The Harmon Sink, which lies near the west coast of central Guam, is one of the striking surface features of the Northern Guam Lens Aquifer, a highly permeable carbonate island karst aquifer in uplifted Cenozoic limestone. Surrounded by the island's densest industrial and urban areas, the sink collects storm water from a surrounding industrial park and from the adjacent airport to the southeast. In recent years, it has also received large discharges of sewage from failing lift stations. The island's premier tourist district is located in the nearby coastal zone, in Agana Bay 2 to 4 km west, and in Tumon Bay 1 to 3 km north. There has been concern that contaminants entering the sink or airport may be carried to these bays by groundwater discharging in the coastal zone. A dye trace was therefore conducted to help characterize groundwater transport from the sink and the airport to the adjacent coastal zone. Dye receptors were placed at seeps and springs in each bay and in sampling wells installed 150 m down gradient from each injection point. The results of the dye trace are consistent with a highly permeable triple porosity system:

1. Dye injected at the surface in a dry streamway along the axis of the sink was detected on the fourth day after injection at sites in Agana Bay 2500 m west-southwest. Dye injected at the water table from a well on the airport about a kilometer south of the Harmon Sink injection point, was detected on the sixth day after injection at sites in Agana Bay 2150 m west. The general direction of transport to Agana Bay from each site is nearly perpendicular to the regional hydraulic gradient but consistent with mapped regional fracture orientation.
2. Dye from the Harmon Sink injection was detected on the 17^th day after injection at sites in Tumon Bay 1400 m northwest. Dye from the airport injection was detected on the eighth day after injection at sites in Tumon Bay 1400 m north-northwest. The transport rate to Tumon Bay was thus very rapid, though slower than transport to Agana Bay. The general direction of transport to Tumon Bay from each site is consistent with the regional hydraulic gradient.
3. Dye from the Harmon Sink injection was detected in the sampling well 150 m down gradient four days after the injection (with daily sampling). Dye from the airport injection was also detected in the sampling well 150 m down gradient when the well was sampled for the first time on the fourth day following injection. From these observations, we hypothesize that the fastest transport (to Agana Bay) was controlled by relatively open, regional-scale fracture pathways. The rapid, but intermediate transport rate (to Tumon Bay) may be controlled by gradient-driven flow, probably through enhanced secondary pathways in the general direction of the gradient. Local flow (from injection points to nearby sampling wells) is apparently controlled by gradient-driven diffuse flow through the aquifer matrix.