Several methods for estimating streamflow characteristics for low and peak flows for streams on Tutuila, American Samoa are presented. These methods, derived from frequency, flow-duration, and regression analyses, allow the user to compute estimates for the 7-day low flow with 2-year and 10-year recurrence intervals; the median flow; the mean flow; and the 2-, 5-, 10-, 25-, 50, and 100- year peak flood flows at gaged and ungaged streams. In addition, frequency techniques using the log-Pearson type III distribution were also used to compute for gaged sites the 1-, 7-, 14-, 30-, 60-, 90-, 120-, and 183-day low flows with 2-, 5-, 10-, and 20-year recurrence intervals; the 1-, 7-, 15-, 30- day high flows with 2-, 5-, 10-, and 25-year recurrence intervals; and the 2-, 5-, 10-, 25-, 50-, 100- year peak flows. The regression equations and estimated streamflow characteristics are based on streamflow data collected from 1958 through 1990 at 11 continuous-record gaging stations with 9 to 32 years of record, 75 low-flow partial-record stations, and 49 miscellaneous sites.
Low-flow regression analysis using the baseflow index basin characteristic provided results with standard errors of estimate ranging from 13.3 to 43.8 percent. These standard errors are comparable to the average values at the low-flow partial-record stations. Regional analysis, using the method of residuals, divided Tutuila into two hydrologic regions which correspond to differences in geology. Peak-flow regression equations had coefficients of determination ranging from 0.60 to 0.66 and standard errors of estimate ranging from 44.0 to 47.4 percent. The large standard errors result from the large variability of flood peaks compared with drainage area. The geology of Tutuila also affects the magnitude of observed flood peaks.
