Day ahead irradiance forecast variability characterization using satellite data

Solar variability corresponds to strong variations of the solar irradiance, caused mainly by the presence of clouds. Practical uses of solar resource data, such as the design of photovoltaic solar plants, usually employ several years of hourly data, neglecting subhourly features. The effect of clouds on short-time variability can differ by cloud type, suggesting that some cloud effects could be ignored when working with hourly data. In this work, we compare statistics of solar variability calculated at different time intervals and separate the analysis by cloud categories. We use 1 min solar data and cloud radar products from the Atmospheric Radiation Measurement (ARM) cloud, aerosol, and complex terrain interactions campaign in Córdoba, Argentina, where a wide variety of clouds exist. We classify the clouds based on their vertical position and observe solar variability using the mean and standard deviation of the clear sky index for varying time intervals of 5, 15, 30, and 60 min. Time intervals affect the mean and standard deviation of the clear sky index differently for each cloud type: longer time intervals neglect small variability and overestimate the mean clear sky index of low and mid-clouds, while high clouds do not change as much. The effect is also palpable when measuring ramps: the percentile 95 of the ramps obtained for 1 min is 21 times greater compared to 1 h. This ratio varies per cloud type with the strongest differences occurring for mid-clouds, having ramps that are 73 times stronger.

Satellite data have the potential to significantly enhance railway operations and drive the digitization of the rail sector. In the context of railways, satellite data primarily refers to the use of Global Navigation Satellite System (GNSS) data for applications such as navigation, positioning, and signalling. However, remote sensing data from Earth Observation (EO) satellites remain comparatively underutilized in railway applications. While the use of GNSS data in railways is well documented in the literature, research on EO-based remote sensing methods remains relatively limited. This paper aims to bridge this gap as it presents a comprehensive review of the use of satellite data in railway applications, with a particular focus on the underexplored potential of EO data. It provides the first in-depth analysis of EO techniques, primarily examining the use of synthetic aperture radar (SAR) and optical satellite data for key applications for infrastructure managers and railway operators, such as assessing track stability, detecting deformations, and monitoring surrounding environmental conditions. The goal of this review is to explore the diverse range of EO-based applications in railways and to identify emerging trends, including the integration of thermal EO data and the novel use of SAR for dynamic and predictive analyses. By synthesizing existing research and addressing knowledge gaps, the presented review underscores the potential of EO data to transform railway infrastructure management. Enhanced spatial resolution, frequent revisit cycles, and advanced AI-driven analytics are highlighted as key enablers for safer, more reliable, and cost-effective solutions. This review provides a framework for leveraging EO data to drive innovation and improve railway monitoring practices.

Solar irradiance forecasting using a ground-based sky imager developed at UC San Diego

Separated irradiance components are used to calculate the plane-of-array irradiance which is necessary for accurately predicting the energy output of solar photovoltaic (PV) systems. Solar irradiance decomposition or separation models are required for calculating the diffuse fraction ( K d ), direct normal irradiance (DNI), and diffuse horizontal irradiance (DIF) from global horizontal irradiance (GHI) measurements. To improve K d estimates, we introduce a novel, simple logistic multipredictor model. The new model only requires measured GHI and calculable predictors: clear-sky GHI, a clearness index ( K t ), apparent solar time, solar zenith angle, deviation in clear-sky K t , a clear-sky index, a persistence index, and a cloud enhancement index. Using 1-yr quality-controlled 1-min data of K d , DIF, and DNI from 14 stations spread across all 6 climatological regions of South Africa, the new model and the other 14 previously developed K d models are comprehensively validated. Error metrics and a mean linear ranking method determine the best model for estimating K d , DNI, and DIF, based on the sky condition, site, climate zone, and overall, in South Africa. Under all skies, the new model is proven to provide the most accurate K d , DIF, and DNI estimates. The new model, Engerer3, and Louche models are superior at estimating DNI, K d , and DIF under clear skies, respectively. In overcast skies, the nested model is superior at estimating K d , while the Yang4 model is well suited for predicting DIF and DNI. To ascertain the new model’s broader applicability, we recommend validation in other climate zones and geographical locations. Significance Statement This study is conducted in a region where solar decomposition models have not been developed at a very high temporal resolution and with a diverse spectrum of weather systems due to altitude variations and wide proximity ranges to coastlines. The presence of six unique climatic zones in South Africa underscores regional diversity, yet the study’s results hold applicability beyond the southern African context. Decomposition modeling is an essential stage in the model chain and, hence, extremely valuable for solar resource assessment and forecasting. Given that final uncertainties in solar resource estimates correlate with financial risks in utility-scale projects, a reduction of uncertainty in decomposition models is critical for reducing the uncertainty of solar photovoltaic (PV) energy forecasts.

Two probabilistic forecasting models for the clear-sky index, based on the Markov-chain mixture distribution (MCM) and copula clear-sky index generators, are presented and evaluated. In terms of performance, these models are compared with two benchmark models: a Quantile Regression (QR) model and the Persistence Ensemble (PeEn). The models are tested on minute resolution clear-sky index data, which was estimated from irradiance data for two different climatic regions: Hawaii, USA and Norrkoping, Sweden. Results show that the copula model generally outperforms the PeEn, while the MCM and QR models are superior in all tested aspects. Comparing MCM and QR reliability, the QR is superior, while the MCM is superior in mean CRPS and skill score. The MCM model is proposed as a potential benchmark for probabilistic solar forecasting. The MCM model is available in Python as SheperoMah/MCM-distribution-forecasting at GitHub.

The validation of optical satellite data products is a central but challenging component of the space missions. In order to validate the satellite images, ground data is used for reference and allows also the assessment of the associated total uncertainty budget. Overall, when comparing ground data and satellite measurements three main uncertainty sources need to be considered: i) instrument characterisation, ii) algorithm retrieval performances and iii) spatial representativeness. These key components affect the proper comparison of ground measurements with satellite data and, thus, have to be carefully examined. JB devices (FloX and RoX) are hyperspectral instruments acquiring optical field data with standardized hardware and routines. They have collected a legacy of data for over half of a decade using a comprehensive and readily implemented open-source data processing chain, considering the individual laboratory characterization of each instrument’s optical performance. Thus, the instruments are capable of providing valuable data products for the purpose of satellite validation. In particular, the FloX (Fluorescence BoX, JB Hyperspectral Devices GmbH) is the first commercially available device for the measurement of solar-induced chlorophyll fluorescence (SIF). The instrument was developed with the support of the scientific community following the specification of the Fluorescence Explorer mission (FLEX) by the European Space Agency (ESA), expected to be launched in 2024. The FloX features a high performing spectrometer (FWHM: 0.3 nm, SSI: 0.15, SNR: 1000) and allows stand-alone measurement of SIF emission at canopy level on the ground. Furthermore, the FloX enables the continuous measurements of spectral down-welling and up-welling radiance in the VIS-NIR range using an additional spectrometer to cover a larger spectral range and allows the automatic computation of reflectance as well as various vegetation indices (VIs). The instrument synchronously acquires upwelling and downwelling radiance during each measurement cycle, automatically optimizes the integration time according to light conditions and acquires the dark current and internal quality flags to ensure high quality data products. In addition to SIF and VIs, the FloX produces time series of high-resolution radiometric parameters, suitable for the investigation of the optical properties from the monitored targets. In the last years over 60 FloX units have been deployed worldwide.Within a current ESA project, we are investigating the instrument uncertainty sources, with the final aim of defining a preliminary version of the FLEX validation plan. At the same time, currently deployed instruments in 10 location around the world were used to examine the agreement of the ground measurements with available satellite product (i.e. Sentinel-2). This approach reversed the common practice of validating satellite data with ground measurements by using the globally available, standardized L2A products of Sentiel-2 evaluating the conformance of ground-measured data products across a network of standardized instruments. An unprecedented alignment of satellite and ground data was achieved, confirming high validity of data products from the network of automated field spectrometers around the globe.In summary, in this contribution we provide an overview of how field spectroscopy systems can be used in the framework of specific activities with the purpose of satellite validation.

A new statistical approach for deriving global solar radiation from satellite images

Post-processing of solar irradiance forecasts from WRF model at Reunion Island

Predictions of photovoltaic (PV) energy supply, based on data statistics of weather and PV records, are required in short‐intra and day‐ahead planning of PV plant operations. Numerical weather prediction (NWP), based on physical consideration, can simulate the progress of local cloudiness, although their prognoses are usually delayed by a few hours and are not provided with the quality desired by PV operators. Differential polynomial neural network (D‐PNN) is an unconventional hybrid regression technique, based on a novel learning strategy and able to resolve some problems in modeling patterns in high‐dynamical and chaotic non‐deterministic systems. D‐PNN splits the general partial differential equation (PDE) of the kth order into a summation form of two‐variable PDEs of a predetermined low order. PDE derivatives are transformed and separated into the Laplace counterparts, from which the unknown node originals are restored using the inverse L‐operation. D‐PNN searches for the best 2‐input couples, inserting node by node into its changeable tree structure, to expand the modular sum model by usable PDE components. The PDE‐modularization enhances the D‐PNN capability in modeling high uncertainty in weather patterns in the ground‐layer atmosphere, analogous to NWP. The designed PV intra‐ and day‐ahead prediction schemes, using hourly increasing and fixed‐time prediction horizons, were compared with different computing approaches based on differential, probabilistic and statistical learning. Machine learning (ML) models calculate the output PV power in each single hour separately or in an all‐day processing series with trained input delay. The results are analyzed and interpreted for each day and prediction hour. The quantitative results for both prediction procedures are comparable, showing the average intraday 11.6% and all‐day 9.6% (or rather 12.3% respectively) inaccuracy with respect to the daily peak power. Training data samples with spatial overlap in the regional scope and ML initialized similarity intervals are related to the fixed‐day or hourly increased input–output delay. This approach can compensate for unpredictable midterm variances in local weather patterns (caused by terrain or convection anomalies) and possible errors in model initialization. PV power was converted to and restored from the clear sky index in the modeling input/output, which represents its relative values, disregarding the actual solar cycle time.

An accurate estimate of sun irradiance is important for solar energy management in urban areas with complicated atmospheric conditions. The urban solar irradiance (USI) can be predictively researched with a variety of models; however, basing this entirely on one model often leads to other important conditions being omitted. A hybrid framework is suggested in this study, integrating the Perez–Ineichen PI model with a Deep Neural Network (DNN) model for predicting USI in Wuhan, China. The PI model predicts clear-sky irradiance labels based on atmospheric parameters normalized against the National Solar Radiation Database for greater accuracy. The model is trained on the Clear Sky Index with real-time atmospheric parameters gained from ground station measurements and satellite images. Following correlation analysis using bands from Sentinel-2 to find suitable bands for the model, the algorithm was prepared for atmospheric parameters, including cloud cover, aerosol concentration, and surface reflectance, all of which impact solar radiation. The architecture incorporates attention methods for important atmospheric parameters and skip connections for greater training stability. Results from the Deep Neural Network-Selected bands (DNN-S) and Deep Neural Network-All bands (DNN-A) models gave different performances, with the DNN-S model yielding better accuracy with a RMSE of 69.49 W/m2 clear-sky, 87.60 W/m2 cloudy-sky, and 72.57 W/m2 all-sky. The results were validated using hyperspectral imagery, along with cloud mask, solar area, and surface albedo-derived products, confirming that the USI estimates are supported by the high precision and consistency of Sentinel-2-derived irradiance estimates.

Towards the intrahour forecasting of direct normal irradiance using sky-imaging data

Photovoltaic (PV) power is generated by two common types of solar components that are primarily affected by fluctuations and development in cloud structures as a result of uncertain and chaotic processes. Local PV forecasting is unavoidable in supply and load planning necessary in integration of smart systems into electrical grids. Intra- or day-ahead modelling of weather patterns based on Artificial Intelligence (AI) allows one to refine available 24 h. cloudiness forecast or predict PV production at a particular plant location during the day. AI usually gets an adequate prediction quality in shorter-level horizons, using the historical meteo- and PV record series as compared to Numerical Weather Prediction (NWP) systems. NWP models are produced every 6 h to simulate grid motion of local cloudiness, which is additionally delayed and usually scaled in a rough less operational applicability. Differential Neural Network (DNN) is based on a newly developed neurocomputing strategy that allows the representation of complex weather patterns analogous to NWP. DNN parses the n-variable linear Partial Differential Equation (PDE), which describes the ground-level patterns, into sub-PDE modules of a determined order at each node. Their derivatives are substituted by the Laplace transforms and solved using adapted inverse operations of Operation Calculus (OC). DNN fuses OC mathematics with neural computing in evolution 2-input node structures to form sum modules of selected PDEs added step-by-step to the expanded composite model. The AI multi- 1…9-h and one-stage 24-h models were evolved using spatio-temporal data in the preidentified daily learning sequences according to the applied input–output data delay to predict the Clear Sky Index (CSI). The prediction results of both statistical schemes were evaluated to assess the performance of the AI models. Intraday models obtain slightly better prediction accuracy in average errors compared to those applied in the second-day-ahead evening approach. However, the first-morning strategy application includes a much more complicated and time-consuming procedure, so in conclusion, the second one-mode series processing could be recommended. Model statistics are generally more successful than NWP data processing due to limitations, delays, and unavailability of free radiation or cloudiness forecasts in the systems. An all-day model enables the prediction of complete PVP periods in sequenced computing with acceptable operational quality in the late afternoon, which is inevitable in day-ahead management and load scheduling in smart grids based on PV energy.

Spot panchromatic satellite image had been employed to study and know the difference Between ground and satellite data( DN ,its values varies from 0-255) where it is necessary to convert these DN values to absolute radiance values through special equations ,later it converted to spectral reflectance values .In this study a monitoring of the environmental effect resulted from throwing the sewage drainages pollutants (industrial and home) into the Tigris river water in Mosul, was achieved, which have an effect mostly on physical characters specially color and turbidity which lead to the variation in Spectral Reflectance of the river water ,and it could be detected by using many remote sensing techniques. The contaminated areas within the water of the river which represents the difference in the reflectance values were isolated and signed, as well as the field estimations, which had been done by using spectrometer device, which gave an acceptable agreement with satellite data considering time difference between these estimations. satellite imagery analysis program ERDAS version 8.4 was used to determine the values of Spectral Reflectance in the satellite image. A geographic information systems through the ARC INFO has been used to draw photo map of the study area determined it specific sites of measuring the Reflectance, which represent areas that are near the sources of pollution and the other various regions along the river.

The geomagnetic components from eight low-latitude ground stations and the ATS 1 satellite are studied for substorms occurring on December 24 and 25, 1967, and on August 15, 1968. A simple, four-parameter, field-aligned line current model is employed, and the linear inverse problem is constructed to determine the set of model parameters that best fit the data. A straightforward review of linear inversion theory is included in the theory section. The four model parameters are the magnitude of the current, the L shell on which the current flows, and the azimuthal positions of the eastern and western portions of the circuit. Both growth phases and expansion phases are treated. The best fits to the ground and satellite data are shown to be independent of the L parameter for values of L ≳ 3.5 when only ground data are treated and for values of L ≳ 8.5 when satellite data are included. The four-parameter model, although it produces good qualitative fits to expansion phase data, is unable to fit the data within 2 standard deviations of the error in the observations. The model is able to fit the growth phase of one substorm within 1 standard deviation. In one substorm with an apparent growth phase a better fit of the data during the expansion phase was obtained by combining the line current model of the growth phase with an additional four-parameter line current model.

O aproveitamento da energia solar vem se tornando viável economicamente e, atualmente, é a que vem apresentando maior crescimento mundial. O conhecimento da disponibilidade de energia solar na superfície é um fator essencial para incentivar novos investimentos em sistemas de geração de energiasolar. Em locais onde não existem sistemas de medida, estimativas da incidência de radiação solar podem ser realizadas por interpolação/extrapolação de medidas realizadas em estações vizinhas ou através de modelos numéricos de transferência radiativa. Ambas são formas aproximadas de estimativa, e seus empregos dependem da quantificação de suas incertezas. Neste trabalho, dados de radiação solar coletados em 18 plataformas de coleta de dados (PCDs), situadas no sudeste do Brasil, foram empregados na geração de extrapolações e interpolações considerando diferentes distâncias das PCDs. Os desvios apresentados por estas estimativas crescem com a distância, e são comparáveis com os desvios do modelo satelital de transferência radiativa BRASIL-SR apenas para distâncias inferiores a 60 km. Os métodos de interpolação/extrapolação apresentaram desvios dentro do intervalo limitado pelos desvios RMSE (root mean square errors) máximo e mínimo observados com o uso do modelo. Dessa forma, pode-se concluir que o uso de modelos físicos de transferência radiativa alimentados por dados de satélites representam a melhor solução para a estimativa da disponibilidade de energia solar em locais distantes das estações radiométricas.