Maurice Borgeaud

On the characterization of agricultural soil roughness for radar remote sensing studies

The surface roughness parameters commonly used as inputs to electromagnetic surface scattering models (SPM, PO, GO, and IEM) are the root mean square (RMS) height s, and autocorrelation length l. However, soil moisture retrieval studies based on these models have yielded inconsistent results, not so much because of the failure of the models themselves, but because of the complexity of natural surfaces and the difficulty in estimating appropriate input roughness parameters. In this paper, the authors address the issue of soil roughness characterization in the case of agricultural fields having different tillage (roughness) states by making use of an extensive multisite database of surface profiles collected using a novel laser profiler capable of recording profiles up to 25 m long. Using this dataset, the range of RMS height and correlation values associated with each agricultural roughness state is estimated, and the dependence of these estimates on profile length is investigated. The results show that at spatial scales equivalent to those of the SAR resolution cell, agricultural surface roughness characteristics are well described by the superposition of a single scale process related to the tillage state with a multiscale random fractal process related to field topography.

A study of vegetation cover effects on ERS scatterometer data

The scatterometer flown onboard the European remote-sensing satellites ERS-1 and ERS-2 is a vertically polarized radar operating at 5.3 GHz (C-band) and has a spatial resolution of 50 km. In a number of studies, the sensitivity of the ERS scatterometer to vegetation has been demonstrated, but it is not yet clear which vegetation parameters are of primary importance to explain the ERS scatterometer signal. In this paper, the effects of land cover and seasonal vegetation development are investigated by comparing ERS scatterometer data with land cover information, normalized difference vegetation index (NDVI) data sets, and meteorological observations. As a study area, the Iberian Peninsula was chosen. The Iberian Peninsula is characterized by the Mediterranean climate that has a wet winter and a dry summer. This allows the authors to better differentiate the effects of the annual vegetation and precipitation cycle on the temporal evolution of the backscattering coefficient /spl sigma//spl deg/. It is shown that the ERS scatterometer has only limited capabilities for monitoring the vegetation development within a given year because most of the temporal variability of /spl sigma//spl deg/ is due to soil moisture changes. On the other hand, it might be of merit for vegetation discrimination on large scales (regional to global) because the percentage area of forests, bushes, and shrubs within one ERS scatterometer pixel is found to explain a significant part of the spatial variability of the signal.

Monitoring soil moisture over the Canadian Prairies with the ERS scatterometer

The capability of the scatterometers onboard the European Remote Sensing Satellites (ERS-1 and ERS-2) for soil moisture retrieval is investigated. The ERS scatterometer consists of three antennas that illuminate the Earths surface from three different viewing directions. This allows the authors to study the dependence of the backscattering coefficient /spl sigma//sup 0/ on the azimuth and the incidence angle. An analysis of ERS scatterometer data over the Canadian Prairie region shows that land surfaces are slightly anisotropic with respect to the azimuth angle. It is proposed to consider the azimuthal anisotropy as an additional error source to /spl sigma//sup 0/. The variation of /spl sigma//sup 0/ with the incidence angle was found to be linked to vegetation, but independent of soil moisture. Based on these observations, a method for the normalization of the backscattering coefficient with respect to the incidence angle is proposed. The normalized backscattering coefficient at an incidence angle of 40/spl deg/, /spl sigma//sup 0/(40), is sensitive to vegetation and, in the case of moderate vegetation (grassland to sparsely forested areas), to the soil moisture content. Soil moisture maps derived from ERS-1 scatterometer measurements are compared to maps representing conditions on annually cropped land showing agreement. Results suggest that, over the Canadian Prairies, estimates of the total water content in the soil profile might be possible with an accuracy of about 10% of field capacity if little or no rainfall has occurred for three days before radar image acquisition.

On current limits of soil moisture retrieval from ERS-SAR data

Assesses the feasibility of retrieving soil moisture content over smooth bare-soil fields using European Remote Sensing synthetic aperture radar (ERS-SAR) data. The roughness conditions considered in this study correspond to those observed in agricultural fields at the time of sowing. Within this context, the retrieval possibilities of a single-parameter ERS-SAR configuration is assessed using appropriately trained neural networks. Three sources of error affecting soil moisture retrieval (inversion, measurement, and model errors) are identified, and their relative influence on retrieval performance is assessed using synthetic datasets as well as a large pan-European database of ground and ERS-1 and ERS-2 measurements. The results from this study indicate that no more than two soil moisture classes can reliably be distinguished using the ERS configuration, even for the restricted roughness range considered.

Interpreting ERS SAR signatures of agricultural crops in Flevoland, 1993-1996

This paper discusses an analysis of ERS SAR imagery of agricultural crops in Flevoland, The Netherlands, over a four-fear period (1993 to 1996) to study the stability of multitemporal radar signatures from one year to the next. Direct comparisons of the multitemporal profiles of crop signatures are made to derive limits on their stability and to examine the differences between them from one year to the next. Sharp rises (of several dB) in temporal crop signatures are linked to variations in rainfall, freezing, and incident angle (due to imaging passes from different orbit tracks). Model simulations confirm the plausibility of these mechanisms and emphasize their importance for quantitative monitoring of agricultural crop development. The possibility of timing critical phases of the crop growth cycle is highlighted using field-to-field variations with particular regard to the emergence and closure of sugar beet. The interyear comparison also enables generalized comments to be made regarding the performance and stability of crop classification algorithms from one year to another. Only summer months are consistently identified as helping to distinguish broad-leaved crops from cereals. There is some evidence that other times of the year assist in distinguishing specific crops, but this evidence is not stable from one year to another.

The use of ERS-1/2 Tandem interferometric coherence in the estimation of agricultural crop heights

In this study, the relationship between ERS-1/2 SAR Tandem interferometric coherence and the height of sugar beet, potato, and winter wheat is investigated using measurements of crop growth and coherence. The development of both coherence and crop height is observed to be approximately linear during the early growing season, and linear crop-specific relationships between Tandem coherence and the heights of the studied crops are derived. The coherence of all the studied crops decreases as the crop heights increase, a probable explanation for this is that as the crop grows, it screens the ground more effectively, and a greater part of the incident radar energy is backscattered from vegetation that decorrelates more rapidly than the more stable soil. The relatively dense canopies of the root crops sugar beet and potato screen the ground considerably more effectively than the cereal crop winter wheat. This study gives a strong indication that ERS-1/2 Tandem interferometric coherence is related to the height of agricultural crops, and that this relationship can be used to retrieve crop heights.

Kernel Low-Rank and Sparse Graph for Unsupervised and Semi-Supervised Classification of Hyperspectral Images

In this paper, we present a graph representation that is based on the assumption that data live on a union of manifolds. Such a representation is based on sample proximities in reproducing kernel Hilbert spaces and is thus linear in the feature space and nonlinear in the original space. Moreover, it also expresses sample relationships under sparse and low-rank constraints, meaning that the resulting graph will have limited connectivity (sparseness) and that samples belonging to the same group will be likely to be connected together and not with those from other groups (low rankness). We present this graph representation as a general representation that can be then applied to any graph-based method. In the experiments, we consider the clustering of hyperspectral images and semi-supervised classification (one class and multiclass).

Semi-Supervised Novelty Detection Using SVM Entire Solution Path

Very often, the only reliable information available to perform change detection is the description of some “unchanged” regions. Since, sometimes, these regions do not contain all the relevant information to identify their counterpart (the changes), we consider the use of unlabeled data to perform semi-supervised novelty detection (SSND). SSND can be seen as an unbalanced classification problem solved using the cost-sensitive support vector machine (CS-SVM), but this requires a heavy parameter search. Here, we propose the use of entire solution path algorithms for the CS-SVM in order to facilitate and accelerate parameter selection for SSND. Two algorithms are considered and evaluated. The first algorithm is an extension of the CS-SVM algorithm that returns the entire solution path in a single optimization. This way, optimization of a separate model for each hyperparameter set is avoided. The second algorithm forces the solution to be coherent through the solution path, thus producing classification boundaries that are nested (included in each other). We also present a low-density (LD) criterion for selecting optimal classification boundaries, thus avoiding recourse to cross validation (CV) that usually requires information about the “change” class. Experiments are performed on two multitemporal change detection data sets (flood and fire detection). Both algorithms tracing the solution path provide similar performances than the standard CS-SVM while being significantly faster. The proposed LD criterion achieves results that are close to the ones obtained by CV but without using information about the changes.

Cluster validity measure and merging system for hierarchical clustering considering outliers

Clustering algorithms have evolved to handle more and more complex structures. However, the measures that allow to qualify the quality of such clustering partitions are rare and have been developed only for specific algorithms. In this work, we propose a new cluster validity measure (CVM) to quantify the clustering performance of hierarchical algorithms that handle overlapping clusters of any shape and in the presence of outliers. This work also introduces a cluster merging system (CMS) to group clusters that share outliers. When located in regions of cluster overlap, these outliers may be issued by a mixture of nearby cores. The proposed CVM and CMS are applied to hierarchical extensions of the Support Vector and Gaussian Process Clustering algorithms both in synthetic and real experiments. These results show that the proposed metrics help to select the appropriate level of hierarchy and the appropriate hyperparameters. HighlightsCluster validity measure for arbitrary shaped clusters with outliers.Cluster merging system grouping cluster cores based on the outliers� structure.Truly hierarchical variants of support vector and Gaussian process clustering.Benefits for unsupervised change detection applications are presented.

On the scattering from natural surfaces: the IEM and the improved IEM

A comparative study regarding the scattering from natural surfaces is conducted by means of real measurements and the integral equation model (IEM) and the improved integral equation model (I-IEM). A large set of experiments have been conducted and hereafter illustrated.