The current land cover mapping paradigm relies on automatic classification of satellite images, with supervised methods being the most used, implying training data to have a crucial role. Aspects such as training sample size and quality should be carefully considered. This paper proposes assessing the use of a detailed class nomenclature to reinforce class diversity in the training sample. A Random Forest (RF) classification of Sentinel-2 multi-temporal data was conducted. Additionally, the effect of sample size and class distribution were evaluated. The results indicate that the use of a detailed nomenclature provided better results in terms of classification accuracy. With respect to sample distribution, adopting class sizes proportional to their occurrence in a reference land cover map exhibited superior performance in comparison to an equal size approach. The effect of sample size on classification performance was limited, as previous studies with RF suggested.

This paper presents an experimental crop classification of the 10 most abundant annual crop types in Portugal, using a study area located in Alentejo region. This region has great diversity of land uses as well as multiple crop types. Sentinel-2 2018 intra-annual time-series imagery is considered in the experiment. The Portuguese Land Parcel Identification System (LPIS) is used to extract automatic training samples. LPIS information is automatically processed with the help of auxiliary datasets to filter out crop areas more likely to have been mislabeled. Classification is obtained using random forest. Validation is performed using an independent dataset also based on LPIS. A global accuracy of 76% is obtained. The novelty of the methodology here presented shows that LPIS can be used together with auxiliary data for crop type mapping, helping to characterize the agriculture land diversity in Portugal.

Southern Portugal is characterized by disperse tree cover of Cork and Holm oaks in an agro-forestry system known as montado. Mapping these trees has been historically very difficult as they occur in isolation or in groups with different understory vegetation, including grass and shrubland. Automatic classification for binary tree/non-tree map production has been used elsewhere, but with limited success in the context of montado. Here, the potential of Sentinel-2 data was explored to map oaks using pure and mixed pixels to train a random forest. The output depicts a gradient of tree cover that can be transformed into a crisp map. The accuracy assessment of the latter shows commission and omission errors of 17% and 18%.

Tree species classification with satellite data has become more and more popular since Sentinel-2 launch. We compared efficacy and effectiveness of Extreme Gradient Boosting (XGBoost) and Light Gradient Boosting Machine (LGBM) with widely used in remote sensing Random Forest (RF), Support Vector Machine (SVM) and K-Nearest Neighbour (KNN) algorithms. Analyses were performed over an area in Portugal with multi-temporal Sentinel-2 data registered in April, June, August and October 2018. The selected classes were: cork oak, holm oak, eucalyptus, other broadleaved, maritime pine, stone pine and other coniferous. Algorithm efficacy was measured through F1-score and accuracy while efficiency was measured through the median time needed for each fit. XGBoost and LGBM outperformed efficacy of other algorithms, which was already high (above 90% for the best variant of each algorithm). In terms of efficacy, LGBM overcame all algorithms, including XGBoost.

Classification accuracy of remote sensing images with supervised learning depends on the quality and characteristics of training samples. Size is a key aspect of a sample and its impact on classification depends on several factors, including the classifier employed, dimension on the feature space and land cover characteristics. Random Forest classifier is considered to be of low sensitivity to variations in sample size. However, further investigation is required when feature spaces are large and training is performed with spectral subclasses of the land cover classes to be mapped. This paper proposes to assess the impact of sample size in the classification accuracy of Random Forest using multitemporal Sentinel-2 data and a detailed set of training subclasses to produce a map with general land cover classes. The results revealed similar classification accuracies after major reductions in sample size.

In this article, a methodology to retrieve the precipitable water vapor (PWV) from a differential interferometric time series is presented. We used external data provided by atmospheric weather models (e.g., ERA-Interim reanalysis) to constrain the initial state and by Global Navigation Satellite System (GNSS) to phase ambiguities elimination introduced by phase unwrapping algorithm. An iterative least-square is then used to solve the optimization problem. We applied the presented methodology to two time series of differential PWV maps estimated from synthetic aperture radar (SAR) images acquired by the Sentinel-1A, over the southwest part of the Appalachian Mountains (USA). The results were validated using an independent GNSS data set and also compared with atmospheric weather prediction data. The GNSS PWV observations show a strong correlation with the estimated PWV maps with a root-mean-square error less than 1 mm. These results are very encouraging, particularly for the meteorology community, providing crucial information to assimilate into numerical weather models and potentially improve the forecasts.

A Global Navigational Satellite System (GNSS) tomography experiment has been performed for 1 week, introducing Atmospheric Infrared Sounder (AIRS) remote sensing data to initiate and update a 3D wet refractivity hourly solution series of the troposphere. Some qualitative and quantitate studies have been performed, taking advantage of a local radiosonde campaign with a 4-hour sampling data. 3D wet refractivity maps with an accuracy close to 2 g/m3are obtained.

Global Navigation Satellite System (GNSS) tomography provides 3-D reconstructions of atmosphere wet refractivity, related to water vapor. A simulated analysis of the integration of Global Positioning System and future Galileo data is presented. Atmospheric refractivity is derived from radiosonde data acquired over the Lisbon area. The impact of Galileo data on the tomographic reconstruction is assessed. Furthermore, horizontal anomalies are added to a reference vertical profile of atmospheric refractivity to reproduce low-level dry or wet air intrusions, a phenomenon commonly observed in meteorological data acquired by both radiosonde and satellites. The dependence of tomographic solution on the GNSS network density is also analyzed. Better reconstruction capabilities in the lower layers are observed when increasing the network density.

The integration of interferometric synthetic aperture radar (InSAR) and GPS tomography techniques for the estimation of the 3-D distribution of atmosphere refractivity is discussed. A methodology to use the maps of the temporal changes of precipitable water vapor (PWV) provided by InSAR as a further constraint in the GPS tomography is described. The aim of the methodology is to increase the accuracy of the GPS tomography reconstruction of the atmosphere's refractivity. The results, which are obtained with SAR and GPS data acquired over the Lisbon area, Portugal, are presented and assessed. It has been found that the reconstruction of the atmospheric refractivity is closer to the real atmospheric state with a mitigation of the smoothing effects due to the usual geometrical constraints of the GPS tomography.

GPS tomography provides a unique opportunity to sense the 3D state of the atmosphere. However, the setting of the model grid can affect the water vapor solution since GPS data can be insufficient to cover all the domain, leading to an ill-posed conditioning, usually solved by constraints. In this work we present the result of a simulation analysis to study the impact of Galileo data on the reconstruction of 3D atmospheric water vapor. GPS tomography results are compared with those provided by merging GPS and Galileo data in order to ascertain the enhancement of the 3D water vapor refractivity reconstruction.

Microwave sensing of the atmosphere with GPS data, particularly using GPS tomography, provides a unique opportunity to measure the 3D state of the atmospheric water vapor, since it acquires data in a high temporal sampling. However, this technique requires a full domain coverage which is not fulfilled by GPS data, leading to an ill-posed conditioning only solved by constraints. SAR interferometry can be used to get integrated water vapor maps with high spatial resolution, but with a temporal frequency depending on the SAR acquisitions. In this work we describe a methodology to include integrated water vapor maps provided by SAR interferometry into the GPS tomography processing scheme. The differential water vapor spatial distribution occurred between the acquisition time of master and slave images is used to constrain the GPS tomography solution. Results are obtained over Lisbon area using GPS and ENVISAT-ASAR data and validated with regional radiosonde data.