Modeling suspended sediment distribution patterns of the Amazon River using MODIS data Edward Park ⁎, Edgardo M. Latrubesse University of Texas at Austin, Department of Geography and the Environment, 305 E. 23rd Street, CLA 3.306, Austin, TX 78712, USA abstract article info Article history: Received 21 August 2013 Received in revised form 4 March 2014 Accepted 11 March 2014 Available online 1 April 2014 Keywords: Suspended sediment distribution Large rivers Floodplain Remote sensing Amazon Patterns of surface sediment concentration distribution in rivers are significant for understanding fluvial morphodynamics and environmental characteristics of the rivers and their floodplains. In the case of the Amazon Basin, complexity in sediment pattern distribution is affected by the anabranching channel pattern of the Amazon River, inputs from tributaries (some of which are among the largest rivers on Earth) and the existence of huge and complex floodplains. In this paper, patterns of surface sediment distribution are modeled based on Moderate Resolution Imaging Spectroradiometer (MODIS) data over the Amazon River by estimating surface sediment concentrations. Specifically, we aim to 1) detect the regional and seasonal variability of surface sediment in the main channel, 2) observe the influence of tributaries into the main system, 3) identify channel-floodplain interactions, and 4) investigate the internal variability of surface sediment along the main channel system. Field surface sediment concentration data from three gauging stations representing the upstream, midstream, and downstream sections of the Amazon River between 2000 and 2010 were used to calibrate 1328 MODIS daily surface reflectance images. Robust empirical models were derived between field surface sediment concentration and surface reflectance data from each station (0.79 b R 2 b 0.92, slopes significant at 99% confidence level) from 752 selected data after quality control. We applied empirical models to 2112 8-day composite surface reflectance images to generate surface sediment distribution maps since 2000. Overall, this study successfully demonstrated the capability of our MODIS-based model to capture the spatial and temporal variability of surface sediments in the Amazon River Basin, the largest river system on Earth. © 2014 Elsevier Inc. All rights reserved. 1. Introduction Understanding the patterns of sediment transport, erosion and deposition is critical in studying large rivers because sediment plays a major role in the hydrophysical and ecological functioning, evolution of the channel–floodplain system, and biogeochemical cycle (Bayley, 1995; Filizola, Guyot, Wittmann, Martinez, & Oliveira, 2011; Latrubesse, Stevaux, & Sinha, 2005; Mertes & Magadzire, 2007, among others). In addition, human induced environmental impacts can trigger widespread and remarkable changes on the fluvial system by altering sediment quantity and quality. Therefore, mapping aggradational landforms in rivers (i.e. islands, bar, natural levees, and delta) and analyzing multi-temporal recent channel changes have been areas of focus for river scientists across disciplines concerning the fluvial environments. However, the assessment of sediment fluxes has been concentrated on hydro-sedimentological techniques and, until recently, the identifi- cation of patterns of suspended sediment transport in terms of morphodynamics was poorly understood. The examination of surface water quality through remote sensing techniques has come to the attention of river scientists since the first earth-observing satellite was launched in the 1972. Due to the optical properties of surface sediments, nonlinear signals from sensors moun- ted on satellites show robust association with color of surface water (Albanakis, 1990; Baker & Lavelle, 1984; Bhargava & Mariam, 1991a; Curran & Novo, 1988; Doxaran, Froidefond, & Castaing, 2002; Forget, Ouillon, Lahet, & Broche, 1999; Holyer, 1978; Jensen et al., 1989; Novo, Hansom, & Curran, 1989; Topliss, Almos, & Hill, 1990; Witte & Heinlein, 1981). In particular, the strong responses within the visible light portion of the electromagnetic spectrum as a function of sediment concen- tration at the water surface enabled the retrieval of absolute values of a river's surface sediment load from space, as surface reflectance is significantly controlled by the scattering from suspended matters on the water surface (Kirk, 1989; Miller & McKee, 2004). This method has been validated with root mean square errors (RMSE) of estimated sur- face sediment concentrations ranging 10–20 mg/l per pixel (Mertes, Smith, & Adams, 1993; Warrick, Mertes, Washburn, & Siegel, 2004) by a number of different remote sensing instruments including Landsat, SeaWiFS, SPOT and MODIS through mapping surface waters in various regions across the Earth (e.g. Bi et al., 2011; Doxaran, Froidefond, & Castaing, 2002; Harrington, Schiebe, & Nix, 1992; Li, Huang, & Fang, Remote Sensing of Environment 147 (2014) 232–242 ⁎ Corresponding author at: University of Texas at Austin, SAC 4.178, 2201 Speedway, Austin, TX 78712, USA. Tel.: +1 512 230 4603; fax: +1 512 471 5049. E-mail address: geo.edpark@utexas.edu (E. Park). http://dx.doi.org/10.1016/j.rse.2014.03.013 0034-4257/© 2014 Elsevier Inc. All rights reserved. Contents lists available at ScienceDirect Remote Sensing of Environment journal homepage: www.elsevier.com/locate/rse