On the merging of optical and SAR satellite imagery for surface water mapping applications

Optical and Synthetic Aperture Radar (SAR) imagery from satellite platforms provide a means to discretely map surface water; however, the application of the two data sources in tandem has been inhibited by inconsistent data availability, the distinct physical properties that optical and SAR instruments sense, and dissimilar data delivery platforms. In this paper, we describe a preliminary methodology for merging optical and SAR data into a common data space. We apply our approach over a portion of the Mekong Basin, a region with highly variable surface water cover and persistent cloud cover, for surface water applications requiring dense time series analysis. The methods include the derivation of a representative index from both sensors that transforms data from disparate physical units (reflectance and backscatter) to a comparable dimensionless space applying a consistent water extraction approach to both datasets. The merging of optical and SAR data allows for increased observations in cloud prone regions that can be used to gain additional insight into surface water dynamics or flood mapping applications. This preliminary methodology shows promise for a common optical-SAR water extraction; however, data ranges and thresholding values can vary depending on data source, yielding classification errors in the resulting surface water maps. We discuss some potential future approaches to address these inconsistencies.