The use of spaceborne radar data to model inundation patterns and trace gas emissions in the central Amazon floodplain

Abstract

River floodplains are the dominant wetland habitat in the Amazon river basin, providing important habitation for aquatic flora and fauna, and playing a key role in sustaining regional fish production. The annual inundation pulse has been identified as the dominant environmental factor affecting aquatic biota on the floodplain, and the characteristics of this pulse, in terms of timing, duration and amplitude, vary spatially on the floodplain as a function of fluctuations in river stage height and topography. River floodplains are furthermore globally significant sources of methane (CH4) and other trace gases essential to climate regulation. Refined information on wetland distributions and dynamics are currently needed to improve estimates of habitat availability and to calculate regional contributions of trace gases, especially CH4, to the troposphere. This paper describes how multitemporal time series of spaceborne L-band Synthetic Aperture Radar (SAR) data from the Japanese Earth Resource Satellite 1 (JERS-1) were used to generate a model of the spatial and temporal variation of inundation on the floodplain of a typical black water river in the Central Brazilian Amazon and how this model was utilized, together with in situ measurements of river stage heights and CH4 fluxes, to model regional estimates of CH4 emissions. We also demonstrate how a JERS-1 SAR time series can be used to map the spatial variation of flood duration on the floodplain, a key factor controlling local variations in plant biodiversity. For both applications, the availability of adequate time series of satellite sensor data is the prime factor affecting the reliability and accuracy of the flood models and the spatial details of the flood duration map. The availability of in situ data, especially daily river height measurements, was also critical for the development of the flooding model and for the subsequent decoupling of the model from the satellite sensor data.