Variation of dissolved organic carbon transported by two Chinese rivers: The Changjiang River and Yellow River

Abstract

Real-time monitoring of riverine dissolved organic carbon (DOC) and the associated controlling factors is essential to coastal ocean management. This study was the first to simulate the monthly DOC concentrations at the Datong Hydrometric Station for the Changjiang River and at the Lijin Hydrometric Station for the Yellow River from 2000 to 2013 using a multilayer back-propagation neural network (MBPNN), along with basin remote-sensing products and river in situ data. The average absolute error between the modeled values and in situ values was 9.98% for the Changjiang River and 10.84% for the Yellow River. As an effect of water dilution, the variations of DOC concentrations in the two rivers were significantly negatively affected by discharge, with lower values reported during the wet season. Moreover, vegetation growth status and agricultural activities, represented by the gross primary product (GPP) and cropland area percent (CropPer) in the river basin, respectively, also significantly affected the DOC concentration in the Changjiang River, but not the Yellow River. The monthly riverine DOC flux was calculated using modeled DOC concentrations. In particular, the riverine DOC fluxes were affected by discharge, with 71.06% being reported for the Changjiang River and 90.71% for the Yellow River. Over the past decade, both DOC concentration and flux in the two rivers have not shown significant changes.

Keywords: Dissolved organic carbon; Multilayer back-propagation neural network; The Changjiang River; The Yellow River; Variation.