwater salinity in the coastal region of Bangladesh (Figure 4).

Figure 4    Contoured salinity in (a) the ‘shallow (first) aquifer’, and (b) the ‘deep (third) aquifer’ in coastal Bangladesh (from Zahid et al. 2013. Distribution of Groundwater Salinity and Its Seasonal Variability in the Coastal Aquifers of Bengal Delta, Book Chapter Nine in Adaptation to the Impact of Climate Change on Socio-economic Conditions of Bangladesh (edited by Zahid et al.), pp.170–193.
Table 1    Overview of high-frequency (hourly) hydraulic monitoring since March/April 2013 and depth specific sampling in 2013 and 2014
Site Depth (m) Monitoring Sampling
Kachua

Bangladesh

25

165
280

head, barometric

head
head

May, 2013
Laksmipur

Bangladesh

91

152
244

head, barometric

head
head

May, 2013
Barisal

Bangladesh

105

151
236

head, barometric

head
head

May, 2013
Kuakata (Kalapara)

Bangladesh

122

180
268

head, barometric

head
head

May, 2013
Khulna

Bangladesh

61

164
323

head, barometric

head
head

March, 2014
Gabura

Bangladesh

67

116
212

head, barometric

head
head

March, 2014
Kamgachi

West Bengal

17

60
100

head, barometric

head
head

March, 2014
Sahishpur

West Bengal

37 head March, 2014
Digha

West Bengal

250

250

head, barometric

head

to be sampled

Preliminary findings

Evidence from nested monitoring stations: depth-specific profiles of groundwater heads and chemistry

Groundwater heads

Groundwater heads were monitored using automatically recording pressure transducers installed in nested piezometers that represent an in-kind project contribution by the Bangladesh Water Development Board. At each site, records have been collected at three depths in vertical profile to maximum depth up to 320 m between May 2013 and June 2014 (Table 1). A selection (one coastal, one inland) of initial results is illustrated in Figure 5.

Figure 5    Groundwater head records in vertical profile and barometric pressure from Gabura (coastal site) and Lakshmipur (inland site) — note raw data require additional corrections for salinity. Depth values (m) in the plot titles indicate the depth below datum for the piezometers. The barometric data is shown as a pale blue line.

Vertical hydraulic gradients consistent with regional gravitational flow in the BAS are over-printed by the effects of intensive pumping, where it occurs. Vertically upward hydraulic gradients reflecting regional groundwater discharge are restricted to extreme coastal locations. Transient variations in deep groundwater head, remote from the effects of pumping, are dominated by the elastic response of the aquifer sediments to surface loading: periodic (tidal) loading in the vicinity of tidal water bodies (Figure 6), and episodic loading by terrestrial surface water during the monsoon season. In most records, episodic deflections of groundwater head in the order of 0.1 m and up to 0.5 m, near simultaneous with depth, are clearly resolvable. These deflections are several times larger than previously published results[1] of passive piezometric monitoring at 300 m depth.

The high susceptibility of the BAS sediments to compaction reduces the barometric effect, which is obscured by the larger scale impacts of groundwater pumping, and surface loading. Inland, where tidal loading is absent, small amplitude earth tide responses are ubiquitous in the background signal. At greater than 100 m depth, we interpret the seasonal recovery and recession in groundwater heads to be a record of annual changes in terrestrial water storage, largely independent of groundwater flow.

These results have important implications for management of the deep groundwater resources. In most hydrogeological environments, groundwater flow models are calibrated against measured groundwater heads. This study shows that for the Bengal Aquifer System, the sensitivity of deep groundwater heads to loading means that deep groundwater levels are not valid for calibration of uncoupled models of transient groundwater flow since rising heads are not indicators of recharge in this environment and falling heads do not indicate aquifer drainage.

Figure 6     Groundwater head recorded in piezometers between 67 and 212 m depth, Gabura (southern Bangladesh), 2012–2013. Note correspondence with monthly tidal cycle consistent with surface mechanical loading (top), and the semi-diurnal tidal periodicity with phase shift between 67 m and 212 m depth (bottom).

Groundwater chemistry

As shown in Figure 7, two anthropogenic tracers CFC-12 and SF6 provide consistent age profiles through the BAS at the eight monitoring sites. The differences in absolute % modern values, for waters with only a small proportion of tracer generally greater than 50 mbgl, are an artefact of the analytical sensitivity and precision of the two methods (see Figure 7 caption for details). SF6 is present in modern recharge at fmol· L-1 concentrations, CFCs are 3 orders of magnitude higher at pmol· L-1 concentrations.

Shallow groundwater, less than 50 mbgl, shows a range of % modern water with 0% at Kachua, 3% at Lakshmipur, 6% at Kuakata and Kamgachhi and >10% at Khulna and Sahishpur. Shallow sites were not available at Barisal but these are likely to be >10% since this was observed at this site in groundwater at 100 mbgl. Overall, slight declines in % modern CFC-12 with depth for groundwater below 50 mbgl (Figure 7a) are detectable at Khulna, Barisal and Lakshmipur with high historical rates of pumping. These trends are not, however, clearly evident in the % modern SF6 results (Figure 7b).

Two public supply sites in Khulna with contrasting historical pumping regimes but in close proximity (<200 m bgl) to the BDWB monitoring site, were sampled at a depth of 290 m (Figure 7) for comparison. There is a noteworthy anomaly for one of these two sites where there is evidence of a significantly higher component of modern recharge (>15%) in both CFC-12 and SF6 results. Although both sites have been used since the early 1980s, the site with a higher component of modern recharge has been pumped on a 16h cycle compared to a 7h cycle for the other site. Tracer differences are corroborated by stable isotope data that show distinct signatures in the two deep sites (Figure 8a). These observations strongly indicate vertical connectivity between the shallower and deeper depths of the BAS at this site that has most likely been induced by enhanced pumping.

Figure 7c reveals a widespread salinity at coastal sites, most notably at depths shallower than 200 mbgl. At some sites (e.g. Kuakata, Kachua), salinity is also a problem at depths of 270 to 280 mbgl where SEC is approaching or in excess of 2000 μS·cm-1. Figure 7d (note log scale on x axis) shows a decline in arsenic with depth from >100 μg·L-1 at shallow sites (<50 mbgl) to <5 μg·L-1 — below the WHO guideline value for drinking-water (10 μg·L-1) — at deeper sites (>100 mbgl). Exceptions occur at Barisal and Kuakata where As concentrations are higher (10 μg·L-1) in the deepest groundwater (>270 mbgl) than at shallower depths between 100 and 250 mbgl (<2 μg·L-1).

Figure 8a shows that for stable isotope ratios of oxygen and hydrogen, there is a different regression line for the deeper (>100 mbgl) groundwaters relative to shallow groundwaters (<100 mbgl). The deep sites (Figure 8) give a regression line of δ2H=8.3δ18O + 10.2 per mil VSMOW, consistent with the global meteoric water line (Craig, 1961); the slight offset is consistent with colder meteoric temperatures compared to present day. Shallow sites have a regression line of δ2H=6.8δ18O + 0.82 per mil VSMOW, either indicative of fractionation processes prior to recharge or of a deviation in the current regional meteoric water line. These data suggest a long-term shift in regional climate and associated meteoric signatures and/or a change in recharge sources or processes. In addition, in some instances the isotope signatures may be used to demonstrate mixing processes and leakage as in the case of the two deep public supply sites in Khulna which are characterised by distinct isotope and % modern recharge signatures. The relevance of these data in identifying recharge sources for deep groundwater will be tested using groundwater flow pathways modelled under different pumping regimes in allied work.[2]

Figure 7    Depth profiles for groundwater residence time tracers a) CFC-12 and b) SF6, data presented as % modern recharge, c) SEC (μs·cm-1) and d) total dissolved arsenic (μg·L-1), note the log scale for the x-axis. Residence time tracer results have been corrected for temperature and excess air. A binary mixing model, assuming mixing between modern recharge and tracer ‘dead’ water, has been used to estimate the % modern recharge. Measurement precision is within ±5% for CFC-12 with a detection limit of 0.05 pmol/L. Measurement precision is within ±10% for SF6 with a detection limit of 0.1 fmol/L.
Figure 8    Cross-plot of δ18O vs δ2H, (a) grouped by site and (b) grouped by SEC (μs·cm-1) categories relevant for drinking water and irrigation standards. Figure 8a has been annotated to illustrate Khulna sites with different depths and pumping regimes. Regression lines for shallow sites (<100 mbgl) shown as a solid line, regression line for deeper sites (>100 mbgl) shown as a dashed line.

Projected intensification of rainfall in Bangladesh

The analysis of General Circulation Model (GCM) projections for the twenty-first century over Bangladesh (Figure 9) suggests an increase in mean rainfall consistent with projected increases in rainfall across the tropics more generally. This result is similar in the analysis of GCMs contributing to the IPCC’s AR4 and AR5. At the broader scale, this finding is part of a wider quasi-global rich-get-richer pattern in which regions of moisture convergence (divergence) are expected to experience increased (decreased) rainfall. It is noteworthy that projected changes to extreme (90th percentile) monthly rainfall driving seasonal groundwater recharge are of greater magnitude than changes projected for mean monthly rainfall. Further experiments within the 5° domain (see Figure 3) demonstrate that these results (i.e. increased rainfall and preferential intensification of extreme rainfall) are robust irrespective of sub-domain location. The projected changes in the higher moments of the rainfall distribution are an important dimension to non-stationarity in future climate and, as shown here, have important implications for groundwater recharge. In areas with suitably permeable surficial sediments, more intensive seasonal rainfall could benefit groundwater recharge if sufficient groundwater storage is made available through dry season abstraction (i.e. ‘Ganges Water Machine’[3]). On-going research is focussed on recharge processes where sub-daily (hourly) time series records of rainfall and groundwater levels have been collated at two sites of contrasting surface geology in central Bangladesh (Savar, Bhuapur).

Figure 9    Projected changes in precipitation over the period 2070–2099 relative to 1961–1990 for the 5° box (Figure 3) in Bangladesh from multi-model ensembles of CMIP3 (AR4) under the A1B emissions scenario (23 GCMs) (left), CMIP5 (AR5) under the RCP8.5 scenario (21 GCMs including 8 new ESMs) (middle), and CMIP5 (AR5) under the RCP8.5 scenario (13 GCMs without new ESMs) (right). Tukey box plots are of changes in monthly precipitation for each model in the MME. Dots indicate individual models within the MME sample, boxes show the inter-quartile range and median and circles show the mean of the MME sample.

Intensive abstraction of shallow groundwater flushes arsenic

Constructed GRMs (Generalised Regression Models) both nationally and regionally within Holocene deposits reveal inverse associations between observed As concentrations and three covariates[4] (Figure 10): (1) net changes in mean recharge between pre-developed and developed groundwater-fed irrigation periods; (2) hydraulic conductivity of the shallow aquifer; and (3) groundwater-fed irrigation trends (1985–1999). GRMs show further that the spatial variation of As concentrations is well explained by not only surface geology but also statistical interactions (i.e. combined effects) between surface geology and mean groundwater recharge, thickness of surficial silt and clay, and well depth. Collectively, these associations are consistent with the assertion that irrigation-induced recharge serves to flush mobile As from shallow groundwater. Our results suggest that shallow groundwater-fed irrigation redistributes As to the soil where it can continue to pose a threat to human health and food security.

Figure 10    Graphical summary of the key GRM results from (a) national and (b) regional scale analyses[3]. Model coefficients of three important covariates with their error bars (± standard error) are shown in the bottom axis. The top axis shows equivalent change (in percentage) in mean As concentrations with an 1-unit increase of these covariates.

References

  1. Reported in confined sandstones from Kansas (USA): Sophocelous et al., 2006. A rainfall loading response recorded at 300 meters depth: Implications for geological weighing lysimeters. Journal of Hydrology 319: 237–244.
  2. Shamsudduha, M, Zahid, A, and Burgess, W G. (in review) Security of deep groundwater against arsenic ingress in SE Bengal Basin.
  3. 3.0 3.1 Revelle, R, and Lakshminarayama, V. 1975. The Ganges Water Machine. Science 188: 611–616.
  4. Shamsudduha, M, Taylor, R G, and Chandler, R. 2015. A generalised regression model of arsenic variations in the shallow groundwater of Bangladesh. Water Resources Research 51: 685–703.