Estimating inundation extent using CYGNSS data: A conceptual modeling study

Abstract

Mapping inundation dynamics and flooding extent is important for a wide variety of applications, from providing disaster relief and predicting infectious disease transmission to quantifying the effects of climate change on Earth's hydrologic cycle. Due to the rapid and highly spatially heterogeneous nature of flooding events, acquiring data with both high spatial and temporal resolutions is paramount, yet doing so has remained a challenge in satellite remote sensing. The potential for Global Navigation Satellite System-Reflectometry (GNSS-R) to help address this challenge has been explored in several studies, the bulk of which use data from the Cyclone GNSS (CYGNSS) constellation of GNSS-R satellites. This work presents a simple forward model that describes how surface reflectivity measured by CYGNSS should change due to flooding for different land surface types. We corroborate our model findings with observations from the Amazon Basin and Lake Eyre, Australia. Both the model and observations indicate that the relationship between surface reflectivity and surface water extent strongly depends on the micro-scale surface roughness of the land and water. We show that the increase in surface reflectivity due to flooding or inundation is greatest in areas where the surrounding land has dense vegetation. In areas where the land surface surrounding inundated areas is perfectly smooth, the increase in surface reflectivity due to flooding is not as strong, and confounding effects of soil moisture and water roughness could lead to large uncertainties in resulting surface water retrievals. However, even a few centimeters of surface roughness will result in several dB sensitivity to surface water, provided that the water is smoother than the land surface itself.