Search

Motivated by the increasing integration among electricity markets, in this paper we propose two different methods to incorporate market integration in electricity price forecasting and to improve the predictive performance. First, we propose a deep neural network that considers features from connected markets to improve the predictive accuracy in a local market. To measure the importance of these features, we propose a novel feature selection algorithm that, by using Bayesian optimization and functional analysis of variance, evaluates the effect of the features on the algorithm performance. In addition, using market integration, we propose a second model that, by simultaneously predicting prices from two markets, improves the forecasting accuracy even further. As a case study, we consider the electricity market in Belgium and the improvements in forecasting accuracy when using various French electricity features. We show that the two proposed models lead to improvements that are statistically significant. Particularly, due to market integration, the predictive accuracy is improved from 15.7% to 12.5% sMAPE (symmetric mean absolute percentage error). In addition, we show that the proposed feature selection algorithm is able to perform a correct assessment, i.e. to discard the irrelevant features. [ABSTRACT FROM AUTHOR]

Abstract: The present paper addresses the issue of urban sprawl in Europe from a multidimensional point of view, identifying the most sprawled areas and characterizing them in terms of population size. The literature is reviewed to categorize and extract the most relevant six dimensions that define the concept and several indices are specified to implement them. These are then calculated for a sample of the main European cities that uses several sources to obtain the best possible dataset to measure urban sprawl. All this information is brought together using the self-organizing map (SOM) algorithm to be visualized and further studied, taking advantage of its properties as a data-reduction as well as a clustering technique. The analysis locates the hot-spots of urban sprawl in Europe in the centre of the continent, around Germany, and characterizes such urban areas as small, always half the size of the average city of the sample. [Copyright &y& Elsevier]

Abstract: This paper describes a methodology to explore the (future) spatial distribution of biofuel crops in Europe. Two main types of biofuel crops are distinguished: biofuel crops used for the production of biodiesel or bioethanol, and second-generation biofuel crops. A multi-scale, multi-model approach is used in which biofuel crops are allocated over the period 2000–2030. The area of biofuel crops at the national level is determined by a macro-economic model. A spatially explicit land use model is used to allocate the biofuel crops within the countries. Four scenarios have been prepared based on storylines influencing the extent and spatial distribution of biofuel crop cultivation. The allocation algorithm consists of two steps. In the first step, processing plants are allocated based on location factors that are dependent on the type of biofuel crop processed and scenario conditions. In the second step, biofuel crops are allocated accounting for the transportation costs to the processing plants. Both types of biofuel crops are allocated separately based on different location factors. Despite differences between the scenarios, mostly the same areas are showing growth in biofuel crop cultivation in all scenarios. These areas stand out because they have a combination of well-developed infrastructural and industrial facilities and large areas of suitable arable land. The spatially explicit results allow an assessment of the potential consequences of large-scale biofuel crop cultivation for ecology and environment. [Copyright &y& Elsevier]

Abstract: This purposes of this paper are to: 1) demonstrate an objective overshooting top (OT) detection method using Meteosat Second Generation (MSG) Spinning Enhanced Visible and Infrared Imager (SEVIRI) infrared data, 2) produce an OT database for all operational SEVIRI data over Europe and north Africa for the six Northern Hemisphere warm seasons observed by SEVIRI to date, and 3) determine the frequency of OT detections in the vicinity of confirmed severe weather reports recorded within the European Severe Weather Database (ESWD). Algorithm performance is demonstrated for a case where numerous severe storms were present. Qualitative comparisons indicate that most OT detections correspond with the characteristic OT signature in SEVIRI 1km resolution visible channel imagery, but some OTs were left undetected due in part to relatively coarse SEVIRI spatial resolution over the European domain. The SEVIRI OT database shows a strong relationship between OT maxima and regions with high terrain. OTs are found to occur more frequently during the day over land and during the night over water. Inter- and intra-seasonal variability in OT frequency and location are also shown. An OT was found near 47% of the confirmed ESWD events. The OT–severe weather relationship is strong for large hail (53%) and severe wind (52%) events but relatively weak for tornado events (14%). The weak OT–tornado relationship may be related two factors: 1) low-level wind shear is found to be of greater importance than large CAPE and strong updrafts (i.e. OTs) in tornadic storm environments across Europe and 2) a weakening of the storm updraft and collapse of the OT region has been documented prior to tornado formation. The relatively strong overall OT–severe weather relationship suggests that OT detections can be used to increase forecaster confidence that a given storm is severe, especially in regions where frequent, ground based Doppler weather radar data is unavailable. [ABSTRACT FROM AUTHOR]

Abstract: This paper describes the evaluation and performance of the Medium Resolution Imaging Spectrometer (MERIS) Global Vegetation Index (MGVI) algorithm that is implemented in the MERIS ground segment as the primary land surface product. MGVI output values represent the Fraction of Absorbed Photosynthetically Active Radiation (FAPAR) which acts as an indicator of the presence and state of the vegetation canopy. The retrieval algorithm was optimized to be insensitive to the overlying atmosphere, the underlying soil as well as angular effects, using radiative transfer models. This physically based approach for retrieving land biophysical parameters can be extended to a series of sensors; the resulting algorithms are designed to deliver similar geophysical products that are directly comparable and ultimately generate long time series of FAPAR. After presenting the MGVI algorithm, we analyze actual results by inter-comparing the FAPAR values derived from MERIS to similar products derived from the Sea-viewing Wide Field-of-view Sensor (SeaWiFS) that have been generated at the European Commission Joint Research Centre (EC-JRC). The FAPAR products derived from MERIS and SeaWiFS are compared on several dates in 2002 when both instruments simultaneously observed the same geographical regions. The consistency and temporal continuity of the products are then evaluated by comparing the time series of FAPAR values over local sites in Europe during the year 2003 from the two instruments. Finally, multi-annual FAPAR time series obtained by merging MERIS and SeaWiFS products demonstrate the feasibility of monitoring the state of land surfaces with FAPAR products before and after the European drought event in 2003. [Copyright &y& Elsevier]

• Timeseries representational method that presents the collective trend of the data. • A user defined segmentation method of time-series data. • Probabilistic approach of forming representational building blocks for timeseries. • Unsupervised trend based hierarchical clustering of timeseries data. Time series data is a sequence of values recorded systematically over a period which are mostly used for prediction, clustering, and analysis. The two essential features of a time series data are trend and seasonality. Preprocessing of the time series data is necessary for performing prediction tasks. In most of the cases, the trend and the seasonality are removed before applying the regression algorithms. The accuracy of such algorithms depends upon the functions used for the removal of trend and seasonality. Clustering of an unlabeled time series data with the presence of trend and seasonality is challenging. In this paper, we propose a probabilistic representational learning method for grouping the time series data. We introduce five terminologies in our method of clustering namely the trendlets, uplets, downlets, equalets and trendlet string. These elements are the representational building blocks of our proposed method. Experiments on the proposed algorithm are performed with the renewable energy data on the electricity supply system of continental Europe which includes the demand and inflow of renewable energy for the term 2012 to 2014 and UCR-2018 time series archive containing 128 datasets. We compared our proposed representational method with various clustering algorithms using the silhouette score. Mini-batch k-means and agglomerative hierarchical clustering algorithms show better performance in terms of quality, logical accordance with data and time taken for clustering. [ABSTRACT FROM AUTHOR]