Uncertainty in acquiring elemental fluxes from subtropical mountainous rivers

Abstract. Mountainous watersheds in high standing islands of the western tropical and subtropical Pacific have received great international attention regarding its high physical and chemical weathering rates caused by cyclone invasion, friable lithology and high tectonic activity. Since mountainous region is usually difficult to assess, particularly, during severe weather conditions, hydrological responses of elements against full-scale of water discharge (often >2 orders of magnitude) are rarely reported. In this study, we conducted discrete sampling (~3 day interval) throughout four seasons and intensive sampling (hourly) during typhoon floods from three upstream watersheds in Taiwan during 2002–2005. These observations revealing various elemental responses are taken as complete dataset (i.e. reference flux) to evaluate flux uncertainty among constituents caused by different sampling frequency, sample size and estimators. Five constituents are analyzed, including nitrate (NO₃), sulfate (SO₄), dissolved organic carbon (DOC), calcium (Ca), and silicate (Si). Each has specific environmental and geological implications. Direct average, flow-weighted and rating curve methods were applied. Basing on statistical analyses, flow-weighted method is the most conservative method, and is recommended to use for all constituents if few samples are available. The rating curve method is suggested, only when water samples in high-flows are available. Direct average method is only appropriate for stable constituents, such as Si. These findings, such as concentration-discharge variation, sampling frequency effect, and flux estimator assessment, offer fundamental knowledge while estimating geochemical fluxes from small mountainous rivers in Oceania region.