The performance prediction of an optical communications link over maritime environments has been extensively researched over the last two decades. The various atmospheric phenomena and turbulence effects have been thoroughly explored, and long-term measurements have allowed for the construction of simple empirical models. The aim of this work is to demonstrate the prediction accuracy of various machine learning (ML) algorithms for a free-space optical communication (FSO) link performance, with respect to real time, non-linear atmospheric conditions. A large data set of received signal strength indicators (RSSI) for a laser communications link has been collected and analyzed against seven local atmospheric parameters (i.e., wind speed, pressure, temperature, humidity, dew point, solar flux and air-sea temperature difference). The k-nearest-neighbors (KNN), tree-based methods-decision trees, random forest and gradient boosting- and artificial neural networks (ANN) have been employed and compared among each other using the root mean square error (RMSE) and the coefficient of determination (R2) of each model as the primary performance indices. The regression analysis revealed an excellent fit for all ML models, indicative of their ability to offer a significant improvement in FSO performance modeling as compared to traditional regression models. The best-performing R2 model found to be the ANN approach (0.94867), while random forests achieved the most optimal RMSE result (7.37).

Aerosols and clouds are the most important constituents in the atmosphere that affect the incoming solar radiation, either directly through absorbing and scattering processes or indirectly by changing the optical properties and lifetime of clouds. Under clear skies, aerosols become the dominant factor that affect the intensity of solar irradiance reaching the ground. Under cloudy skies, the high temporal and spatial variability of cloudiness is the key factor for the estimation of solar irradiance. In this study, recent research activities related to the climatology and the prediction of solar energy in Greece are presented with emphasis on new challenges in the climatology of global horizontal irradiance (GHI) and direct normal irradiance (DNI), the changes of DNI due to the decreasing aerosol optical depth and the short-term (15–240 min) forecasts of solar irradiance with the collaborative use of neural networks and satellite images.

A novel method for the short-term (15–240 min) forecasting of cloudiness in Greece is presented by taking into account that this is the main atmospheric factor responsible for the spatial and temporal distribution of surface solar irradiance. Images from the Spinning Enhanced Visible and Infrared Imager onboard the Meteosat Second Generation satellite, for a 3-year time period and with high spatial and temporal resolution (0.05°, 15 min), were processed to retrieve the cloud clearness index (CCI) and used for the training and testing of an artificial neural network (ANN). The estimated and the measured values of CCI are in good agreement and emphasis is given to the spatial distribution of the seasonal errors. The ANN was trained according to pre-classified areas that present similar cloud characteristics and could provide estimations of surface solar irradiance in synergy with models that calculate surface irradiance under clear skies.

Time-resolved characterization of solar irradiance at the ground level is a critical element in solar energy analysis. Siting of nodes in a network of solar irradiance monitoring stations (MS) is a multi-faceted problem that directly affects the determination of the solar resource and its spatio-temporal variability. The present work proposes an objective framework to optimize the deployment of solar MS over a sub-continental region. There are two main components in the proposed methodology. The first employs cluster analysis using the affinity propagation algorithm, to select the optimal number of clusters (regions with coherent solar microclimates) upon internal coherence criteria. The second component employs stochastic prediction and validation, through the use of a Bayesian maximum entropy method, and selects the optimal MS configuration, according to geostatistical criteria, among the solutions recommended by the cluster analysis. We apply this two-pronged methodology to determine clusters and optimal locations for global horizontal irradiance monitoring across the state of California. In this proof-of-concept study, 3 disparate MS configurations are examined within the cluster partition. The subsequent geostatistical analysis indicates that all configurations rank almost equally well based on different statistical error measures. The optimal configuration can be singled out depending on desired criteria of choice.

We propose and analyze a spatio-temporal correlation method to improve forecast performance of solar irradiance using gridded satellite-derived global horizontal irradiance (GHI) data. Forecast models are developed for seven locations in California to predict 1-h averaged GHI 1, 2 and 3 h ahead of time. The seven locations were chosen to represent a diverse set of maritime, mediterranean, arid and semi-arid micro-climates. Ground stations from the California Irrigation Management Information System were used to obtain solar irradiance time-series from the points of interest. In this method, firstly, we define areas with the highest correlated time-series between the satellite-derived data and the ground data. Secondly, we select satellite-derived data from these regions as exogenous variables to several forecast models (linear models, Artificial Neural Networks, Support Vector Regression) to predict GHI at the seven locations. The results show that using linear forecasting models and a genetic algorithm to optimize the selection of multiple time-lagged exogenous variables results in significant forecasting improvements over other benchmark models. (C) 2015 Elsevier Ltd. All rights reserved.

This work presents a cluster analysis for the determination of coherent zones of Global Horizontal Irradiance (GHI) for a utility scale territory in California, which is serviced by San Diego Gas & Electric. Knowledge of these coherent zones, or clusters, would allow utilities and power plants to realize cost savings through regional planning and operation activities such as the mitigation of solar power variability through the intelligent placement of solar farms and the optimal placement of radiometric stations. In order to determine such clusters, two years of gridded satellite data were used to describe the evolution of GHI over a portion of Southern California. Step changes of the average daily clear-sky index at each location are used to characterize the fluctuation of GHI. The k-means clustering algorithm is applied in conjunction with a stable initialization method to diminish its dependency to random initial conditions. Two validity indices are then used to define the quality of the cluster partitions as well as the appropriate number of clusters. The clustering algorithm determined an optimal number of 14 coherent spatial clusters of similar GHI variability as the most appropriate segmentation of the service territory map. In addition, 14 cluster centers are selected whose radiometric observations may serve as a proxy for the rest of the cluster. A correlation analysis, within and between the proposed clusters, based both on single-point ground-based and satellite-derived measurements evaluates positively the coherence of the conducted clustering. This method could easily be applied to any other utility scale region and is not dependent on GHI data which shows promise for the application of such clustering methods to load data and/or other renewable resources such as wind. (C) 2014 Elsevier Ltd. All rights reserved.

We propose a novel methodology to select candidate locations for solar power plants that take into account solar variability and geographical smoothing effects. This methodology includes the development of maps created by a clustering technique that determines regions of coherent solar quality attributes as defined by a feature which considers both solar clearness and solar variability. An efficient combination of two well-known clustering algorithms, the affinity propagation and the k-means, is introduced in order to produce stable partitions of the data to a variety of number of clusters in a computationally fast and reliable manner. We use 15 years worth of the 30-min GHI gridded data for the island of Lanai in Hawaii to produce, validate and reproduce clustering maps. A family of appropriate number of clusters is obtained by evaluating the performance of three internal validity indices. We apply a correlation analysis to the family of solutions to determine the map segmentation that maximizes a definite interpretation of the distinction between and within the emerged clusters. Having selected a single clustering we validated the clustering by using a new dataset to demonstrate that the degree of similarity between the two partitions remains high at 90.91%. In the end we show how the clustering map can be used in solar energy problems. Firstly, we explore the effects of geographical smoothing in terms of the clustering maps, by determining the average ramp ratio for two location within and without the same cluster and identify the pair of clusters that shows the highest smoothing potential. Secondly, we demonstrate how the map can be used to select locations for GHI measurements to improve solar forecasting for a PV plant, by showing that additional measurements from within the cluster where the PV plant is located can lead to improvements of 10% in the forecast. (C) 2014 Elsevier Ltd. All rights reserved.

Methods applied on efficient planning of ground-based monitoring networks of surface solar irradiance could provide valuable scientific results and be useful for accurate monitoring and efficient planning of solar energy applications. Based on the dominance of cloud effect on solar irradiance and the advantage of the high spatial resolution of a geostationary satellite, a novel method is presented for optimizing the location of measuring sites for the newly built Hellenic Network of Solar Energy (www.helionet.gr). The k-means algorithm is used for cluster analysis and the validation of the clustering method reveals that the variability of surface solar irradiance due to cloudiness over Greece could be sufficiently monitored with the establishment of 22 ground-based instruments. The spatial representativeness of the proposed sites is also assessed. The proposed number of stations could be considered as the basis to build the climatology of surface solar irradiance over Greece. (C) 2013 Elsevier Ltd. All rights reserved.

Methods applied on efficient planning of ground-based monitoring networks of surface solar ultraviolet (UV) irradiance could provide valuable scientific results and be useful for accurate monitoring and assessment of climatic values for UV-health studies. Based on the dominance of cloud effect on solar irradiance and the advantage of the high spatial resolution of a geostationary satellite, a novel method is presented for optimizing the location of UV measuring sites. The k-means algorithm is used for cluster analysis and the validation of the clustering method reveals that the variability of surface solar UV irradiance due to cloudiness over Greece could be sufficiently monitored with the establishment of 23-28 ground-based instruments.

The paper presents a semi-supervised weather classification method based on 850-hPa isobaric level maps. A preprocessing step is employed, where isolines of geopotential height are extracted from weather map images via an image processing procedure. Α feature extraction stage follows where two techniques are applied. The first technique implements phase space reconstruction, and yields multidimensional delay distributions. The second technique is based on chain code representation of signals, from which histogram features are derived. Similarity measures are used to compare multidimensional data and the k-means algorithm is applied in the final stage. The method is applied over the area of Greece, and the resulting catalogues are compared to a subjective classification for this area. Numerical experiments with datasets derived from the European Meteorological Bulletin archives exhibit an up to 91 % accurate agreement with the subjective weather patterns.

Classification of weather maps at various isobaric levels as a methodological tool is used in several problems related to meteorology, climatology, atmospheric pollution and to other fields for many years. Initially the classification was performed manually. The criteria used by the person performing the classification are features of isobars or isopleths of geopotential height, depending on the type of maps to be classified. Although manual classifications integrate the perceptual experience and other unquantifiable qualities of the meteorology specialists involved, these are typically subjective and time consuming. Furthermore, during the last years different approaches of automated methods for atmospheric circulation classification have been proposed, which present automated and so-called objective classifications. In this paper a new method of atmospheric circulation classification of isobaric maps is presented. The method is based on graph theory. It starts with an intelligent prototype selection using an over-partitioning mode of fuzzy c-means (FCM) algorithm, proceeds to a graph formulation for the entire dataset and produces the clusters based on the contemporary dominant sets clustering method. Graph theory is a novel mathematical approach, allowing a more efficient representation of spatially correlated data, compared to the classical Euclidian space representation approaches, used in conventional classification methods. The method has been applied to the classification of 850hPa atmospheric circulation over the Eastern Mediterranean. The evaluation of the automated methods is performed by statistical indexes; results indicate that the classification is adequately comparable with other state-of-the-art automated map classification methods, for a variable number of clusters.

Atmospheric circulation plays a major role in the stable isotopic composition of precipitation. In this study, the relationship between the synoptic patterns and the stable isotopic composition (δ18O & δ2H) of precipitation is investigated using event-based rainfall data. The aim of this paper is the generation of isotopic time series using a combined synoptic classification technique. Using the classification software developed within the COST733 action, we generated synoptic catalogues utilizing various classification methods for two meteorological parameters: the geopotential height at the isobaric level of 500 hPa and the thickness (500–1,000 hPa) and we propose an efficient technique to generate a representative classification catalogue based on the above parameters. An ANN is trained using this catalogue in order to classify each one of the meteorological parameters. The output scheme is compared with the initial catalogues of the COST733 action using statistical indices both in terms of explaining the variance of the classified meteorological fields and in terms of providing classes with statistically distinct isotopic signatures. Finally, using the proposed classification, the isotopic composition of the synoptic classes is determined and used to reconstruct isotopic time series.

Weather maps refer to meteorological data that characterize the atmospheric circulation in a region. The classification of weather maps into categories becomes an important task for understanding regional climate. Towards this goal, manual and semiautomatic techniques have been used, requiring manpower and supervision. In this paper, we propose a machine vision based method for the classification of weather maps into distinct classes. The chain code descriptor is applied to extract the feature of isobaric lines and we introduce the Double-Side Chain Code (DSCC) histogram for feature representation. Handling DSCC histograms as multidimensional vectors, the k-nearest neighbors (k- NN) algorithm classifies the objects to an appropriate number of classes, based on closest training set in the feature space. This method provides an automated and more 'objective' classification scheme, applying straightforward to the input weather map's image.

Locally Linear Embedding (LLE), Isometric Mapping Isomap) are two relatively new nonlinear dimensionality reduction algorithms also used in face recognition applications. Their main aim is to create a low-dimensional embeddings of the original high-dimensional data, laying the face data points on a 'face manifold'. In this work in order to test their performance we applied LLE and Isomap in two face databases together with principal component analysis (PCA), their linear counterpart, varying as parameters the (i) number embedding dimensions and (ii) the number of neighbours. Furthermore, at the final stage we used a data ranking algorithm, which ranks the data with respect to the intrinsic manifold structure and its geometric properties. Experimental results indicate the superiority of the data ranking algorithm on face manifolds against the classical Euclidean distance measure.