The integration of field measurements and satellite observations to determine river solid loads in poorly monitored basins

The use of satellite imagery to assess river sediment discharge is discussed in the context of poorly monitored basins. For more than three decades, the Peruvian hydrological service SENAMHI has been maintaining several gauging stations in the lower part of the Amazon River catchment. This network has been recently supplemented by the Hydro-geodynamics of the Amazon Basin (HYBAM) program, which has a water quality monitoring network distributed over five locations and allows the assessment of river discharge and surface suspended sediment (SSS) concentration. In this paper, the three stations that are located near the confluence of the Maranon and Ucayali Rivers, which form the Amazon River, are reviewed in detail. Two of the stations provide a complete time series of 10-day SSS samples over the studied period. The third station, along the Ucayali River, failed to provide valid estimates of sediment concentration at the river surface. The objective is to use satellite data as a substitute for the missing records in order to assess the Ucayali River sediment discharge, which has never been directly assessed before. An additional goal was to extend the river sediment discharge records for the other two stations. Water reflectance, assessed from the time series of MODIS satellite images, is calibrated using field-sampling campaigns to provide satellite-based SSS estimates. Validation is achieved using an independent dataset consisting of the 10-day SSS samples derived from the HYBAM network. Over a 4-year period between 2004 and 2008, there is greater than 10% agreement between satellite-derived data and network data for the two stations that provided complete field records. Based on satellite-derived SSS estimates assessed from 2000 to 2009, the river sediment balance is shown to be consistent between upstream and downstream stations. The use of satellite data and their integration with field data in the context of poorly monitored basins is discussed, and different cases are proposed.