Negligible contribution of reservoir dams to organic and inorganic transport in the lower Mimi River, Japan

Abstract

Rivers fulfill an essential ecological role by forming networks for material transport from upland forests to coastal areas. The way in which dams affect the organic and inorganic cycles in such systems is not well understood. Herein, we investigated the longitudinal profiles of the various components of the water chemistry across three cascade dams in Japan: the Yamasubaru Dam, Saigou Dam, and Ohuchibaru Dam, which are situated along the sediment-productive Mimi River in different flow conditions. We analyzed the following water quality components: suspended solids (SS), turbidity, total iron (TFe), dissolved iron (DFe), total organic carbon (TOC), total nitrogen (TN), total phosphorus (TP), humic substance (HS), and major ionic components (Na+, Mg2+, Ca2+, Cl−, NO3−, and SO42−) in the downstream channels of the three dams during the low–intermediate-flow and high-flow events from 2012 to 2014. We estimated hourly loads of each component using hourly turbidity data and discharge data (i.e., L–Q model) separately, and the results are integrated to estimate the annual fluxes. The annual fluxes between the methods were compared to verify predictability of the conventional L–Q models. Annual flux of TOC, TN, DFe, and HS estimated by the turbidity displayed similar values, whereas the flux of SS, TFe, and TP tended to increase downstream of the dams. Among the dams, estimated flux proportions for TP and TFe were higher during high-flow events (74%–94%). Considering geographic conditions (e.g., absence of major tributary between the dams), the result implies that accumulated TP and TFe in the reservoirs may be flushed and transported downstream with SS over the short height dams during flood events. Assuming this process, the reservoir dams probably make only a fractional contribution to the organic and inorganic transport in the catchment studied. The percent flux errors for SS, TFe, and TP fluxes ranged from −7.2% to −97% (except for the TP flux in 2013), which highlights the risk of underestimating these components when using an L–Q model.