G. De Grandi

Polarimetric SAR speckle filtering and its implication for classification

This paper proposes a new approach in polarimetric synthetic aperture radar (SAR) speckle filtering. The new approach emphasizes preserving polarimetric properties and statistical correlation between channels, not introducing crosstalk, and not degrading the image quality. In the last decade, speckle reduction of polarimetric SAR imagery has been studied using several different approaches. All of these approaches exploited the degree of statistical independence between linear polarization channels. The preservation of polarimetric properties and statistical characteristics such as correlation between channels were not carefully addressed. To avoid crosstalk, each element of the covariance matrix must be filtered independently. This rules out current methods of polarimetric SAR filtering. To preserve the polarimetric signature, each element of the covariance matrix should be filtered in a way similar to multilook processing by averaging the covariance matrix of neighboring pixels. However, this must be done without the deficiency of smearing the edges, which degrades image quality and corrupts polarimetric properties. The proposed polarimetric SAR filter uses edge-aligned nonsquare windows and applies the local statistics filter. The impact of using this polarimetric speckle filtering on terrain classification is quite dramatic in boosting classification performance. Airborne polarimetric radar images are used for illustration.

The use of decision tree and multiscale texture for classification of JERS-1 SAR data over tropical forest

The objective of this paper is to study the use of a decision tree classifier and multiscale texture measures to extract thematic information on the tropical vegetation cover from the Global Rain Forest Mapping (GRFM) JERS-1 SAR mosaics. We focus our study on a coastal region of Gabon, which has a variety of land cover types common to most tropical regions. A decision tree classifier does not assume a particular probability density distribution of the input data, and is thus well adapted for SAR image classification. A total of seven features, including wavelet-based multiscale texture measures (at scales of 200, 400, and 800 m) and multiscale multitemporal amplitude data (two dates at scales 100 and 400 m), are used to discriminate the land cover classes of interest. Among these layers, the best features for separating classes are found by constructing exploratory decision trees from various feature combinations. The decision tree structure stability is then investigated by interchanging the role of the training samples for decision tree growth and testing. We show that the construction of exploratory decision trees can improve the classification results. The analysis also proves that the radar backscatter amplitude is important for separating basic land cover categories such as savannas, forests, and flooded vegetation. Texture is found to be useful for refining flooded vegetation classes. Temporal information from SAR images of two different dates is explicitly used in the decision tree structure to identify swamps and temporarily flooded vegetation.

Measurement of topography using polarimetric SAR images

A processing technique for polarimetric synthetic aperture radar (SAR) data has been developed which produces profiles of terrain slopes and elevations in the azimuthal (or along-track) direction. This technique estimates the average shift in orientation angle of copolarization backscatter caused by azimuthal tilts of the scattering plane. Using P-band data, tests of this technique have been made for an area in the Black Forest near Villingen/Schwenningen in Baden-Wurttemberg, Germany. The radar measured slope and derived elevation profiles have low rms errors and high correlation values when compared with a stereo-photograph digital-elevation map (DEM) for the area. This algorithm is capable of adaptively making transitions from the forested areas to nearby regions with open-terrain. Subsequent tests of the algorithm have been conducted using polarimetric SAR L-band data for a mountainous, nonforested, region in the Mojave Desert (Ft. Irwin, CA) where an accurate DEM also was available. Complete elevation and slope mapping of the terrain in two dimensions using this technique is possible when azimuthal elevation profiles are produced throughout the range extent of the SAR image.

The Global Rain Forest Mapping Project JERS-1 radar mosaic of tropical Africa: development and product characterization aspects

The Global Rain Forest Mapping Project (GRFM) is an international collaborative effort initiated and managed by the National Space Development Agency of Japan (NASDA). The main goal of the project is to produce a high resolution wall-to-wall map of the entire tropical rain forest domain in four continents using the L-band SAR onboard the JERS-1 spacecraft. The processing phase, which entails the generation of wide area radar mosaics from the raw SAR data, was split according to the geographic area. In this paper, the focus is on the part related to Africa. The GRFM projects goal calls for the coverage of a continental scale area of several million km 2 using a sensor with the resolution of tens of meters. In the case of the African continent, this entails the assemblage of some 3900 high resolution SAR scenes into a bitemporal mosaic at 100 m pixel spacing and with known geometric accuracy. While this fact opens up an entire new perspective for vegetation mapping in the tropics, it presents a number of technical challenges. In this paper, we report on the solutions adopted in the GRFM Africa mosaic development and discuss some quantitative and qualitative aspects related to the characterization and validation of the GRFM products. In particular, the mosaic geolocation and its validation are discussed in detail. Indeed, the internal geometric consistency (subpixel accuracy in the coregistration of the two dates), and the absolute geolocation (residual mean squared error of 240 m with respect to ground control points) are key features of the GRFM Africa mosaic. Other important aspects that are discussed are the multiresolution decomposition approach, which allows for tracking the evolution of natural phenomena with scale; the internal semi-automatic radiometric calibration, which minimizes artifacts in the mosaic; and the thematic information content for vegetation mapping, which is illustrated by a few examples elaborated by visual interpretation. Experience gained so far indicates that the GRFM products constitute an important source of information for global environmental studies.

An overview of the JERS-1 SAR Global Boreal Forest Mapping (GBFM) project

Boreal ecosystems play an essential role in global climate regulation. Forests constitute pools of terrestrial carbon and are generally considered as global sinks of atmospheric CO/sub 2/, contributing to attenuating the greenhouse effect. Large amounts of carbon are also stored in boreal lakes, bogs and wetlands, partially released as CH/sub 4/ and other trace gases to the atmosphere during the spring and summer months. Human activities in the forest zone are however reducing the size of the carbon pool and climate change is triggering shorter winters and earlier thaw onset, changing the natural equilibrium. Given its global importance, there is a need to map and monitor the boreal zone, and as the changes occur on all from local, regional to global scales, fine resolution information over vast areas is required. The Global Boreal Forest Mapping (GBFM) project is an international collaborative undertaking initiated by NASDA in 1996, as a follow-on to the tropical-focused Global Rain Forest Mapping (GRFM) project [A. Rosenqvist et al., (2000)]. Utilising the L-band Synthetic Aperture Radar (SAR) on the Japanese Earth Resources Satellite (JERS-1). one of the main objectives of the GBFM project is the generation of extensive, pan-boreaL SAR image mosaics, to provide snap-shots of the forest wetland and open water status in the mid-1990s. Mosaics over Canada, Alaska. Siberia and Europe have been generated, available on the Internet and on DVD free of charge for research and educational purposes. The GBFM project also entails research activities in North America, Siberia and northern Europe, aimed at advancing scientific applications of L-band SAR data in the boreal zone.

Polarimetric SAR speckle filtering and its impact on classification

Speckle reduction of polarimetric SAR imagery has been studied using several different approaches. Most of these approaches exploited the statistical independence between HH, HV and VV channels. The statistical characteristics, such as correlation between channels, and polarimetric signature preservation, were not addressed. This paper proposes a new approach in polarimetric SAR filtering. This new approach emphasizes introducing no cross-talk, preserving polarimetric properties and statistical correlations between channels. In addition, the image sharpness is better maintained. The impact of using this polarimetric speckle filtering on terrain classification is also studied. NASA/JPL Les Landes polarimetric P-band and C-band SAR data is used for illustration.

Contextual clustering for image labeling: an application to degraded forest assessment in Landsat TM images of the Brazilian Amazon

The modified adaptive pappas clustering (MPAC) algorithm, previously published in the image processing literature, is proposed as a valuable tool in the analysis of remotely sensed images where texture information is negligible. Owing to its contextual, adaptive, and multiresolutional labeling approach, MPAC preserves genuine but small regions, is easy to use (i.e., it requires minor user interaction to run), and is robust to changes in input parameters. As an application example, an MPAC-based three-stage classifier is applied to degraded forest detection in Landsat Thematic Mapper (TM) scenes of the Brazilian Amazon, where intermediate states of forest alterations caused by anthropogenic activities can be characterized by image structures 1-3 pixels wide. In three TM images of the Para test site, where classification results are validated by means of qualitative and quantitative comparisons with aerial photos, degraded forest areas cover 13% to 45% of the image ground coverage. In the Mato Grosso test site, the degraded forest class overlaps with 1) 10% of the closed-canopy forest detected by the deforestation mapping program of the Food and Agriculture Organization (FAO, 1992), and 2) 19% of the closed-canopy forest detected by the Tropical Rain Forest Information Center (TRFIC, 1996). These figures are in line with the conclusions of a study where present estimates of annual deforestation for the Brazilian Amazon are speculated to capture less than half of the forest area that is actually impoverished each year.

Spiral eddy detection using surfactant slick patterns and polarimetric SAR image decomposition techniques

Surfactant slicks are widely dispersed throughout the oceans. Current driven features, such as spiral eddies, can be made visible by associated slick patterns. In this study a combined algorithm using the Cloude-Pottier decomposition and the Wishart classifier is presented to produce accurate maps of slick patterns. The study then uses the classified slick patterns to detect spiral eddies. The competing background wave-field is eliminated. Satellite SAR instruments performing wave spectral measurements, or operating as wind scatterometers, regard the slicks as a measurement error-term. The maps produced by the algorithm facilitate the flagging of slick contaminated pixels within the image

The GBFM radar mosaic of the Eurasian Taiga: selected topics on geo-location and preliminary thematic products

In the context of the Global Boreal Forest Mapping project (GBFM), an initiative of the Japan Aerospace Exploration Agency (JAXA), a continental scale radar mosaic of the Eurasian Taiga was compiled. The mosaic is composed of some 520 strip-images (typically covering 80 km by 2500 km each) acquired in 1997-98 by the L-band SAR aboard the JERS-1 spacecraft. The mosaic was assembled in two phases. Coverage in the first stage included the area between the Ural Mountains in the west, Bering Strait in the east, Arctic Ocean in the north and the Korean Peninsula in the south. In the second phase an extended version was produced that comprises the European part of the Boreal ecosystems west of the Ural Mountain and up to the European Union region. Pixel spacing of the high resolution final products is 100 m and map projection is Albers equal-area conical. In this paper selected topics are presented related to a revision of the mosaic geometry. This step was called for to improve the internal consistency and assure proper absolute geo-location of the mosaics with respect to reference data sets. It consists of: i) a data representation (virtual frames) for handling the strip-images in smaller units that are more effective for dealing with local distortions; and ii) inclusion of control points derived from the Landsat GeoCover data sets. Results characterizing the mosaic geometric accuracy in terms of root mean square residuals are reported. Finally, we present, as a first thematic result, a vegetation map at coarse resolution (900 m) derived by a combination of the Global Land Cover 2000 map and the GBFM radar mosaic

Wavelet polarimetric SAR signature analysis of sea oil spills and look-alike features

Marine dark features in SAR imagery related to oil spills, biogenic look-alikes and low wind areas are analyzed by means of the wavelet polarimetric signature (WASP) tool. The WASP encapsulates in a graph the dependency of the wavelet variance on dyadic scale and polarization state. Experiments on SIR-C/X-SAR C-band data showed the effectiveness of this analysis in characterizing textural features of the areas of study.