A method for generating forest/non-forest maps from TanDEM-X interferometric data

Abstract

Land-cover classification by means of remote sensing data is of fundamental importance for a broad range of commercial and scientific uses. In particular, the detection of vegetated areas is of great interest for global change research and for applications in agriculture, cartography, geology, forestry, as well as regional planning. At a global level, only in the very last years high-resolution forest classification maps have been produced using global mosaics of multi-spectral Landsat sensor data, together with a forest cover change maps from 2000 to 2012 at 30 m resolution. Recently, the Advanced Land Observing Satellite (ALOS) L-band SAR from JAXA has produced a forest map with 25 m spacing from global mosaics collected from 2007 to 2010, by exploiting the variability in L-band of the HH and HV backscatter coefficients to discriminate between forest and non-forest areas. The aim of this paper is to present the first global forest/non-forest map obtained from TanDEM-X interferometric data. The TanDEM-X (TerraSAR-X add-on for Digital Elevation Measurement) mission comprises the two twin satellites TerraSARX (TSX) and TanDEM-X (TDX), which fly in close orbit formation of a few hundred meters distance and act as a large single-pass SAR interferometer (InSAR) since October 2010, with the main goal of producing a global and consistent Digital Elevation Model (DEM) with an unprecedented accuracy. The quality of interferometric data pairs is strongly influenced by the specific characteristics of the illuminated land-cover type. Over forested areas the presence of multiple scatterers at different heights and within a single resolution cell results in an increase of the interferometric phase uncertainty (the so-called volume decorrelation), whose intensity depends on several factors, such as the acquisition geometry, the sensor parameters, and the canopy density. Hence, from each bistatic TanDEM-X coherence map the volume decorrelation contribution can be estimated and, from this, a forest/non-forest map is derived. This paper shows the developed approach for discriminating between forested and non-forested areas from TanDEM-X interferometric data, and presents some examples aimed at verifying the validity of the proposed method. From this, global mosaics are generated from the TanDEM-X quicklook data set at a final resolution of 25 m x� 25 m. Global statistics as well as validation with external data will be included in the final version of the paper.