S. Le Hegarat-Mascle

Application of Dempster-Shafer evidence theory to unsupervised classification in multisource remote sensing

The aim of this paper is to show that Dempster-Shafer evidence theory may be successfully applied to unsupervised classification in multisource remote sensing. Dempster-Shafer formulation allows for consideration of unions of classes, and to represent both imprecision and uncertainty, through the definition of belief and plausibility functions. These two functions, derived from mass function, are generally chosen in a supervised way. In this paper, the authors describe an unsupervised method, based on the comparison of monosource classification results, to select the classes necessary for Dempster-Shafer evidence combination and to define their mass functions. Data fusion is then performed, discarding invalid clusters (e.g. corresponding to conflicting information) thank to an iterative process. Unsupervised multisource classification algorithm is applied to MAC-Europe91 multisensor airborne campaign data collected over the Orgeval French site. Classification results using different combinations of sensors (TMS and AirSAR) or wavelengths (L- and C-bands) are compared. Performance of data fusion is evaluated in terms of identification of land cover types. The best results are obtained when all three data sets are used. Furthermore, some other combinations of data are tried, and their ability to discriminate between the different land cover types is quantified.

Estimation of watershed soil moisture index from ERS/SAR data.

Abstract The aim of this article is to show that a watershed hydrological index could be derived from ERS/SAR measurements. Indeed, it is well known that, over bare soil, the SAR signal is a function of the geometric and dielectric surface properties. The problem to estimate soil moisture is to free from the effects of the space and time fluctuations of soil roughness and from the vegetation cover attenuation and scattering. The methodology presented here is based on the selection of land cover types or “targets,” for which the SAR signal is mainly sensitive to soil water content variations, and for which the vegetation and the roughness effects (in SAR signal) can be estimated and removed if needed. This method has been validated over an agricultural watershed in France. We show that the accuracy of the retrieved soil moisture is ±0.04–0.05 cm 3 /cm 3 , except during May and June, when vegetation cover is too dense to get reliable soil information.

Soil moisture estimation from ERS/SAR data: toward an operational methodology

Previous studies have shown the possibility of using European Remote Sensing/synthetic aperture radar (ERS/SAR) data to monitor surface soil moisture from space. The linear relationships between soil moisture and the SAR signal have been derived empirically and, thus, were a priori specific to the considered watershed. In order to overcome this limit, this study focused on two objectives. The first one was to validate over two years of data the empirical sensitivity of the radar signal to soil moisture, in the case of three agricultural watersheds with different soil compositions and land cover uses. The slope of the observed relationship was very consistent. Conversely, the offset could change, making the soil moisture retrieval only relative (and not absolute). The second one was to propose an operational methodology for soil moisture monitoring based on ERS/SAR data. The implementation of this methodology is based on two steps: the calibration period and the operational period. During the calibration period, ground truth campaigns are performed to measure vegetation parameters (to correct the SAR signal from the vegetation effect), and the ERS/SAR data is processed only once a field land cover map is established. In contrast, during the operational period, no vegetation field campaigns are performed, and the images are processed as soon as they are available. The results confirm the relevance of this operational methodology, since no loss of performance (in soil moisture retrieval) is observed between the calibration and operational periods.

Backscattering behavior and simulation comparison over bare soils using SIR-C/X-SAR and ERASME 1994 data over Orgeval☆

Abstract During April 1994, the three-frequency radar system flew on the Space Shuttle Endeavour, known as SIR-C/X-SAR mission (Shuttle Imaging Radar C/X-Synthetic Aperture Radar). Over the Orgeval watershed (France), the ground condition stayed very wet throughout the 5-day SAR mission. The SAR imagery allows a data collection over a range of roughness conditions on bare soils. Three classes were identified: very smooth sowings with crusted top layer, cloddy surfaces, and different ploughed fields for future crops. To complement the Shuttle Radar data (three frequencies L, C, X, incidence range from 44° to 57°), the helicopter-borne scatterometer ERASME (C- and X-bands, copolorized configurations) was used. Merging of the two databases was possible. As a result, incidence angles ranging from 25° to 50° are available in C- and X-bands for the copolarized cross sections. Then the major objective of the article is, over this available data collection, to begin the validation of current surface backscattering models to natural surfaces, the theoretical integral equation model (IEM) of Fung et al. (1992) and the empirical model of Oh et al. (1994). It shows adequacies and limits. The IEM model reproduces well radar scatter over smooth surfaces, but fails over rough surfaces, predicting a flatter response with incidence angle than the observed signals in C- and X-bands. Difference in backscatter response due to direction angles (perpendicular and parallel to the row direction) is difficult to reproduce over smooth surfaces by this model integrating anisotropic surface but may be due to the unadequacy of the surface representation. The Oh algorithm agrees well with the backscatter response over rough surfaces at medium incidence angle, but fails with a systematic underestimation over smooth conditions. As a conclusion, further developments are necessary on derivation of theoritical solutions over rough surfaces and on validation of semiempirical algorithms over data sets of various training sources (radar and natural conditions).

INTRODUCTION OF NEIGHBORHOOD INFORMATION IN EVIDENCE THEORY AND APPLICATION TO DATA FUSION OF RADAR AND OPTICAL IMAGES WITH PARTIAL CLOUD COVER

Abstract Two ways of introducing spatial information in Dempster–Shafer evidence theory are examined: in the definition of the monosource mass functions, and, during data fusion. In the latter case, a “neighborhood” mass function is derived from the label image and combined with the “radiometric” masses, according to the Dempster orthogonal sum. The main advantage of such a combination law is to adapt the importance of neighborhood information to the level of radiometric missing information. The importance of introducing neighborhood information has been illustrated through the following application: forest area detection using radar and optical images showing a partial cloud cover.

Surface soil moisture estimation from the synergistic use of the (multi-incidence and multi-resolution) active microwave ERS Wind Scatterometer and SAR data

This paper presents an original methodology to retrieve surface (<5 cm) soil moisture over low vegetated regions using the two active microwave instruments of ERS satellites. The developed algorithm takes advantage of the multi-angular configuration and high temporal resolution of the Wind Scatterometer (WSC) combined with the SAR high spatial resolution. As a result, a mixed target model is proposed. The WSC backscattered signal may be represented as a combination of the vegetation and bare soil contributions weighted by their respective fractional covers. Over our temperate regions and time periods of interest, the vegetation signal is assumed to be principally due to forests backscattered signal. Then, thanks to the high spatial resolution of the SAR instrument, the forest contribution may be quantified from the analysis of the SAR image, and then removed from the total WSC signal in order to estimate the soil contribution. Finally, the Integral Equation Model (IEM, [IEEE Transactions on Geoscience and Remote Sensing, 30 (2), (1992) 356]) is used to estimate the effect of surface roughness and to retrieve surface soil moisture from the WSC multi-angular measurements. This methodology has been developed and applied on ERS data acquired over three different Seine river watersheds in France, and for a 3-year time period. The soil moisture estimations are compared with in situ ground measurements. High correlations (R 2 greater than 0.8) are observed for the three study watersheds with a root mean square

Land cover discrimination from multitemporal ERS images and multispectral Landsat images: A study case in an agricultural area in France

More and more remote sensing data corresponding to various wavelength domains is becoming available. Visible/infrared data were first used for land cover classification. However, radar data are becoming more widely used for hydrological and agricultural applications. This paper discusses the performance, for land cover type discrimination, of an optical image acquisition and a multitemporal radar series. For the majority of land cover types existing within the test site (representative of northern European agricultural areas), both ERS multitemporal SAR and Landsat multispectral visible/infrared classifications lead to good results, with the latter being more robust. For better identification of cultures that are less represented, the complementarity of the two datasets may be exploited using an efficient data fusion algorithm based on the Dempster-Shafer evidence theory. The performance of this combination was verified on two successive vegetation cycles.

Performance of change detection using remotely sensed data and evidential fusion : comparison of three cases of application

The detection of changes affecting continental surfaces has important applications in hydrological, meteorological and climatic modelling. Using remote sensing data, numerous change indices have already been proposed. Previous work showed the interest of combining several of these to improve change detection performance, using the Dempster–Shafer evidence theory framework. This study analyses the performance of different change indices and their combination in different cases of application: forest logging either in pine forest or in mixed forest, and winter vegetation cover of fields in intensive farming areas, in comparison to the forest fire case presented in previous work. The interest of indices derived from Information Theory, some of which are original, is shown.

Unsupervised Subpixelic Classification Using Coarse-Resolution Time Series and Structural Information

In this paper, a new method is presented for a subpixelic land cover classification using both high-resolution structural information and coarse-resolution (CR) temporal information. To that aim, the linear mixture model is used for pixel disaggregation. It enables us to describe a CR time series in terms of the mixture of classes that are represented within each pixel. Then, the Bayes rule and the maximum a posteriori criterion lead to the definition of an energy function whose minimum corresponds to the researched optimal classification. A theoretical analysis of the labeling errors that may be obtained using this energy function is provided, raising the main parameters for labeling performance. The optimal classification is computed by combining linear regressions and simulated annealing, leading to an unsupervised algorithm. The method is validated with numerical results obtained on two different agricultural scenes (i.e., the Danubian plain and the Coet Dan watershed).

Derivation of wild vegetation cover density in semi-arid regions: ERS2/SAR evaluation

In this paper, a simple model is proposed for measuring the vegetation cover over soil surfaces from radar signals acquired in semi-arid regions. In such regions, vegetation is characterized by the presence of clumps which partially cover the soil surface. The proposed model describes the relationship between the percentage of covered surface and the measured radar signal. Model simulations over Tunisian test areas, where ground parameters are controlled, are performed and compared with actual ERS2 radar measurements. A very good agreement is found. The model is then used to derive a map of the vegetation cover density for the whole studied site (in Tunisia). The approach used here is based upon supervised classification with classes defined by inverting the model and taking into account ERS calibration error. Each of the four classes thus defined exhibits a good classification rate, greater than 85%. Finally, two important applications for natural resources management are presented: vegetation monitoring and soil moisture monitoring.