Weekly mapping of surface water extent in the intertropical wetlands using spaceborne GNSS reflectometry

Abstract

As of today, the uncertain distribution and dynamics of inundations in the tropical wetlands cause large uncertainties in the quantification of the world’s methane emissions. Long-term, global remote sensing observations currently rely on optical sensors, affected by cloud cover and vegetation, and on passive microwave imagery with a low spatial resolution. Global Navigation Satellite System Reflectometry (GNSS-R) L-band observations from space onboard the Cyclone GNSS (CYGNSS) mission can provide information about the dynamics of surface water at moderate to low spatial resolution and with a frequent revisit. In this article, we propose a methodology to compute the fraction of water contained in 0.1°pixels (∼10 km) using CYGNSS reflectivity and Above Ground Biomass (AGB) maps. We produced a 1-year time series of water fractions with a weekly time sampling for the full coverage of CYGNSS between ±38°of latitude, from August 2018 to July 2019. We evaluated the results against regional reference maps of surface water based on MODIS, JERS-1, and PALSAR, and against the Global Inundation Extent from Multi-Satellite (GIEMS) dataset. CYGNSS water fractions represent well the spatial distribution and the seasonality of inundation extent for most of the tropical wetlands, with a global RMSD of 15.0 % against GIEMS. We found a likely overestimation when the soil moisture (SM) exceeds 0.3-0.4 cm3/cm3, i.e. for saturated soils where GNSS-R reflections are coherent, as the SM was not parameterized in our retrieval model. Despite this, the results are consistent in the entire intertropical band including over densely vegetated areas. These weekly water fractions for 2018-2019 with a 0.1°pixel size are distributed to contribute to further comparisons and hydrological researches.