Thomas Melzer

Appearance models based on kernel canonical correlation analysis

This paper introduces a new approach to constructing appearance models based on kernel canonical correlation analysis (kernel-CCA). Kernel-CCA is a non-linear extension of CCA, where a non-linear transformation of the input data is performed implicitly using kernel methods. Although, in this respect, it is similar to other generalized linear methods, kernel-CCA is especially well suited for relating two sets of measurements. The benefits of our method compared to standard feature extraction methods based on PCA will be illustrated experimentally for the task of estimating an objects pose from raw brightness images.

The ASCAT Soil Moisture Product: A Review of its Specifications, Validation Results, and Emerging Applications

Many physical, chemical and biological processes taking place at the land surface are strongly influenced by the amount of water stored within the upper soil layers. Therefore, many scientific disciplines require soil moisture observations for developing, evaluating and improving their models. One of these disciplines is meteorology where soil moisture is important due to its control on the exchange of heat and water between the soil and the lower atmosphere. Soil moisture observations may thus help to improve the forecasts of air temperature, air humidity and precipitation. However, until recently, soil moisture observations had only been available over a limited number of regional soil moisture networks. This has hampered scientific progress as regards the characterisation of land surface processes not just in meteorology but many other scientific disciplines as well. Fortunately, in recent years, satellite soil moisture data have increasingly become available. One of the freely available global soil moisture data sets is derived from the backscatter measurements acquired by the Advanced Scatterometer (ASCAT) that is a C-band active microwave remote sensing instrument flown on board of the Meteorological Operational (METOP) satellite series. ASCAT was designed to observe wind speed and direction over the oceans and was initially not foreseen for monitoring soil moisture over land. Yet, as argued in this review paper, the characteristics of the ASCAT instrument, most importantly its wavelength (5.7 cm), its high radiometric accuracy, and its multiple-viewing capabilities make it an attractive sensor for measuring soil moisture. Moreover, given the operational status of ASCAT, and its promising long-term prospects, many geoscientific applications might benefit from using ASCAT soil moisture data. Nonetheless, the ASCAT soil moisture product is relatively complex, requiring a good understanding of its properties before it can be successfully used in applications. To provide a comprehensive overview of themajor characteristics and caveats of the ASCATsoil moisture product, this paper describes the ASCAT instrument and the soil moisture processor and near-real-time distribution service implemented by the European Organisation for the Exploitation of Meteorological Satellites (EUMETSAT).A review of themost recent validation studies shows that the quality of ASCAT soil moisture product is – with the exception of arid environments –comparable to, and over some regions (e.g. Europe) even better than currently available soil moisture data derived from passive microwave sensors. Further, a review of applications studies shows that the use of the ASCAT soil moisture product is particularly advanced in the fields of numerical weather prediction and hydrologic modelling. But also in other application areas such as yield monitoring, epidemiologic modelling, or societal risks assessment some first progress can be noted. Considering the generally positive evaluation results, it is expected that the ASCAT soil moisture product will increasingly be used by a growing number of rather diverse land applications.

Nonlinear Feature Extraction Using Generalized Canonical Correlation Analysis

This paper introduces a new non-linear feature extraction technique based on Canonical Correlation Analysis (CCA) with applications in regression and object recognition. The non-linear transformation of the input data is performed using kernel-methods. Although, in this respect, our approach is similar to other generalized linear methods like kernel-PCA, our method is especially well suited for relating two sets of measurements. The benefits of our method compared to standard feature extraction methods based on PCA will be illustrated with several experiments from the field of object recognition and pose estimation.

Stroke detection of brush strokes in portrait miniatures using a semi-parametric and a model based approach

The arrangement of brush strokes is an important criterion in classifying portrait miniatures. In order to detect single brush strokes we used both a model based and a semi-parametric, neural network approach. The performance of both operators is evaluated and compared experimentally.

Probabilistic Fusion of

A novel sensor fusion algorithm for retrieving the freeze/thaw (f/t) state from scatterometer data is presented: It is based on a probabilistic model, which is a variant of the Hidden Markov model, and it computes the probability that the landscape is frozen, thawed, or thawing for each day. By combining Ku- and C-band scatterometer data, the distinct backscattering properties of snow, soil, and vegetation at the two radar bands are exploited. The parameters that are necessary for inferring the f/t state are estimated in an unsupervised fashion, i.e., no training data are required. Comparison with model and in situ temperature data in a test area in Siberia/northern China indicates that the approach yields promising results (typical accuracies exceeding 90%); difficulties are encountered over bare rock and areas where large fluctuations in soil moisture are common. These limitations turn out to be closely linked to the inherent assumptions of the probabilistic model.

\hbox{K}_{\rm u}

Since December 2008, the European Organisation for the Exploitation of Meteorological Satellites has been operationally distributing a global 25-km surface soil moisture product derived from the Advanced Scatterometer (ASCAT) onboard the meteorological operational platform (METOP) satellite METOP-A. Soil moisture is retrieved by using the semiempirical change detection method originally developed by the Vienna University of Technology (TU Wien) for the Active Microwave Instrument (AMI) flown on the European Remote Sensing (ERS) satellites ERS-1 and ERS-2. With the launch of the first of the three Meteorological Operational Platforms (METOP-A) in October 2006, ASCAT onboard METOP-A inherits and continues the role of his predecessor AMI. The original soil moisture retrieval algorithm (TU Wien model) was expected to be almost directly applicable for ASCAT with only minor changes, since the configuration and technical design is similar to the ERS scatterometers. Since the TU Wien model requires a robust historic long-term reference of scattering parameters, the initial near real-time METOP ASCAT soil moisture product had to rely on the model parameters derived from over 15 years of ERS-1/2. However, the combination of ASCAT backscatter measurements and ERS-1/2 historic long-term reference introduced some artifacts in the soil moisture product. The objectives of this paper were to analyze and investigate the impact of the ERS-1/2 historic long-term reference on the soil moisture retrieval. An error model has been developed to quantify the effects of the two main error sources: differences in spatial resolution and absolute calibration. The results of the study show that a simple model is able to describe the artifacts in the initial near real-time METOP ASCAT soil moisture product, which frequently occur in areas characterized by sharp backscatter contrasts. The expected overestimation of soil moisture using ERS-1/2 model parameters due to a calibration bias between AMI and ASCAT could be modeled as well.

- and C-band Scatterometer Data for Determining the Freeze/Thaw State

The analysis of multitemporal synthetic aperture radar (SAR) images requires normalization to a common incidence angle when the images are acquired in varying view geometry. The dependence of radar backscatter on the incidence angle is known to vary with vegetation cover and will therefore change throughout the year. This letter tries to quantify the effect of neglecting this seasonal variability in an empirical incidence angle normalization for a region under Mediterranean climatic conditions. A methodology is presented to assess, at monthly intervals, the seasonal variability of the angular dependence of ASAR Wide Swath (WS) backscatter over Calabria, Italy. It is observed that the angular dependence of backscatter strongly fluctuates temporally, depending on the land cover. The angular dependence of ASAR WS backscatter has a similar seasonal behavior as that of the Advanced Scatterometer for large parts of the study site when both are resampled to a common spatial and temporal resolution. It is found that errors that are larger than the radiometric accuracy of the sensor may be introduced when this temporal variability is ignored in an angular normalization.

Error Assessment of the Initial Near Real-Time METOP ASCAT Surface Soil Moisture Product

Validation of soil moisture observations from remote sensing platforms is quickly becoming a routine task as operational soil moisture products continue to be developed. Despite this, there are no agreed upon validation procedures or open source implementations of common tasks and methods. This makes the reproduction of scientific results difficult, especially since it is not yet common that software is include in publications. The presented work aims to improve this situation by showing an open source toolbox that makes it easy to perform common validation tasks. This toolbox is also used as the backend of a web application that aims to simplify the comparison of different validation methods.

Seasonality in the Angular Dependence of ASAR Wide Swath Backscatter

In the recent literature, gradient-based (filtered) eigenspaces have been used as a means to achieve illumination insensitivity. In this paper we show that filtered eigenspaces are also inherently robust w.r.t. (non-Gaussian) noise and occlusions. We argue that this robustness stems essentially from the sparseness of representation and insensitivity w.r.t. shifts in the mean value. This is also demonstrated experimentally using examples from the field of object recognition and pose estimation.

Open source toolbox and web application for soil moisture validation

The article gives an in-depth overview of historical and recent advancements in the soil moisture retrieval from European C-band scatterometer missions. The TU Wien soil moisture retrieval model is outlined and complemented with an overview of available global soil moisture products derived from ERS ESCAT and MetOp ASCAT. Finally, two methodologies are presented for the creation of a consistent soil moisture data record for climate change research.