Characterizing floodplain evolution by joint analysis of SAR, GOES IR and local precipitation data: Case study of Rio Colorado, Bolivia

Dryland rivers in an endorheic basin experience downstream decrease of channel width and depth as the consequence of transmission losses by percolation and evapo-transpiration. Major changes in the river morphology take place during short peak discharge periods when the volume of water and sediment by far exceeds the river capacity and, as a consequence, meander bends are cut off, and the river path may change position by avulsion. Successive avulsions create a complex network of cross-cutting abandoned river channels. The Río Colorado, located in the southeast part of the Altiplano basin in Bolivia, is such a river system. This system consists of a very low-gradient lacustrine coastal plain onto which is deposited a 400 km2 sheet of fluvial sediment over the last 4000 year. Traditional studies to monitor the morphological changes at the terminus of the river system are based on field data acquisition of the fluvial sediments. This is time consuming and only covers a small area of the total fluvial morphology. The combination of field measurements and remote sensing imagery allows for the analysis of the development of the entire river system terminus at a longer temporal scale. Changes of alluvial surfaces affect the reflectance of objects and patterns on the ground, which is recorded by satellite images. In this study we show the response of the delta to rainfall events of various intensities in the Rio Colorado watershed. We combine precipitation data in the watershed with spaceborne synthetic aperture radar (SAR) data. We use two data sets that contain precipitation information: rainfall estimates based on Geostationary Operational Environmental Satellite (GOES) weather satellite IR window brightness temperatures and direct measures of rainfall from rain gauges in the vicinity of the delta. To detect changes on the surface water content we use the backscatter intensity or amplitude images from the European Remote Sensing Satellite (ERS) Synthetic Aperture Radar (SAR) dataset. Our analysis shows that the response of the radar intensity to a major rainfall event consists of three phases. Just after a major rainfall event the radar intensity in inundated areas of the delta drops significantly with respect to its mean amplitude. As the time after inundation increases the radar signal is brighter than the long term average, after 10-20 days the region has generally reverted to its long term average radar intensity. This first phase corresponds well with the physical response of radar signals. When the delta is inundated the radar signal bounces away from the sensor, reducing the backscatter. The reason for the increased amount of backscatter during the second phase is less straightforward. Our fieldwork in the region suggests it correlates with an increase in roughness of the terrain possibly due to evaporation leaving a layer of salts. In the final phase, the return to the long term average is possibly achieved by aeolian erosion of the surface, reducing the roughness and thereby the amount of backscatter. In this study we demonstrate the feasibility of augmenting in-situ measurements with several types of spaceborne remote sensing data to monitor and characterize the evolution of the Rio Colorado floodplain in terms of spatial and temporal resolution, and report our results on the join analysis.