Abstract
This paper presents a new solution approach to solve the resource-constrained project scheduling problem in the presence of three types of logical constraints. Apart from the traditional AND constraints with minimal time-lags, these precedences are extended to OR constraints and bidirectional (BI) relations. These logical constraints extend the set of relations between pairs of activities and make the RCPSP definition somewhat different from the traditional RCPSP research topics in literature. It is known that the RCPSP with AND constraints, and hence its extension to OR and BI constraints, is NP-hard. The new algorithm consists of a set of network transformation rules that removes the OR and BI logical constraints to transform them into AND constraints and hereby extends the set of activities to maintain the original logic. A satisfiability (SAT) solver is used to guarantee the original precedence logic and is embedded in a metaheuristic search to resource feasible schedules that respect both the limited renewable resource availability as well as the precedence logic. Computational results on two well-known datasets from literature show that the algorithm can compete with the multi-mode algorithms from literature when no logical constraints are taken into account. When the logical constraints are taken into account, the algorithm can report major reductions in the project makespan for most of the instances within a reasonable time.

Abstract
Software Defined Networks (SDNs) provides a separation between the control plane and the forwarding plane of networks. The software implementation of the control plane and the built in data collection mechanisms of the OpenFlow protocol promise to be excellent tools to implement Machine Learning (ML) network control applications. A first step in that direction is to understand the type of data that can be collected in SDNs and how information can be learned from that data. In this work we describe a simple architecture deployed in an enterprise network that gathers traffic data using the OpenFlow protocol. We present the data-sets that can be obtained and show how several ML techniques can be applied to it for traffic classification. The results indicate that high accuracy classification can be obtained with the data-sets using supervised learning.

Abstract
Satellite altimetry allows the study of sea-level long-term variability on a global and spatially uniform basis. Here quantile regression is applied to derive robust median regression trends of mean sea level as well as trends in extreme quantiles from radar altimetry time series. In contrast with ordinary least squares regression, which only provides an estimate on the rate of change of the mean of data distribution, quantile regression allows the estimation of trends at different quantiles of the data distribution, yielding a more complete picture of long-term variability. Trends derived from basin-wide averaged regional mean sea level time series are robust and similar for all quantiles, indicating that all parts of the data distribution are changing at the same rate. In contrast, trends are not robust and diverge across quantiles in the case of local time series. Trends are under- (over-)estimated in the western (eastern) equatorial Pacific. Furthermore, trends in the lowermost quantile (0.05) are larger than the median trend in the western Pacific, while trends in the uppermost quantile (0.95) are lower than the median trend in the eastern Pacific. These differences in trends in extreme mean sea level quantiles are explained by the exceptional effect of the strong 1997-1998 El Nino-Southern Oscillation (ENSO) event.

Abstract
The classification ofmultivariate time series in terms of their corresponding temporal dependence patterns is a common problem in geosciences, particularly for large datasets resulting from environmental monitoring networks. Here a wavelet-based clustering approach is applied to sea level and atmospheric pressure time series at tide gauge locations in the Baltic Sea. The resulting dendrogram discriminates three spatially-coherent groups of stations separating the southernmost tide gauges, reflecting mainly high-frequency variability driven by zonal wind, from the middle-basin stations and the northernmost stations dominated by lower-frequency variability and the response to atmospheric pressure.

Abstract
The seasonal cycle accounts for about 40 % of the total sea level variability in the Baltic Sea. In a climate change context, changes are expected to occur, not only in mean levels but also in the seasonal characteristics of sea level. The present study addresses the quantification of changes in the seasonal cycle of sea level from a set of century-long tide gauge records in the Baltic Sea. In order to obtain robust estimates of the changes in amplitude and phase of the seasonal cycle, we apply different methods, including continuous wavelet filtering, multi-resolution decomposition based on the maximal overlap discrete wavelet transform, auto-regressive-based decomposition, singular spectrum analysis and empirical mode decomposition. The results show that all methods generally trace a similar long-term variability of the annual cycle amplitudes, and we focus on discrete wavelet analysis as the natural counterpart of classical moving Fourier analysis. In contrast to previous studies suggesting the existence of long-term changes in the seasonal cycle, in particular an increase of the annual amplitude, we find alternating periods of high and low amplitudes without any clear indication of systematic long-term trends. The derived seasonal patterns are spatially coherent, discriminating the stations in the Baltic entrance from the remaining stations in the Baltic basin, for which zonal wind accounts for typically more than 40 % of the variations in amplitude.

Abstract
Atmospheric Electric Potential Gradient (PG) measurements were carried out in three sites forming a triangular array in Southern Portugal. The campaign was performed during the summer, characterized by Saharan dust outbreaks; 16th-17th July 2014 dust event is considered. Short time-scale oscillations of the PG at two of the stations and a mid time-scale suppression of the PG in the three stations are found. Results are interpreted as evidencing long-range dust electrification; attributed to the air-Earth electrical current creating a bipolar charge distribution inside of the dust layer. The relevance of using arrays of sensors, instead of single sited, is highlighted.