Abstract
Grapevine phenology observations are essential for ecological adaptability of grape varieties, crop management and crop modelling. Phenological events have traditionally been ground based, with observations mainly providing information concerning grape varieties over a limited spatial area and few in-season observations. Time-series of satellite imagery can rapidly provide a synoptic and objective view of grape vegetation dynamics that may be used for vineyard management. Ten-day VEGETATION image composites from 1999 to 2007 were used to examine temporal profile in the Normalized Difference Vegetation Index (NDVI) and their relationship with ground based observation of grapevine phenology. In Portugal is Douro wine region, 2 suitable tests sites with over 70% or more of their area occupied by grapevines were selected. A number of NDVI metrics were obtained for each year through logistic model adjusted to NDVI time series after noise reduction using a Savitzky-Golay filter. The comparison of ground-based vineyard phenology and satellite-derived flowering, show an average spread deviation of 3 days. The satellite derived full canopy date was significantly correlated to the veraison date (r=0.87; n=7; p<0.02). The data set provided by the VEGETATION sensor proved to be a valuable tool for vineyard monitoring, mainly for inter-annual comparisons on regional scale.

Abstract
There are several data fusion methods widely used to produce a high resolution multi-spectral image from a pair of images - a panchromatic high resolution and a multi-spectral lower resolution image. Although the fused images can be visually satisfactory, it is not clear whether they provide additional information for quantitative measurements made from satellite images. A methodology to evaluate data fusion algorithms is proposed, based on the production of synthetic images that reproduce real satellite images. An experiment was conducted testing the performance of six data fusion methods in the production of NDVI values for land parcels from SPOT HRG and Landsat TM data. The fusion methods evaluated were: Brovey, IHS Hexcone, IHS Cylinder, PCA, Wavelet IHS and Wavelet Single Band. The best data fusion method overall was found to be Wavelet IHS, although better results were obtained by using directly the lower resolution multi-spectral data instead. The software tools developed and a number of test images datasets are freely available at the SITEF website (www.fc.up.pt/sitef).

Abstract
"Lameiros" are ancestral semi-natural meadows, essential elements of mountain landscapes in Northern Portugal. In the "lameiros" a traditional irrigation system is used and water is applied all year around. They are mainly used for forage production for autochthonous bovine feeding, but they are also important for the water and nutrients cycle regulation, erosion control and as barrier to forest fires propagation. Although recognized for their economical, environmental, landscaping, cultural and genetic value, the perpetuation of these "lameiros" could be at risk, at medium term, due to human desertification in the mountain regions and to the announced constraints in use of water resources. To preserve these landscapes it is essential to know them better and to better characterize them. Therefore a monitoring program using remote sensing tools is now being developed to evaluate different patterns of "lameiros", and their spatial extent and evolution. Two important questions are determinant in this program: the selection of the most appropriate spatial resolution for monitoring "lameiros", and the availability of satellite historical data. In this context, NDVI were compared in two selected test sites, with and without full irrigation. Data were derived from several field campaigns with a spectroradiometer and using different sensors: i) Landsat 5 and Landsat 7 (30m pixel), ii) SPOT 4 and SPOT 2 (20m pixel), iii) SPOT 5 (10m pixel). The NDVI temporal series produced were evaluated considering "lameiros" management and weather data. Results obtained so far indicate that the SPOT images provide data at the most adequate scale.

Abstract
The segmentation of satellite images is a valuable tool to obtain useful information about the spatial distribution of different land cover types. The use of segmentation algorithms instead of the traditional pixel-by-pixel classifiers used to produce land cover maps results on images that exhibit a more homogeneous distribution of classes, showing the piecewise spatial continuity of the real world. Several segmentation and classification methods are being developed to properly handle the high dimensionality of hyperspectral images. An example is a Bayesian segmentation procedure based on discriminative classifiers with a Multi-Level Logistic Markov-Gibbs prior. This method adopts the Fast Sparse Multinomial Logistic Regression as discriminative classifier, a method that promotes sparsity by including a Laplacian prior. However, the use of this type of prior requires an extensive search to for the best parameter of sparsity. In this work, a modification to this method is introduced. Instead of using the Laplacian Prior to enforce the sparsity of FSMLR classifier, the Jeffreys prior is used. This prior avoids the need to proceed to an extensive search for the best parameter, and also keeps the sparsity of the densities estimators, resulting on a faster and competitive segmentation procedure. The results of the application of this new approach to the benchmarked dataset Indian Pines show the effectiveness of the proposed method when compared with that using the Laplacian prior.

Abstract
Methods for learning sparse classification are among the state-of-the-art in supervised learning. Sparsity, essential to achieve good generalization capabilities, can be enforced by using heavy tailed priors/regularizers on the weights of the linear combination of functions. These priors/regularizers favour a few large weights and many to exactly zero. The Sparse Multinomial Logistic Regression algorithm [1] is one of such methods, that adopts a Laplacian prior to enforce sparseness. Its applicability to large datasets is still a delicate task from the computational point of view, sometimes even impossible to perform. This work implements an iterative procedure to calculate the weights of the decision function that is O(m(2)) faster than the original method introduced in [1] (m is the number of classes). The benchmark dataset Indian Pines is used to test this modification. Results over subsets of this dataset are presented and compared with others computed with support vector machines.

Abstract
Remote sensing techniques can be used to calculate suspended sediment concentrations and to understand the flux and distribution of sediments driven by mechanisms such as tides and waves, river discharges, etc. The main objective of this study is the quantification of the Total Suspended Matter (TSM) concentration in the sea breaking zone for a particular area of the Portuguese coast, around Aveiro. The methodology used was based on in situ measurements and multi spectral satellite images. In situ experimental techniques (maritime platform, aerial platform, simulation on the beach and water sample collection in the breaking zone) were used to determine a relationship between the TSM concentration and the seawater reflectance in the breaking zone. Spectral reflectance was measured with a spectroradiometer and water samples were simultaneously collected. Empirical relationships were established between TSM concentration and the equivalent reflectance values for sensors SPOT/HRVIR, TERRA/ASTER and Landsat/TM at visible and Near Infra Red (NIR) bands computed from the experimental data. Satellite images from ASTER, SPOT HRVIR and Landsat TM were used together with the same empirical models. These satellite images were calibrated and atmospherically corrected. Equations of linear, polynomial, logarithmic, power and exponential models were tested for the satellite image bands on the visible and near infrared. The coefficients of determination (R-2) were also calculated for each model. The results obtained from the two approaches, in situ measurements and directly from the multi spectral satellite images, were analysed.