Lead Investigators

Jenny Jay (UCLA), Thomas Harmon (UC Merced), Eddie Kohler (UCLA), Deborah Estrin (UCLA), Charles Harvey (MIT)

Overview

Arsenic (As) in well water has led to the largest environmental poisoning in history; tens of millions of people in the Ganges Delta continue to drink groundwater that is dangerously contaminated with arsenic. Despite the tragic public health implications of this problem, we do not yet have a complete answer to the question of why dissolved arsenic concentrations are so high in the groundwater of the Ganges Delta. Answering this question has proven to be a complex scientific challenge requiring a combined understanding of biogeochemical, hydrological, and geological processes.  A current working hypothesis is that As is mobilized in the near surface environment where sediments are weathered by seasonal changes in the redox state that drive a cycle of pyrite oxidation and iron oxide reduction, and the dissolved As is transported into the aquifers by recharge. 

In January 2006, we deployed an embedded sensing system in a rice paddy in Bangladesh to evaluate the relationship between irrigation and arsenic contamination in the groundwater. We deployed 42 ion-selective electrodes (ISEs) to monitor ammonium, calcium, carbonate, chloride, pH, oxidation-reduction potential, and nitrate, and 8 soil temperature, moisture and pressure sensors at 3 different depths in 3 locations. The network collected 26,000 measurements over a period of 12 days. We observed several unexpected phenomena (already discussed in last year's report) which have spawned further study, and subsequent deployments; several UCLA students will be leaving in March, 2007 to deploy more sensors and collect physical samples for lab analysis.

Approach

Data from 2006 sensor deployments at our field site show large daily oscillations in soil water pressures and aqueous geochemistry that occur within the slower seasonal changes driven by the monsoonal climate. We propose to investigate the mechanisms leading to these daily oscillations, and the effects of this cycling on As mobility and transport.  The importance of these newly recognized diurnal processes will be investigated throughout the annual cycle of flooding, drying and rice cultivation.

The ultimate science aims for this project are:
Specific Aim 1: We hypothesize that hydrological processes (due to barometric changes and plant evapotranspiration) and plant-induced chemical changes (due to root oxygen release and surface water changes) drive the daily oscillations observed at our site.

Specific Aim 2: We hypothesize that these daily oscillations may drive As mobilization and transport at our site. Oscillations in redox conditions, coupled with oscillating recharge fluxes, can mobilize and transport arsenic and other chemicals into aquifers.

Systems/Experiments

San Joaquin River Deployment.  We deployed 21 embedded networked sensors at the confluence of the Merced and San Joaquin Rivers. This deployment is discussed earlier in the report. We deployed Confidence with this wireless network of 14 ion-selective electrodes (ISEs) and 7 temperature sensors. Our short-term goal was to validate the embedded networked system, and the functioning of Confidence. Our longer-term goal is to design a system that addresses the problems we experienced in Bangladesh, and deploy this system in Bangladesh in December, 2007.

We took two steps to validate this system. First, we deployed our sensors alongside the sensors connected to Hobo data-loggers described earlier in the report.  This was useful in order to validate the data collection hardware we used in our system, as the Hobo data-loggers are rugged, and have been field-tested at many sites, providing a good baseline validation point. Second, we periodically extracted physical water samples from each of the sites in order to further validate the data we collected. This process is also described earlier in the report, and we do not expand further. Using the output of the Hach Kit as ground truth, we validate our system by comparing sensor output to this reading. We find that of the 14 ISEs, and 7 temperature sensors, 5 ISEs and 1 temperature sensor are definitely faulty at some point during the deployment and require attention. Confidence detects 5 of these 6 faults.

For each of the 7 functioning sensors, Confidence never reports a fault. Data from two example sensors are shown in Figure 1. For all deployment graphs, circles corresponds to ammonium, and squares correspond to nitrate; the small points correspond to sensor readings, and the big points correspond to physical sample readings. Confidence never reports a fault for any of these functioning sensors.

Figure 1

3 of the 4 faults Confidence detected were broken or dying sensors. Within several days of deployment, Confidence notified us to check the ammonium sensor connected to mote 22, and the nitrate sensor connected to mote 10. We disconnected the sensors from the board and connected them to a portable meter, and validated the readings obtained fromthe board. We then compared the sensor output with nitrate and ammonium levels measured from physical samples extracted near each sensor tip. Data from these sensors and the physical samples are shown in Figure 2. From the graph it is easy to see that readings from the sensors are very far from the ground truth, indicative of a dead sensor. We later removed the sensors and re-calibrated them, validating that the sensors were completely broken. Confidence also notified us to check the nitrate sensor connected to mote 16. Physical samples validated that the sensor was not operating correctly. And re-calibration of the sensor revealed that it was not completely dead, but only responded to changes in very high concentrations.

Figure 2

The final fault was simply a disconnected temperature sensor wire. Once notified, we re-connected the wire and fixed the problem. The results from Confidence monitoring of network quality were not as good unfortunately. When there is only one fault, Confidence always notified us. However, we did not have time to add the ability for Confidence to detect multiple correlated faults before the deployment. So, when there were multiple correlated faults, Confidence did not detect them. We have since added and evaluated this functionality.

Bangladesh Field Trip 2007.  The upcoming field deployment at the experimental rice paddy in Munshiganj, Bangladesh will consist of two components.  In the first, oxygen sensors will be deployed to determine the chemical composition of an unsaturated zone that was observed under the plough pan.  The origin and composition (i.e. redox status) of this layer is unknown, but has great bearing on the geochemistry of the surrounding groundwater.  Introduction of oxygen to the subsurface by plant- or infiltration-driven processes can drive redox cycling that can lead to arsenic mobilization (oxidation of arsenic-bearing pyrite minerals, then subsequent reductive mobilization).

Secondly, we will deploy javelins with nitrate, ammonium, carbonate, and chloride sensors in transects from the bund toward the center of the field.  We aim to determine if the diurnal cycling in redox active species that was observed in the 2006 deployment is repeatable.  We will identify location and times in the rice paddy where this cycling is most prevalent, and which will inform decisions regarding locations for further studies (arsenic measurements, enclosure studies).

Accomplishments

• San Joaquin River Deployment. The results from this deployment are encouraging. We will be ready to deploy this improved system in Bangladesh once we integrate the new sensor-board and test the final result in Pt. Mugu.
• Second trip to Bangladesh field site.  Tiffany Lin and Christine Lee will travel to Bangladesh from March 10, 2007 to March 22, 2007.
• Proposal to fully fund the Bangladesh research was submitted to NSF
  in Fall 2006 and received favorable reviews.  The proposal has been
  "wait-listed" subject to the availability of funds.

Future Directions

In the next year, we will test hypotheses outlined in Specific Aim 2.  Namely, we hypothesize that the observed daily oscillations in redox state may drive As mobilization at our site.  Through 1) As measurements alongside our embedded sensor deployments, 2) enclosure studies in the field, 3) and microcosm studies, we will investigate effects of diurnal oscillations in hydrological regime and redox condition on the biogeochemistry of As at our field site.

1. Temporally dense As speciation measurements alongside embedded sensors deployments.  Through the 2006 and 2007 deployments in Bangladesh, we will gain an understanding of the location, extent, and timing or interesting redox changes.  Deployments in the next phase will focus on key locations, and involve analysis of As cycling through time at these sites.  We will install lysimeters next to each pylon, and groundwater will be withdrawn As speciation and analysis.  Geochemical reactions will be modeled using PHREEQCI Version 2.

2. Enclosure studies in the field.  Metals rings will be fabricated in Munshiganj and used for enclosure studies to test the hypothesis that live plants and roots impact subsurface As mobilization.  Two sets of control enclosures (two each of slotted and solid-walled enclosures) will be compared with sets of duplicate enclosures with the tops of the plants cut off (leaving dead roots in place) and with plants and major roots removed (12 enclosures in all).  Geochemical changes in the surface water and the groundwater in each enclosure can be monitored throughout the experiment with sensors.  Through comparing the various treatments, and inhibiting photosynthesis in some treatments in additional experiments (microbial photosynthesis can be inhibited through darkness), we will be able to distinguish the effect of surface water chemistry changes from changes in the root zone.

3. Laboratory microcosms.  We can test factors such as rice type, soils type, amount of nitrate in irrigation water, water level changes on geochemistry and As mobilization. We can also correlate sensor measurements taken in the microcosms with studies of the microbial community.

Using this experimental microcosm, we will investigate the effects of the following on As speciation in sediments and mobilization to the dissolved phase: (a) presence versus absence of plants; (b) variant of rice plant (previous work has shown Boro (dry season) rice accumulated less arsenite and arsenate compared with Aman (wet season) rice); (c) growth phase of rice (While the rice growth cycle has been shown to result in short-term fluctuations in microbial guilds, no research has yet tied this to As cycling); and (d) soil type.  We will use soil from our site, as well as other typical soils in Bangladesh; and (d) water management.  This will be accomplished by varying the hydraulic regime in the microcosms.  Effects of daily oxidizing conditions in the upper sediments can be investigated.

People

Faculty:

• Thomas Harmon, UC Merced, School Engineering
• Deborah Estrin, UCLA, Computer Science
• Charles Harvey, MIT, Civil and Environmental Engineering
• Jenny Jay, UCLA, Civil and Environmental Engineering
• Eddie Koehler, UCLA, Computer Science

Researchers:

• Alex Rat’ko, UC Merced, School of Engineering

Graduate Students:

• Christine Lee, UCLA, Civil and Environmental Engineering
• Tiffany Lin, UCLA, Civil and Environmental Engineering
• Nithya Ramanathan, UCLA, Computer Science
• Sarah Rothenberg, UCLA, Environmental Science and Engineering

Undergraduates:

• Chris Butler, UC Merced, Environmental Engineering