An Isotopic and Microbiological Tracer Approach to Assessing Recharge Mechanisms in Surface Water Affected Wells on Tutuila, American Samoa Christopher Shuler 1,2 *, Marek Kirs 2 , Henrietta Dulai 1,2 , Craig R. Glenn 1,2 , Randel DeWees 3 , and Aly El-Kadi 1,2 . 1 Dept. Geology and Geophysics, 2 Water Resources Research Center; University of Hawaii, 96822. 3 American Samoa Community College, 96799 Email: *cshuler@hawaii.edu Acknowledgements Fa’afetai Tele Lava to the people and agencies who made this work possible: American Samoa Power Authority Utu Abe Malae, Danielle Mauga, Taylor Savusa, Katrina Mariner, & operations staff American Smaoa EPA Jason Gambatese & Tim Bodell American Samoa Community College Randy DeWees, Kelley Anderson-Tagarino, Rocco Tinitali, Hugh Fuimaono, Janet and Mona Chang. And Scot Izuka & Joe Fackrell Support for this work was provided by: The Water Resources Research Institute Program, Pacific RISA and the U.S. EPA Region IX Making a Visible Difference Program Background Part of the water supply on Tutuila, the territory’s main island, is subject to one of the longest boil-water-advisories in U.S. history. Turbidity and bacteria spikes during heavy rainfall events show many of the island’s most productive wells (Fig. 1) receive groundwater under the direct influence of surface water (GUDI). However, it’s unclear whether surface water reaches the wells through improperly sealed well casings (Fig 2-A), or through the aquifer matrix itself (Fig 2-B). Here these hypotheses are examined with environmental tracers over seasonal and event- based timescales to constrain recharge timing and material transport from surface to production well pump. Tracers include the isotopic composition of water ( δ 2 H & δ 18 O), microbial indicators (Escherichia coli & total coliform (TC)), and physical tracers such as turbidity. ‐11.00 ‐9.00 ‐7.00 ‐5.00 ‐3.00 ‐1.00 1.00 5/1/2014 10/28/2014 4/26/2015 10/23/2015 4/20/2016 10/17/2016 4/15/2017 δ 18 O Groundwater and precipitation δ 18 O Samples Asu Faga Ili Moa Wells near collectors δ 2 D & δ 18 O in precipitation and groundwater Methods – Water samples for δ 2 H & δ 18 O analysis were collected monthly at six groundwater wells and at four precipitation sampling locations outfitted with cumulative precipitation collectors (Fig. 3). (1) Results – δ 2 H & δ 18 O values in precipitation show clear seasonal variability. In comparison, well waters show little variation throughout the year or between different sites, and consistently match an average mixed composition of water from different seasons (Fig. 4). Significance – Low seasonal variability in groundwater δ 2 H & δ 18 O values shows that regional scale recharge occurs on longer than seasonal timescales. This suggests the water recharged during heavy- rainfall events constitutes a relatively small volume of the total recharge. 0 100 200 300 400 500 600 700 Mmi-67 Taf-61 Mmi-89 Taf-72 Taf-33 Taf-81 E. coli [MPN] GUDI wells During heavy rain 3 days after rain Microbial tracers in groundwater Figure 1: Study region and locations of municipal wells. Wells are labeled by ID number and are color coded by U.S. Environmental Protection Agency (US-EPA) determined GUDI status. Figure 2: Hypothesized flow paths of surface contamination in GUDI wells. Scenario A depicts surface contaminated water (indicated by red lines) traveling through poorly sealed grout or well casing, while Scenario B shows how contaminated water may reach a properly sealed well through highly permeable aquifer matrix. Figure 3: Cumulative precipitation collector Figure 4: δ 18 O values in precipitation (colored lines) and in groundwater from wells located near precipitation collectors (black lines). Figure 6: E.coli counts vs. 48 hour cumulative rainfall in monthly samples at both GUDI and non-GUDI wells. Green lines represent liner best fit of the rainfall- E. coli relationship. r 2 values generally indicate fair correlation. The rainfall-TC relationship (not shown) was similar; however, r 2 values were generally lower (0.3, 0.88. 0.6 and 0.1 for wells 33, 81, 60, and 89 respectively). Figure 7: Heavy rain event E. coli sampling at GUDI wells. Two sample times were selected, during a heavy rain event (4 inches of rainfall in previous 48 hours) ( blue bars), and 11 days later ( orange bars) during a period of low 48 hour rainfall (0.07 inches). Methods - GUDI and non-GUDI wells were sampled monthly and during high-rain events for TC and E. coli . In groundwater, these bacteria indicate low aquifer-filtration capacity (2) and short travel times, as sub-surface E. coli die off rates can be >50% per day (3) . Results – Non-GUDI wells had insignificant TC levels and no E. coli detections, whereas GUDI wells consistently showed TC and E. coli presence. Monthly samples showed no seasonal trend . However, E. coli counts correlated with 48 hour rainfall totals (Fig 6). Rain event sampling indicated E. coli spikes occur during events, but persist marginally past 72 hours (Fig. 7). Significance – Rapid E. coli response suggests surface water recharge occurs on very short timescales, (<48 hours). Order-of-magnitude E. coli concentration discrepancies in adjacent GUDI wells shows the source may be heterogeneous or contribute variable amounts of water to different wells. Well: Taf-81, Status: GUDI, Ave. response to peak: 29 hrs., Ave. peak length: 5.5 days Figure 8: American Samoa Community College student interns and faculty assisted with sampling and analysis. Well: Taf-61, Status: GUDI, Ave. response to peak: 12 hrs., Ave. peak length: 36 hrs. Well: Taf-77, Status: GUDI, Ave. response to peak: 80 hrs., Ave. peak length:11 days Well: Moa-80, Status: GUDI, Ave. response to peak: 2.5 days, Ave. peak length: 3 days Well: Ili-84, Status: Non-GUDI, Ave. response to peak: na, Ave. peak length: na References: (1) Scholl, M. A., Ingebritsen, S. E., Janik, C. J., & Kauahikaua, J. P. (1996). Use of precipitation and groundwater isotopes to interpret regional hydrology on a tropical volcanic island: Kilauea volcano area, Hawaii. Water Resources Research, 32(12), 3525-3537. (2) Entry, J. A., & Farmer, N. (2001). Movement of coliform bacteria and nutrients in ground water flowing through basalt and sand aquifers. Journal of Environmental quality, 30(5), 1533-1539. (3) Foppen, J. W. A., & Schijven, J. F. (2006). Evaluation of data from the literature on the transport and survival of Escherichia coli and thermotolerant coliforms in aquifers under saturated conditions. Water Research, 40(3), 401-426. (4) Izuka, S. K., Perreault, J. A., & Presley, T. K. (2007). Areas Contributing Recharge to Wells in the Tafuna-Leone Plain, Tutuila, American Samoa (No. 2007-5167). Geological Survey (US). (5) Eyre, P., and G. Walker. 1991. Geology and ground-water resources of Tutuila American Samoa. Unpublished report in American Samoa Power Authority files. Method – Rainfall and turbidity data originally collected by U.S. EPA for GUDI well designation was reanalyzed. Turbidity data was filtered to identify peaks that correlated with heavy rain events. Result –GUDI wells showed faster and more dramatic turbidity response to rainfall, whereas non-GUDI wells maintained low turbidity levels overall (Fig. 5). Tafuna GUDI well turbidity spikes started 3 - 24 hours (mean = 17.6 hrs.) after rain event peaks, and lasted between 1.5 – 11 days (mean = 4.7 days). Significance – Turbidity spike duration indicates event water remains in the aquifer for multiple days after rainfall, suggesting it originates from an area with more storage than the material around a well annulus. Additionally, the response time of turbidity spikes is consistent with expected groundwater velocities in the Tafuna Aquifer material (4, 5) . Surface water infiltration through casings would be expected produce a faster response, similar to surface runoff peaks in streams, which occur within minutes to hours of heavy rain. This supports the hypothesis that surface water reaches wells through permeable aquifer material. Conclusions Turbidity response to rainfall events • Rapid recharge of surface water occurs quickly, within 3 – 48 hours of heavy rainfall • Turbidity travel time points towards aquifer material over faulty casings as the primary transport medium for surface water • Tafuna Aquifer material does not provide sufficient filtration and holding time to remove microbes from groundwater. • Results suggest abandonment of the Tafuna Well Field is recommended over rehabilitation of existing wells. Figure 5. Turbidity [NTU] (black line) in wells vs. heavy rain events (cyan points). Rain events of < 0.33 inch/hr. were ignored and turbidity spikes without preceding rain events were not analyzed (red points show peaks; grey boxes show durration). Taf- Taf- Moa- Taf- Ili-