Article pubs.acs.org/est
Sealing Rice Field Boundaries in Bangladesh: A Pilot Study Demonstrating Reductions in Water Use, Arsenic Loading to Field Soils, and Methane Emissions from Irrigation Water Rebecca B. Neumann,*,† Lara E. Pracht,† Matthew L. Polizzotto,‡ A. Borhan M. Badruzzaman,§ and M. Ashraf Ali§ †
Department of Civil and Environmental Engineering, University of Washington, Seattle, Washington 98195, United States Department of Soil Science, North Carolina State University, Raleigh, North Carolina 27695, United States § Department of Civil Engineering, Bangladesh University of Engineering and Technology, Dhaka 1000, Bangladesh ‡
S Supporting Information *
ABSTRACT: Irrigation of rice fields in Bangladesh with arseniccontaminated and methane-rich groundwater loads arsenic into field soils and releases methane into the atmosphere. We tested the water-savings potential of sealing field bunds (raised boundaries around field edges) as a way to mitigate these negative outcomes. We found that, on average, bund sealing reduced seasonal water use by 52 ± 17% and decreased arsenic loading to field soils by 15 ± 4%; greater savings in both water use and arsenic loading were achieved in fields with larger perimeter-toarea ratios (i.e., smaller fields). Our study is the first to quantify emission of methane from irrigation water in Bangladesh, a currently unaccounted-for methane source. Irrigation water applied to unsealed fields at our site emits 18 to 31 g of methane per square-meter of field area per season, potentially doubling the atmospheric input of methane from rice cultivation. Bund sealing reduced the emission of methane from irrigation water by 4 to 19 g/m2. While the studied outcomes of bund sealing are positive and compelling, widespread implementation of the technique should consider other factors, such as effect on yields, financial costs, and impact on the hydrologic system. We provide an initial and preliminary assessment of these implementation factors.
■
boundaries of fields.10,17−22 The soil beneath rice field bunds, the raised boundaries around rice fields, is not plowed and puddled each year like the planted portion of the field, and thus, the conductivity of the soil beneath the field boundaries is higher than the rest of the field. During and immediately after irrigation, water applied to rice fields flows into bunds and then down into the subsoil, recharging the aquifer. This water lost down bunds is unavailable to rice plants. However, a significant fraction of the arsenic initially contained within the lost water remains in the planted portion of the field.23 Furthermore, fuel is wasted on pumping this water up from the aquifer and, in methane-rich aquifers, this water may contribute methane to the atmosphere. Thus, water lost down the bund adds arsenic to rice field soils, costs farmers money, and potentially contributes greenhouse gases to the atmosphere, while providing no benefit to the rice plants. The amount of water lost down the bund is a function of the perimeter-to-area ratio of the field. Fields with a greater
INTRODUCTION Groundwater irrigation of rice has improved the quality of life in Bangladesh, increasing food security and improving wages.1 However, groundwater in Bangladesh is severely contaminated with arsenic2 (a carcinogenic chemical) and, in some areas of the country, contains high concentrations of methane3−7 (a potent greenhouse gas). The use of groundwater for irrigation in Bangladesh is known to load arsenic onto rice field soils,8 resulting in continually increasing soil arsenic concentrations over time,9 and we hypothesized that it contributes methane to the atmosphere.10 High arsenic concentrations in rice field soils are associated with increased arsenic concentrations in rice grain and decreased yields,11−14 while the contribution of methane to the atmosphere from groundwater irrigation is largely unknown, only quantified in a few locations around the world,15,16 and not in Bangladesh. Reducing the amount of groundwater applied to rice fields is a straightforward way to reduce the amount of arsenic applied to fields each year and to reduce methane emissions associated with using methane-rich groundwater for irrigation. Previous work conducted in Bangladesh and elsewhere has demonstrated that a notable fraction of water applied to rice fields is not used by the rice plants but instead is quickly lost down the © 2014 American Chemical Society
Received: Revised: Accepted: Published: 9632
January 21, 2014 June 23, 2014 June 25, 2014 June 25, 2014 dx.doi.org/10.1021/es500338u | Environ. Sci. Technol. 2014, 48, 9632−9640
Environmental Science & Technology
Article
perimeter-to-area ratio (i.e., smaller fields) lose proportionately more water out of their bunds than fields with a smaller perimeter-to-area ratio (i.e., larger fields); large perimeter-toarea ratios can result in >90% of applied irrigation water lost down bunds, and small perimeter-to-area ratios can result in almost no loss of applied irrigation water down bunds.10 Within this context, we undertook a field study to investigate the amount of water saved by sealing the bunds of fields with various perimeter-to-area ratios. We connect this water savings with reduced arsenic loading to rice fields and methane emissions from irrigation water. In an attempt to consider multiple dimensions of bund sealing as a rice field management technique, we also provide an initial assessment of the technique’s effect on yields, net financial costs/savings, and impact on the hydrologic system.
season,26 rice field porewater chemistry,23,27 rice field water flow patterns,10 and the impact that groundwater irrigation and recharge water chemistry have on groundwater arsenic concentration patterns in the aquifer.4,5,28 The mean and median perimeter-to-area ratio of rice fields in the area (1 km2) is 0.12 and 0.11 m/m2, respectively (see histogram of perimeter-to-area ratios, Supporting Information section 1). Bund Sealing Experiment. The bund sealing experiment involved six fields separated into three groups of paired fields based on similar perimeter-to-area ratios. The average perimeter-to-area ratio of the groups were approximately 0.075, 0.097, and 0.132 m/m2 (Table 1). These groups span the most commonly occurring perimeter-to-area ratios within our study area (Supporting Information section 1) and included fields for which we could obtain permission from the landowner to conduct our study. To stop field water from flowing into and then down through the bund soil, the bunds of one field in each group were sealed with locally obtained plastic before the rice fields were puddled and plowed. The surface of existing bunds was peeled away using a hoe, and the edge of the plow pan at the bottom of the bund was peeled up. Plastic was laid across the freshly exposed portion of the bund and pushed under the raised edge of the plow pan. The plow pan was then pushed back down to anchor the plastic, and the removed surface of the bund was replaced over the plastic (see Supporting Information section 2 for picture). This bund sealing approach mimics that of Tuong et al.18 who used plastic to seal the bunds of an experimental rice field. Another bund sealing option is to plow through and rebuild the bunds, as Patil et al.21 did in their experimental rice fields. This later approach was not an option at our field site because the bunds represent property boundaries, and plowing through them and rebuilding them would have required permission from all of the property owners in the area. In all six fields, temperature-compensated pressure transducers (In-Situ Inc. Rugged Troll 100) were installed to monitor water levels on a 5 min increment over the growing season. Transducers were placed ∼1.5 m down inside a sealed PVC pipe connected to a PVC screen at the field surface. The screen (covered with a sock to limit clogging) spanned from the plow pan up above the standing water in the field. This setup kept the transducer submerged under enough water to buffer potentially large daily fluctuations in temperature. Simultaneously, atmospheric pressure was measured with a barologger (In-Situ Inc. Rugged Baro Troll) kept in the window
■
METHODS Field Site. The field site is located in Bashailbhog village in the Munshiganj district of Bangladesh, which is approximately 30 km southwest of Dhaka (Figure 1). Dry season rice, boro
Figure 1. Study location and Google Earth image (Imagery date 3/18/ 2013, Copyright 2014 DigitalGlobe) of the field site with the six experimental fields outlined and labeled. Fields with similar perimeterto-area ratios are outlined in the same color. Bunds of the shaded fields were sealed with plastic.
rice, is grown in the area from January to May and is irrigated with arsenic-contaminated (∼390 μg/L As) and methane-rich (∼16 mg/L CH4) groundwater. Previous work conducted at the site has documented the spatial patterns of arsenic input and accumulation in rice field soils,9,24,25 arsenic uptake by rice plants,11 arsenic loss from rice field soils during the monsoon Table 1. Water Use field
status
area (m2)
perimeter (m)
P/A ratio (m/m2)
A I
sealed not sealed
2739.7 2779.3
199.5 211.9
E J
sealed not sealed
1580.4 1758.4
C K
sealed not sealed
avg. sealed: A, E, C avg. unsealed: I, J, K avg. unsealed: J, K a
avg. P/A (m/m2)
irrigation (cm)
irrig. diff. (cm)
frac. irrig. saved
0.073 0.076
0.075
109 ± 6 142 ± 8
34 ± 10
0.24 ± 0.07
162.7 160.8
0.103 0.091
0.097
118 ± 6 197 ± 8
79 ± 10
0.40 ± 0.05
899.7 953.2
114.7 130.5
0.127 0.137
0.132
93 ± 6 241 ± 7
149 ± 9
0.62 ± 0.04
n.a. n.a. n.a.
n.a. n.a. n.a.
n.a. n.a. n.a.
n.a. n.a. n.a.
106 ± 13 194 ± 50 219 ± 31
87 ± 51a 113 ± 34b
0.45 ± 0.29a 0.52 ± 0.17b
Calculation included unsealed fields I, J, and K. bCalculation included unsealed fields J and K. 9633
dx.doi.org/10.1021/es500338u | Environ. Sci. Technol. 2014, 48, 9632−9640
Environmental Science & Technology
Article
both the bund water flux and field water arsenic concentrations are high.23 Data and calculations in Neumann et al.23 suggest that, within the first day following an irrigation event, 50% to 60% of the irrigation arsenic remains in the field water column, with the remainder sequestered into field soils. Thus, for the unsealed fields, it was assumed that the average arsenic concentration in the water lost down the bund was (0.55 ± 0.05) × (390 μg/L), and this concentration was multiplied by the volume of water saved by sealing the bunds to calculate the amount of irrigation arsenic diverted into the bund. This mass of diverted arsenic was subtracted from the total amount of irrigation arsenic applied to the field. Errors for all quantities were propagated through the calculations. Methane Emissions from Irrigation Water. The amount of methane emitted from irrigation water was determined by measuring methane concentrations and the carbon isotopic composition of methane in water samples collected from the irrigation well: at the outlet, at the bottom of the jet of pumped water before it hit the pool of irrigation water, and in the pool just under the jet of pumped irrigation water (see Figure 5A for a photograph). Samples were collected by placing a 50 mL plastic syringe barrel with a 16-gauge needle in the irrigation flow path at the desired location, allowing water to flush through the syringe and needle, and placing a preweighed evacuated 15 mL glass vial (Labco Exetainer, Lampeter, UK) on the syringe needle. These samples were taken on three different days during a two-week long field campaign in December, collecting triplicate samples at each location during each sampling event. Vials were weighed to determine the volume of collected water (between 5 and 10 mL per sample) and were kept cool (4 °C) and upside down (to keep the gas headspace away from the pierced septa) before, during, and after transport back to the University of Washington for analysis. Methane concentration in the gas headspace was measured using GC-FID (SRI 8610C), and Henry’s law was used to calculate the amount of methane remaining dissolved in the sample water. The dissolved concentration was calculated by dividing the total amount of methane in the sample (headspace and water) by the volume of collected water. The change in methane concentration along the irrigation pathway was used to estimate the amount of methane emitted from irrigation water from a given volume of pumped water. To confirm that the change in methane concentration was due to methane emission rather than methane oxidation, a subset of samples was sent to the University of California Davis isotope lab where they were analyzed for 13CH4 with a ThermoScientific PreCon concentration system interfaced to a ThermoScientific Delta V Plus isotope ratio mass spectrometer. Microbial methane oxidation fractionates methane, leaving the methane pool with a carbon isotope signature that is approximately 10 per mil heavier.29 In contrast, advective degassing does not fractionate methane.30 The University of California Davis isotope lab’s long-term standard deviation for 13CH4 is 0.2‰. The lab uses pure CO2 as a reference gas to calculate provisional delta values of the sample peak. These provisional values are adjusted for changes in linearity and instrument drift such that the correct delta values for laboratory standards are obtained. Laboratory standards are commercially prepared methane gas diluted in helium or air and are calibrated against NIST 8559, 8560, and 8561. Further information on the lab’s analysis technique can be found in Yarnes.31
of a nearby village house. Field water levels were determined by subtracting out atmospheric pressure from the field-transducer data and adjusting measurements to height above the plow pan with manual measurements of water levels taken at various points during the season. Irrigation events were defined as a continual positive increase in field water level greater than 2 cm based on the 3 h moving average of the adjusted field waterlevel data. On the basis of informal observations at the field site, farmers usually add at least 2 cm of water during an irrigation event. The amount of water added to the field in an irrigation event was calculated accounting for porosity of unconsolidated soil sitting above the plow pan, which was determined as 0.6 from water content measurements taken on soil cores from Field E during a previous field campaign23 (Supporting Information section 3). The location for the top of the unconsolidated soil was set to the water-level height (above the plow pan) where the data indicated an increase in the rate at which water levels changed with time; field water levels drop more quickly within the unconsolidated soil layer than in ponded water due to the volume occupied by the soil grains. Rainfall measurements made with a totaling rain gauge that was emptied daily were collected from a meteorological station located 4 km southwest of the field site that is run by the Bangladesh Water Development Board. If an apparent irrigation event occurred on the same day as a rain event, the irrigation input was reduced by the amount of recorded rainfall and disregarded as an irrigation event if the resulting irrigation input was less than 2 cm. The resolution of the pressure transducers were conservatively estimated as ±1 cm. This uncertainty was propagated through our calculations for determining the total amount of irrigation water applied to fields. Water saved by sealing bunds was determined by comparing total irrigation water applied to sealed and unsealed fields. Errors were propagated through the calculation. The comparison assumes that other losses of water in the two sets of fields are similar (i.e., evaporation and vertical infiltration) and that farmers of the included fields have similar water management styles (i.e., would use similar amounts of water if bunds were not sealed). Analysis of monthly water use in the fields (Figure 3 and Supporting Information section 6) largely supports these assumptions. The analysis shows that the number of days the fields were dry (i.e., water level at or below the plow pan) was not statistically different (analysis of variance (ANOVA) p-value >0.05) between sealed and unsealed fields. With the exception of Field I, patterns of water use over the irrigation season were similar between the sealed fields and similar between the unsealed fields. Further, at the end of the irrigation season, when the bund sealing technique did not result in a water savings (see Results and Discussion), the amount of water used in sealed and unsealed fields was similar. Field I demonstrated unusual watering patterns relative to the other fields; a large amount of irrigation water was applied early in the season, and little irrigation water was applied later in the season (Figure 3A). Given this unusual behavior, our conclusions and statistical analyses do not include data from Field I. Arsenic Loading to Fields. For sealed fields, it was assumed that all of the arsenic in the applied irrigation water (390 μg/L) entered the field soils. For unsealed fields, some irrigation arsenic is lost down the bunds and thus is diverted from entering field soils. Arsenic enters bunds immediately after an irrigation event (i.e., within the first day of irrigation), when 9634
dx.doi.org/10.1021/es500338u | Environ. Sci. Technol. 2014, 48, 9632−9640
Environmental Science & Technology
Article
Implementation Considerations. Methodological details surrounding our preliminary assessment of factors relating to large-scale implementation of bund sealing as a water management technique are presented in Supporting Information section 4.
■
RESULTS AND DISCUSSION Water Savings. Bund sealing reduced water use in the first part of the irrigation season (December to February). During these early months, the water level in the sealed fields was maintained with very little irrigation input, while in the unsealed fields, the water level oscillated up and down with frequent loss of water and replenishment by irrigation (Figure 2
Figure 3. (A) Cumulative amount of irrigation applied to fields. Black lines indicate fields with unsealed bunds, and gray lines indicate fields with sealed bunds. Field pair A and I is marked with circles, field pair E and J is marked with squares, and field pair C and K is marked with diamonds. (B) Average monthly amount of irrigation added to sealed (Fields A, C, E) and unsealed fields (white bar: Fields I, J, and K; dark gray bar: Fields J and K). Error bars represent plus and minus one standard deviation around the mean. For a given month, stars indicate a statistically significant difference (ANOVA p-value
