This paper presents the analysis of Earth Observation data records collected between 1975 and 2014 for assessing the extent and temporal evolution of the built-up surface in the frame of the Global Human Settlement Layer project. The scale of the information produced by the study enables the assessment of the whole continuum of human settlements from rural hamlets to megacities. The study applies enhanced processing methods as compared to the first production of the GHSL baseline data. The major improvements include the use of a more refined learning set on built-up areas derived from Sentinel-1 data which allowed testing the added-value of incremental learning in big data analytics. Herein, the new features of the GHSL built-up grids and the methods are described and compared with the previous ones using a reference set of building footprints for 277 areas of interest. The results show a gradual improvement in the accuracy measures with a gain of 3.6% in the balanced accuracy, between the first production of the GHSL baseline and the latest GHSL multitemporal built-up grids. A validation of the multitemporal component is also conducted at the global scale establishing the reliability of the built-up layer across time.

Soil erosion can present a major threat to agriculture due to loss of soil, nutrients, and organic carbon. Therefore, soil erosion modelling is one of the steps used to plan suitable soil protection measures and detect erosion hotspots. A bibliometric analysis of this topic can reveal research patterns and soil erosion modelling characteristics that can help identify steps needed to enhance the research conducted in this field. Therefore, a detailed bibliometric analysis, including investigation of collaboration networks and citation patterns, should be conducted. The updated version of the Global Applications of Soil Erosion Modelling Tracker (GASEMT) database contains information about citation characteristics and publication type. Here, we investigated the impact of the number of authors, the publication type and the selected journal on the number of citations. Generalized boosted regression tree (BRT) modelling was used to evaluate the most relevant variables related to soil erosion modelling. Additionally, bibliometric networks were analysed and visualized. This study revealed that the selection of the soil erosion model has the largest impact on the number of publication citations, followed by the modelling scale and the publication's CiteScore. Some of the other GASEMT database attributes such as model calibration and validation have negligible influence on the number of citations according to the BRT model. Although it is true that studies that conduct calibration, on average, received around 30% more citations, than studies where calibration was not performed. Moreover, the bibliographic coupling and citation networks show a clear continental pattern, although the co-authorship network does not show the same characteristics. Therefore, soil erosion modellers should conduct even more comprehensive review of past studies and focus not just on the research conducted in the same country or continent. Moreover, when evaluating soil erosion models, an additional focus should be given to field measurements, model calibration, performance assessment and uncertainty of modelling results. The results of this study indicate that these GASEMT database attributes had smaller impact on the number of citations, according to the BRT model, than anticipated, which could suggest that these attributes should be given additional attention by the soil erosion modelling community. This study provides a kind of bibliographic benchmark for soil erosion modelling research papers as modellers can estimate the influence of their paper.

Large-scale land cover classification from satellite imagery is still a challenge due to the big volume of data to be processed, to persistent cloud-cover in cloud-prone areas as well as seasonal artefacts that affect spatial homogeneity. Sentinel-2 times series from Copernicus Earth Observation program offer a great potential for fine scale land cover mapping thanks to high spatial and temporal resolutions, with a decametric resolution and five-day repeat time. However, the selection of best available scenes, their download together with the requirements in terms of storage and computing resources pose restrictions for large-scale land cover mapping. The dataset presented in this paper corresponds to global cloud-free pixel based composite created from the Sentinel-2 data archive (Level L1C) available in Google Earth Engine for the period January 2017- December 2018. The methodology used for generating the image composite is described and the metadata associated with the 10 m resolution dataset is presented. The data with a total volume of 15 TB is stored on the Big Data platform of the Joint Research Centre. It can be downloaded per UTM grid zone, loaded into GIS clients and displayed easily thanks to pre-computed overviews.

The free, full and open data policy of the EU’s Copernicus programme has vastly increased the amount of remotely sensed data available to both operational and research activities. However, this huge amount of data calls for new ways of accessing and processing such “big data”. This paper focuses on the use of Copernicus’s Sentinel-1 radar satellite for maritime surveillance. It presents a study in which ship positions have been automatically extracted from more than 11,500 Sentinel-1A images collected over the Mediterranean Sea, and compared with ship position reports from the Automatic Identification System (AIS). These images account for almost all the Sentinel-1A acquisitions taken over the area during the two-year period from the start of the operational phase in October 2014 until September 2016. A number of tools and platforms developed at the European Commission’s Joint Research Centre (JRC) that have been used in the study are described in the paper. They are: (1) Search for Unidentified Maritime Objects (SUMO), a tool for ship detection in Synthetic Aperture Radar (SAR) images; (2) the JRC Earth Observation Data and Processing Platform (JEODPP), a platform for efficient storage and processing of large amounts of satellite images; and (3) Blue Hub, a maritime surveillance GIS and data fusion platform. The paper presents the methodology and results of the study, giving insights into the new maritime surveillance knowledge that can be gained by analysing such a large dataset, and the lessons learnt in terms of handling and processing the big dataset.

Motivated by the increasing availability of open and free Earth observation data through the Copernicus Sentinel missions, this study investigates the capacity of advanced computational models to automatically generate thematic layers, which in turn contribute to and facilitate the creation of land cover products. In concrete terms, we assess the practical and computational aspects of multi-class Sentinel-2 image segmentation based on a convolutional neural network and random forest approaches. The annotated learning set derives from data that is made available as result of the implementation of European Union’s INSPIRE Directive. Since this network of data sets remains incomplete in regard to some geographic areas, another objective of this work was to provide consistent and reproducible ways for machine-driven mapping of these gaps and a potential update of the existing ones. Finally, the performance analysis identifies the most important hyper-parameters, and provides hints on the models’ deployment and their transferability.