Space-based detection of wetlands' surface water level changes from L-band SAR interferometry

Abstract

Interferometric processing of JERS-1 L-band Synthetic Aperture Radar (SAR) data acquired over south Florida during 1993–1996 reveals detectable surface changes in the Everglades wetlands. Although our study is limited to south Florida it has implication for other large-scale wetlands, because south Florida wetlands have diverse vegetation types and both managed and natural flow environments. Our analysis reveals that interferometric coherence level is sensitive to wetland vegetation type and to the interferogram time span. Interferograms with time spans less than six months maintain phase observations for all wetland types, allowing characterization of water level changes in different wetland environments. The most noticeable changes occur between the managed and the natural flow wetlands. In the managed wetlands, fringes are organized, follow patterns related to some of the managed water control structures and have high fringe-rate. In the natural flow areas, fringes are irregular and have a low fringe-rate. The high fringe rate in managed areas reflects dynamic water topography caused by high flow rate due to gate operation. Although this organized fringe pattern is not characteristic of most large-scale wetlands, the high level of water level change enables accurate estimation of the wetland InSAR technique, which lies in the range of 5–10 cm. The irregular and low rate fringe pattern in the natural flow area reflects uninterrupted flow that diffuses water efficiently and evenly. Most of the interferograms in the natural flow area show an elongated fringe located along the transitional zone between salt- and fresh-water wetlands, reflecting water level changes due to ocean tides.

Introduction

Interferometric Synthetic Aperture Radar (InSAR) is a powerful technique for detecting small changes (cm level resolution) of the Earth's surface over a wide area. The method is widely used to study earthquake and volcanic induced crustal deformation (e.g., Burgmann et al., 2000, Massonnet and Feigl, 1998, Massonnet et al., 1993), land subsidence (e.g., Amelung et al., 1999, Baer et al., 2002, Bawden et al., 2001, Buckley et al., 2003), landslides (e.g., Colesanti and Wasowski, 2006, Dai et al., 2002, Kimura and Yamaguchi, 2000), mining and large-scale geotechnical projects (e.g., Raucoules et al., 2003, Tesauro et al., 2000), and glacier motion (e.g., Goldstein et al., 1993, Mohr et al., 1998). The above applications measure the displacement of solid surfaces, whether they are rocks, soil, buildings, or ice (glacier). However, water level changes in wetlands may also be measured by InSAR (Alsdorf et al., 2000, Wdowinski et al., 2004a). This application works only in wetlands and floodplains where woody or herbaceous vegetation emerge above the water surface. The horizontal water surface and the vertical vegetation allow a double-bounce reflection of the transmitted radar signal back to the satellite (Richards et al., 1987). In non-vegetated flooded areas the water surface acts as a mirror scattering away the entire radar signal, because of the satellite's off-nadir transmission angle, typically 20°–45°.

Space-borne SAR data have been acquired since the 1970's by several civilian satellites operating in the C- and L-band ranges (5.3 and 1.275 GHz corresponding to 5.66 and 23.5 cm wavelength, respectively). The longer wavelength L-band SAR data (23.5 cm wavelength) is considered more reliable for the InSAR technique in vegetated areas (Rosen et al., 1996, Zebker et al., 1997). The first applications of the InSAR method to floodplains and wetlands were conducted using L-band data acquired over the Amazon basin (Alsdorf et al., 2000) and southern Florida's Everglades wetlands (Wdowinski et al., 2004a). Recent studies have shown that the shorter wavelength C-band SAR data (5.66 cm) can also be used for wetland InSAR, mainly in woody wetland environment (Kim et al., 2005, Lu et al., 2005, Wdowinski et al., 2004a, Wdowinski et al., 2004b).

In this study, we use L-band SAR data to study water level changes and the derived hydrological conditions in the Everglades wetlands. The data were acquired over southern Florida during the years 1993–1996 by the Japanese Earth Resources Satellite (JERS-1). It is the most complete L-band dataset over a large-scale wetland acquired to date, containing seven repeat passes of two tracks that cover most of the Everglades wetlands. Because the Everglades consists of natural and managed wetland portions, as well as diverse vegetation types (woody, herbaceous, prairie, and saltwater mangroves), this large JERS-1 dataset allows a rigorous evaluation of the wetland InSAR method to various wetland environments.

In our previous study (Wdowinski et al., 2004a), we used three JERS-1 passes of one track acquired over six month period in second half of 1994. Here we expand this work in both spatial and temporal coverage. Our 2004 study focused on a limited region, specifically the Water Conservation Areas (WCA) located south of Lake Okeechobee (Fig. 1), where water level changes are largest and are dominated by the operation of human-made water control structures. In the current study, we use a more complete dataset (two tracks with seven repeat orbits) extending over a three year period. We also study a larger area, the entire south Florida wetlands, including both managed and natural flow wetland environments, as well as more diverse vegetation types. We find that the natural and managed flow environments have very different InSAR signatures, reflecting the different conditions in the two environments.