Streamflow and Suspended-Sediment Loads Before, During, and After H-3 Highway Construction, North Halawa, Haiku, South Fork Kapunahala, and Kamooalii Drainage Basins, Oahu, Hawaii, 1983-99

A long-term study (1983?99) was conducted to determine the effects of the H-3 Highway construction on streamflow and suspended-sediment transport on Oahu, Hawaii. Data were collected at five streamflow-gaging stations before, during, and after construction and at two stream-gaging stations during and after construction. Drainage areas at the seven streamflow-gaging stations ranged from 0.40 to 4.01 mi2 and highway construction affected from 4 to 15 percent of these areas. Analysis of covariance and regression techniques were used to assess changes in streamflow and suspended-sediment loads during and after construction, relative to before-construction conditions.

Streamflow at the seven streamflow-gaging stations was compared to streamflow at an index station unaffected by highway construction. Streamflow data were divided into low- and high-flow classes, and the two flow classes were analyzed separately. Additionally, instantaneous peak flows were analyzed at three streamflow-gaging stations. During construction, observed low flows significantly increased by 108 percent at Luluku Stream, a tributary to Kamooalii Stream, and decreased by 31 percent at Kamooalii Stream. After construction, low flows increased by 47 percent at North Halawa Stream near Honolulu compared to low flows during construction. Low flows at Luluku Stream increased by 99 percent after construction compared to before construction. Increased low flows were attributed to removal of vegetation for construction and the increase of impervious areas that reduced infiltration. Decreased low flows were attributed to increased ground-water withdrawals and construction activities.

High flows observed during highway construction compared to before construction increased significantly only at Haiku Stream (by 25 percent). Observed high flows after construction compared to during construction increased significantly only at Kamooalii Stream (by 34 percent). Observed high flows after construction compared to before construction increased by 58 percent only at Luluku Stream. All increases in observed high flows are attributed to increased runoff from land-use changes caused by the highway construction. Instantaneous peak flows increased significantly at Luluku Stream. Luluku Stream had significant increases in low and high flows both during and after construction.

Suspended-sediment loads changed significantly at six out of seven sediment-gaging stations during highway construction. Construction activities increased observed suspended-sediment yields by 222, 426, 60, and 148 percent at North Halawa Stream near Kaneohe, North Halawa Stream near Honolulu, Right Branch Kamooalii Stream, and Haiku Stream, respectively. At Luluku Stream, observed suspended-sediment yields were lower during construction than before construction by 62 percent. After construction, suspended-sediment loads also changed significantly at six out of seven stream-gaging stations. Observed after-construction yields increased at North Halawa Stream near Kaneohe, North Halawa Stream near Honolulu, and Right Branch Kamooalii Stream by 49, 205, and 36 percent, respectively, and decreased at Kamooalii Stream and South Fork Kapunahala Stream by 62 and 71 percent. The observed increases in yields are smaller after construction than during construction indicating that suspended-sediment loads are likely returning to before-construction levels.

The effects of H-3 Highway construction on suspended-sediment loads were generally similar to studies of the effects of highway construction in other areas of the United States where 50 to 85 percent of the sediment loads were attributed to construction activities. The percentages of the observed yields attributable to H-3 Highway construction are similar to the above percentages, ranging from 37 to 81 percent. Decreases in suspended-sediment loads due to highway construction are unique and have not been widely reported in the literature. Where decrease in s