Water Level Changes on the Central Amazon Floodplain Measured with Interferometric JERS-1 SAR Data

The central Amazon floodplain is marked by non-channelized, diffusive flow conditions that prohibit the use of typical in-situ gauging methods for estimating discharge (e.g., rating-curves). In fact, gauging stations are located only on the mainstem and major tributaries, but not in the adjoining floodplains. Logistics and costs prohibit the installation of floodplain gauges, thus predictions of floodplain storage and subsequent discharge evolve from models, such as Muskingum methods, or from remote sensing. Interferometric processing of SIR-C SAR data has previously been used to demonstrate that water levels across the central Amazon floodplain are not horizontal, as assumed in a Muskingum model based approach. Yet, the interferometric SIR-C data exists only for a very brief, 24-hour period, and the method requires a "double-bounce" radar pulse travel path typical of flooded vegetation. Interferometric processing of JERS-1 SAR data over flooded forests of the floodplain provides net changes in water levels during the 44, 88, etc. day orbital repeat cycles. Interferometric coherence during these repeat-cycles is poor, but visible fringe patterns exist, even in unfiltered interferograms. Measured changes in water levels generally agree with channel gauges at Itapeau, Beruri, and Manacapuru, but within-floodplain fluctuations are not planar. Preliminary interpretation of the interferometric data suggests that during early rising water, inundation appears first as a patchwork bordered by small floodplain channels, whereas at mainstem peak stage, floodplain flow appears to sub-parallel the mainstem.