"The combination of two complementary remotely sensed observations (multi-satellite imagery and radar altimetry) allowed us to estimate the amount of surface water – rivers and floodplains – missing in the Amazon compared to its average over 2003–2007," Frédéric Frappart from OMP-GET in Toulouse, France told environmentalresearchweb. "Our results show that surface water is a very important contribution to the total – almost 50%."

The research team used a technique that would ordinarily be employed to monitor sea-level changes, applying it to land-based surface water instead. Other groups have indirectly estimated surface-water storage using Gravity Recovery and Climate Experiment (GRACE) data, which measure the sum of the surface water, soil moisture and groundwater. Frappart and colleagues believe that these data suffer from leakage from other regions, due to the spherical harmonics representation of the GRACE data, which contaminates the signal at the GRACE gridcell resolution.

"The advantage of our method is to directly estimate the water content in the surface reservoir," said Frappart. "Besides, we think that because of the coarse resolution of the GRACE data (∼400 km), the signal measured is a mix of inundated and non-inundated surfaces, and is very smoothed. That is why the estimates are quite different."

While the importance of monitoring stored and moving freshwater in such a large river basin is well known, ground measurements in the Amazon basin are scarce and so the dynamics of freshwater there is unclear. The researchers have for the first time answered questions about how much water is missing and from where, using remote sensing data.

The group believes that the methods it used to derive water levels from multi-satellite datasets over rivers and floodplains offer the first opportunity for continuous monitoring of mass transport in the surface-water reservoir before the launch of the NASA-CNES Surface Water and Ocean Topography (SWOT) mission in 2019.

The team published the study in Environmental Research Letters (ERL).