A GLOBAL BACKSCATTER MODEL FOR C-BAND SAR Daniel Sabel (1) , Marcela Doubková (1) , Wolfgang Wagner (1) , Paul Snoeij (2) , Evert Attema (2) (1) Vienna University of Technology, Institute of Photogrammetry and Remote Sensing, Gusshausstrasse 27-29, 1040 Vienna, Austria, Email: ds@ipf.tuwien.ac.at (2) European Space Agency, ESTEC, Keplerlaan 1, P.O. Box 299, 2200 AG Noordwijk, The Netherlands, Email: paul.snoeij@esa.int ABSTRACT The paper presents the first Global Backscatter Model (GBM) for C-band SAR over land, with a resolution of 1 km. The GBM, which has been derived from 55000 ASAR GM datasets, describes the locations specific backscatter characteristics with a linear model. It can be used in support of SAR image interpretation, processing and simulation. The GBM has been used within the Sentinel-1 project for system performance validation. 1. INTRODUCTION The ENVISAT Advanced Synthetic Aperture Radar (ASAR) offers several different acquisition modes at C- band for continuous Earth observation at global, regional and local scales. As a low-bandwidth background mission, it offers the Global Monitoring (GM) mode with a 1 km resolution. Due to the relatively high temporal sampling of the GM data, it is possible to derive parameters describing the backscattering behaviour of the entire global land surface. Based on time series analysis of more than 55000 GM datasets, the first global backscatter characteristics database at the 1 km scale, called the Global Backscatter Model (GBM), has been derived. The GBM characterises C-band backscatter in HH polarisation for 97% of the global land mass between 60ºS and 70ºN. It is the intention of the Global Backscatter Model to act as a reference database in support of research as well as SAR image interpretation, processing and simulation. 2. DATASET The ASAR sensor provides the GM mode as a low- bandwidth background mission with a resolution of 1 km. Due to the low power consumption, the GM mode can be operated potentially continuously. Together with the 405 km wide swath, made possible through electronic beam steering (ScanSAR), revisit times as short as 4 days can be obtained at the equator. Even shorter revisit times can be obtained at higher latitudes, as can be seen in Fig. 1. Due to the relatively short revisit times, when compared to other SAR data, the GM mode is suitable for analysing temporally dynamic phenomena. The temporally dense data also provides an opportunity to characterise general attributes of the land surface, rather than merely its properties at a certain point in time. Furthermore, due to its relatively low spatial resolution, and therefore relatively low storage and processing requirements, it became feasible to carry out a global analysis using high performance off-the- shelf computers. Figure 1. Actual mean revisit times for ASAR GM acquisitions. Tab. 1 lists technical specifications of the GM data used in the study [1]. The time span of GM data acquisition varied between locations but covers in general the years 2005 to 2008 (4 years). Table 1. Specifications of ASAR GM data. Product type ASA_GM1_1P Centre frequency 5.331 GHz Spatial resolution 1000 m Polarisation HH Swath width 405 km Equivalent no. of looks 7-9 Incidence angle range 17º-42º Due to the different characteristics of backscatter depending on whether or not the soil is frozen or snow- covered, a meteorological analysis was carried out in the analysis, based on the ERA-Interim dataset, which is produced by the European Centre for Medium Range Weather Forecast (ECMWF) [2]. ERA-Interim data are generated for 91 vertical levels at a spatial resolution of 0.25 degrees and at 00, 06, 12 and 18 UTC time steps [3]. The ERA-Interim parameters used in this study are the snow water equivalent and the 2 metre temperature. _________________________________________________ Proc. ‘ESA Living Planet Symposium’, Bergen, Norway 28 June – 2 July 2010 (ESA SP-686, December 2010)