Attention-Based Models for Multivariate Time Series Forecasting: Multi-step Solar Irradiation Prediction

Bangladesh's subtropical climate with an abundance of sunlight throughout the greater portion of the year results in increased effectiveness of solar panels. Solar irradiance forecasting is an essential aspect of grid-connected photovoltaic systems to efficiently manage solar power's variation and uncertainty and to assist in balancing power supply and demand. This is why it is essential to forecast solar irradiation accurately. Many meteorological factors influence solar irradiation, which has a high degree of fluctuation and uncertainty. Predicting solar irradiance multiple steps ahead makes it difficult for forecasting models to capture long-term sequential relationships. Attention-based models are widely used in the field of Natural Language Processing for their ability to learn long-term dependencies within sequential data. In this paper, our aim is to present an attention-based model framework for multivariate time series forecasting. Using data from two different locations in Bangladesh with a resolution of 30 min, the Attention-based encoder-decoder, Transformer, and Temporal Fusion Transformer (TFT) models are trained and tested to predict over 24 steps ahead and compared with other forecasting models. According to our findings, adding the attention mechanism significantly increased prediction accuracy and TFT has shown to be more precise than the rest of the algorithms in terms of accuracy and robustness. The obtained mean square error (MSE), the mean absolute error (MAE), and the coefficient of determination (R2) values for TFT are 0.151, 0.212, and 0.815, respectively. In comparison to the benchmark and sequential models (including the Naive, MLP, and Encoder-Decoder models), TFT has a reduction in the MSE and MAE of 8.4-47.9% and 6.1-22.3%, respectively, while R2 is raised by 2.13-26.16%. The ability to incorporate long-distance dependency increases the predictive power of attention models.

Remote sensing algorithm for retrieving global-scale sea surface solar irradiance

This paper presents a new remote sensing (RS) algorithm for retrieving instantaneous sea surface solar irradiance (SR) by using the XGBoost (XGB) package in RStudio and available remote sensing data along with ground-truth solar irradiance data. By means of XGB, the new RS algorithm, called LSU model, was structurally built with nine key RS parameters, including photosynthetically available radiation (PAR); instantaneous PAR (iPAR); water leaving reflectance Rrs at wavelengths 412, 443, 469, and 488 nm; angstrom; aerosol optical thickness (aot_869); and latitude that represent major sources and sinks of solar irradiance, as model input variables. Among the nine parameters, the most important four parameters are PAR, iPAR, latitude, and aot_869. It was found that the sea surface SR is highly affected by conditions in both the atmosphere and the seawater. The aot_869 is by far the most important factor describing the effects of the atmospheric absorption and scattering of SR before reaching the sea surface. The most important factors describing the effects of seawater characteristics on solar irradiance are PAR, iPAR, and latitude. Comparisons with existing SR models indicate that LSU model is scientifically sound due to the use of major source and sink factors of SR as model input variables. LSU model is also technically accurate due to its fine resolution (1×1 km) and overall best performance in predicting instantaneous SR. More importantly, LSU model is globally applicable as it can be utilized to obtain global-scale SR data for any day, any time, and anywhere in the world.

Application of photovoltaic panels in electric vehicles to enhance the range

There is a significant increase in the number of alternative energy sources and electric vehicles. Therefore, there is a growing need for new technical solutions to increase the distance that an electric vehicle can travel on a single charge. The aim of this study is to assess the possibility of mileage increasing of an electric vehicle by means of commercially available solar energy technologies that require minimal investment. The considered electric car can be recharged from solar panels mounted on its roof during parking stages. Photovoltaic modules can contribute to the vehicle's propulsion or energize its accessories, such as ventilation, air conditioner, heated passenger seats, interior lighting. The results demonstrate feasibility of the proposed solutions for both cases with and without sun-tracking adjustments of solar panels. The calculations show that the vehicle-integrated photovoltaic panels can provide energy for up to 6.32% of the range on a full charge of the battery during the sunniest summer months and up to 1.16% of the range during the least sunny winter months, for the given conditions.

Electrical production data of a domestic grid-connected rooftop PV plant in normal and shading faults conditions associated with solar and meteorological data in a tropical climate

The proposed measured data combines PV plant electrical data with associated solar and meteorological data during normal and faulty conditions. Data are collected regarding a domestic rooftop PV plant of 4 kW, located in the La Réunion Island, in the South-West of Indian Ocean. The present dataset includes healthy behavior and different types of shading faults, identified and labelled by means of a numeric variable. The electrical data (voltage, current and power at AC and DC side as well as produced energy and grid frequency) are collected thanks to PV inverters. Global and diffuse irradiance, PV temperature and ambient temperature are acquired thanks to additional sensors. Electrical and meteorological data sampling frequencies are set to 0.2 Hz and 1 Hz respectively. At present, 12 months of data are available and the database is still being updated. The data streams from each connected device require proper techniques to ensure their persistence. To be able to provide both efficient ingestion and retrieval of these time series collections, the NoSQL database management system InfluxDB has been implemented. The whole dataset is available on Zenodo repository, and can be used, for instance, for PV modeling, PV plant behavior analysis, PV production forecasting and PV Fault Detection and Diagnosis (FDD) tool development.

A novel recurrent neural network approach in forecasting short term solar irradiance

Forecasting solar irradiance is of utmost importance in supplying renewable energy efficiently and timely. This paper aims to experiment five variants of recurrent neural networks (RNN), and develop effective and reliable 5-minute short term solar irradiance prediction models. The 5 RNN classes are long-short term memory (LSTM), gated recurrent unit (GRU), Simple RNN, bidirectional LSTM (Bi-LSTM), and bidirectional GRU (Bi-GRU); the first 3 classes are unidirectional and the last two are bidirectional RNN models. The 26 months data under consideration, exhibits extremely volatile weather conditions in Jinju city, South Korea. Therefore, after different experimental processes, 5 hyper-parameters were selected for each model cautiously. In each model, different levels of depth and width were tested; moreover, a 9-fold cross validation was applied to distinguish them against high variability in the seasonal time-series dataset. Generally the deeper architectures of the aforementioned models had significant outcomes; meanwhile, the Bi-LSTM and Bi-GRU provided more accurate predictions as compared to the unidirectional ones. The Bi-GRU model provided the lowest RMSE and highest R² values of 46.1 and 0.958; additionally, it required 5.25*10^-5 seconds per trainable parameter per epoch, the lowest incurred computational cost among the mentioned models. All 5 models performed differently over the four seasons in the 9-fold cross validation test. On average, the bidirectional RNNs and the simple RNN model showed high robustness with less data and high temporal data variability; although, the stronger architectures of the bidirectional models, deems their results more reliable.

Science Highlights and Final Updates from 17 Years of Total Solar Irradiance Measurements from the SOlar Radiation and Climate Experiment/Total Irradiance Monitor (SORCE/TIM)

The final version (V.19) of the total solar irradiance data from the SOlar Radiation and Climate Experiment (SORCE) Total Irradiance Monitor has been released. This version includes all calibrations updated to the end of the mission and provides irradiance data from 25 February 2003 through 25 February 2020. These final calibrations are presented along with the resulting final data products. An overview of the on-orbit operations timeline is provided as well as the associated changes in the time-dependent uncertainties. Scientific highlights from the instrument are also presented. These include the establishment of a new, lower TSI value; accuracy improvements to other TSI instruments via a new calibration facility; the lowest on-orbit noise (for high sensitivity to solar variability) of any TSI instrument; the best inherent stability of any on-orbit TSI instrument; a lengthy (17-year) measurement record benefitting from these stable, low-noise measurements; the first reported detection of a solar flare in TSI; and observations of two Venus transits and four Mercury transits.

Type II photosensitized oxidation in senescent microalgal cells at different latitudes: Does low under-ice irradiance in polar regions enhance efficiency?

Comparison of Type II photosensitized oxidation of lipids (the photodynamic effect) and photodegradation of chlorophyll (sensitizer photobleaching) in samples of particulate matter collected previously from locations representing a diverse range of latitudes reveals an enhancement of the photooxidation of lipids at the expense of chlorophyll photodegradation in the polar regions. The efficiency of the photodynamic effect appears to be particularly high in sinking particles collected under sea ice and is attributed to the rapid settling of highly aggregated sympagic algae to depths of low light transmission favouring the photodynamic effect at the expense of photobleaching of the sensitizer. Paradoxically, the low efficiency of Type II photosensitized oxidation of lipids observed in temperate and equatorial regions is associated with high solar irradiances in these regions. Type II photosensitized oxidation of lipids in senescent phytoplankton seems thus to be strongly dependent of the intensity of solar irradiance.

Kinetic modeling of the synergistic thermal and spectral actions on the inactivation of viruses in water by sunlight

Sunlight can be an effective tool for inactivating pathogens in water disinfection processes. In clear water, photoinactivation of viruses is driven by the absorption of UVB radiation and it is more efficient at shorter wavelengths. Moreover, the temperature can significantly improve the efficiency of the process. To date, no kinetic model has been reported that describes the simultaneous thermal and spectral effects that occur during the solar inactivation of viruses. This work presents a novel comprehensive kinetic model for the solar inactivation of MS2 coliphage as a function of the water temperature, irradiance, and spectral distribution of the incident radiation. The model is based on a combination of the modified Arrhenius equation, a wavelength-dependent first-order inactivation model with the quantum yield, and thermal parameters estimated from laboratory data. Model predictions have a 9% error with respect to experiments in the temperature range from 30 to 50 °C and UV irradiance range from 15 to 50 W/m². Moreover, the model was validated in three scenarios using different plastic materials that modify the spectral range of the radiation reaching the water, confirming an accurate prediction of inactivation rates for real solar disinfection systems worldwide using containers made of any material.

Effects of shade location and protection from direct solar radiation on the behavior of Holstein cows

Two trials (E1 and E2) were performed to assess the behavior of eight Holstein dairy cows with 367 ± 58 kg of body weight and 10.52 ± 0.08 kg of milk yield. A 4 × 4 Latin square design (four periods of lactation and four levels of solar blockage) with four paddocks was used. Each paddock contained a wood shading structure covered with a cloth that blocked 30% (T1), 50% (T2), 70% (T3), or 100% (T4) of direct solar radiation. In the first trial (E1) each shade structure was located approximately 40 m from the feeder and water troughs; in the second trial (E2), the distance was reduced to 5 m. Air temperature (T_A, °C), relative humidity (R_H, %), wind speed (U, ms^-1), black globe temperature (T_G, K), mean radiant temperature (T_MR, K), radiant heat load (R_HL, W m^-2), and local shortwave radiation (R_S, W m^-2) were recorded at 15-min intervals from 08:00 to 17:00 h. Four behavioral activities were recorded: grazing, eating at the feed trough, ruminating, and idling. For each of these activities, animal posture (lying or upright) and location (under shade or exposed to sunlight) were recorded. The meteorological conditions showed similar variations from 8:00 to 17:00 h between the two trials. However, the air temperatures in E1 were lower (± 2 °C) than those in E2. In a PCA analysis, the first and the second principal components explained 56.87% and 21.85%, respectively, of the total variation in the behavioral variables. Under the E1 conditions, the animals did not seek shade, whereas in E2, the dairy cows spent 35 ± 5% of their time lying and idling in the shade. At a solar radiation blockage of 100%, cows were in the shade more than 60% of the time due to the intensity of solar radiation, which was 722.19 ± 14.59 W m^-2 at 11:45. In a PCA analysis, the first and the second principal components explained 65.18 and 22.3%, respectively, and 87.48% together, of the total variation in the original variables. Consequently, it was possible to develop a shade index (IST) based on the first two components. In E1, animals spent very little time in the shade, spending only 0.15% of total time under the shade, irrespective of blockage. However, E2 cows used shade, reaching almost 80% of time under the shade, at midday, when the blockage was 100%.

Compilation and spatio-temporal analysis of publicly available total solar and UV irradiance data in the contiguous United States

Skin cancer is the most common type of cancer in the United States, the majority of which is caused by overexposure to ultraviolet (UV) irradiance, which is one component of sunlight. National Environmental Public Health Tracking Program at CDC has collaborated with partners to develop and disseminate county-level daily UV irradiance (2005-2015) and total solar irradiance (1991-2012) data for the contiguous United States. UV irradiance dataset was derived from the Ozone Monitoring Instrument (OMI), and solar irradiance was extracted from National Solar Radiation Data Base (NSRDB) and SolarAnywhere data. Firstly, we produced daily population-weighted UV and solar irradiance datasets at the county level. Then the spatial distributions and long-term trends of UV irradiance, solar irradiance and the ratio of UV irradiance to solar irradiance were analyzed. The national average values across all years are 4300 Wh/m2, 2700 J/m2 and 130 mW/m2 for global horizontal irradiance (GHI), erythemally weighted daily dose of UV irradiance (EDD) and erythemally weighted UV irradiance at local solar noon time (EDR), respectively. Solar, UV irradiances and the ratio of UV to solar irradiance all increased toward the South and in some areas with high altitude, suggesting that using solar irradiance as indicator of UV irradiance in studies covering large geographic regions may bias the true pattern of UV exposure. National annual average daily solar and UV irradiances increased significantly over the years by about 0.3% and 0.5% per year, respectively. Both datasets are available to the public through CDC's Tracking network. The UV irradiance dataset is currently the only publicly-available, spatially-resolved, and long-term UV irradiance dataset covering the contiguous United States. These datasets help us understand the spatial distributions and temporal trends of solar and UV irradiances, and allow for improved characterization of UV and sunlight exposure in future studies.

Optimal parameters estimation and modelling of photovoltaic modules using analytical method

This paper introduces a proposed approach to estimate the optimal parameters of the photovoltaic (PV) modules using in-field outdoor measurements and manufacturers' datasheet as well as employing the nonlinear least-squares fitting algorithm. The main goal is to determine the optimal parameter values of the implemented model which are: series resistance, reverse saturation current, photocurrent, ideality factor and shunt resistance in case of the five parameters model. A Microsoft Excel spreadsheet is developed in order to perform modeling and analysis of the parameters analytical initial values using manufacturer datasheet specifications regarding to the changing in solar irradiance and ambient temperature. Then, the sum of the squared residuals between in-field measured and simulated data are calculated and minimized using Excel solver in order to obtain the optimal values of the parameters simultaneously, to describe the best fit for the outdoor measured data. The proposed approach is used to find the optimal parameters of the PV module TRINA TSM-295 using an array tester. The convergence confidences of the estimated parameters are presented and assessed in an easy way. This approach allows all parameters to be optimized, simultaneously. The results are verified and compared with other research studies for different PV cell technologies. The obtained results are useful for the tested PV module manufacturer and assess the performance of the products in different weather conditions.

Changes in grapevine leaf phenolic profiles during the day are temperature rather than irradiance driven

Photosynthesis parameters, adaxial flavonoid index, phenolic profiles and antioxidant capacities of south-facing sun exposed grapevine leaves (Vitis vinifera, Pinot Noir cultivar) were measured hourly between 7 a.m. and 7 p.m. on a clear summer day. Changes in these parameters were statistically compared to changes in environmental conditions, including solar irradiance (photosynthetically active and UV radiations), leaf and air temperature, and relative air humidity. Epidermal UV absorbance, characterised by the flavonoid index, and total extractable phenolic contents were correlated to distinct environmental parameters. The former was positively correlated to irradiance and leaf temperature, while the latter was positively correlated to air temperature. HPLC phenolic profiling identified a positive correlation between air temperature and amounts of the dominant flavonol component, quercetin-3-O-glucuronide. The only phenolic component statistically connected to the flavonoid index was quercetin-3-O-glucoside. This correlation was positive and both parameters decreased during the day, although changes in the amount of this flavonol component showed no correlation to environmental factors. Total antioxidant capacities of leaf extracts were positively correlated to solar UV, and leaf and air temperature, but not to photosynthetically active radiation. Positive correlations of quercetin-3-O-glucoside contents with the flavonoid index, with photosynthesis and with sub-stomatal CO₂ concentration suggest a special protective role of this flavonol. A short-term negative effect of solar UV-A and UV-B on photosynthetic CO₂ uptake was also identified, which was unrelated to changes in stomatal conductance. A hypothesis is presented assuming UV- and photorespiration-derived hydrogen peroxide as the driver of daily changes in leaf antioxidant capacities.

High impact of seasonal temperature changes on acclimation of photoprotection and radiation-induced damage in field grown Arabidopsis thaliana

At temperate latitudes environmental factors such as irradiance, including ultraviolet-B radiation (UV-B, 280-315 nm), temperature and day length vary widely over the course of a year in a concerted way. In the present study physiological acclimation of photoprotection, growth and development of the model organism Arabidopsis thaliana were correlated to these strongly but gradually changing conditions in a one year field study. Plants were sown in the field avoiding any manipulation (and abrupt change) during their life. Developmental rate was strongly dependent on prevailing temperature. Moderate signs of light stress in form of photoinhibition at photosystem II were significantly related to solar irradiances while amount of DNA damage was low and not correlated to UV-B irradiance. Although all the markers were hypothesized to primarily react to radiation, multiple regression analysis showed at least a similarly strong influence of temperature as that of light. Especially for the classical UV screening compounds a positive correlation to UV-B radiation during the course of the year was absent, whereas there was a significant negative correlation between temperature and quercetin content. The sum of violaxanthin cycle pigments was correlated to both, irradiance and temperature, but with opposite sign. Epidermal UV-B transmittance was also much better related to air temperature than to UV-B irradiance. The data show that under natural conditions temperature has at least a similar importance for photoprotective acclimation and partially also for photosensitivity as solar irradiance.

Hourly photosynthetically active radiation estimation in Midwestern United States from artificial neural networks and conventional regressions models

The relationship between hourly photosynthetically active radiation (PAR) and the global solar radiation (R s ) was analyzed from data gathered over 3 years at Bondville, IL, and Sioux Falls, SD, Midwestern USA. These data were used to determine temporal variability of the PAR fraction and its dependence on different sky conditions, which were defined by the clearness index. Meanwhile, models based on artificial neural networks (ANNs) were established for predicting hourly PAR. The performance of the proposed models was compared with four existing conventional regression models in terms of the normalized root mean square error (NRMSE), the coefficient of determination (r (2)), the mean percentage error (MPE), and the relative standard error (RSE). From the overall analysis, it shows that the ANN model can predict PAR accurately, especially for overcast sky and clear sky conditions. Meanwhile, the parameters related to water vapor do not improve the prediction result significantly.

Effects of water-filtered infrared-A and of heat on cell death, inflammation, antioxidative potential and of free radical formation in viable skin--first results

The effects of water-filtered infrared-A (wIRA) and of convective heat on viability, inflammation, inducible free radicals and antioxidative power were investigated in natural and viable skin using the ex vivo Bovine Udder System (BUS) model. Therefore, skin samples from differently treated parts of the udder of a healthy cow were analyzed using the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) test, by prostaglandin E2 (PGE2) measurement and by electron spin resonance (ESR) spectroscopy. Neither cell viability, the inflammation status, the radical status or the antioxidative defence systems of the skin were significantly affected by wIRA applied within 30 min by using an irradiance of 1900 W m(-2) which is of relevance for clinical use, but which exceeded the maximum solar IR-A irradiance at the Earth's surface more than 5 times and which resulted in a skin surface temperature of about 45 °C without cooling and of about 37 °C with convective cooling by air ventilation. No significant effects on viability and on inflammation were detected when convective heat was applied alone under equivalent conditions in terms of the resulting skin surface temperatures and exposure time. As compared with untreated skin, free radical formation was almost doubled, whereas the antioxidative power was reduced to about 50% after convective heating to about 45 °C.

The Space Weather and Ultraviolet Solar Variability (SWUSV) Microsatellite Mission

We present the ambitions of the SWUSV (Space Weather and Ultraviolet Solar Variability) Microsatellite Mission that encompasses three major scientific objectives: (1) Space Weather including the prediction and detection of major eruptions and coronal mass ejections (Lyman-Alpha and Herzberg continuum imaging); (2) solar forcing on the climate through radiation and their interactions with the local stratosphere (UV spectral irradiance from 180 to 400 nm by bands of 20 nm, plus Lyman-Alpha and the CN bandhead); (3) simultaneous radiative budget of the Earth, UV to IR, with an accuracy better than 1% in differential. The paper briefly outlines the mission and describes the five proposed instruments of the model payload: SUAVE (Solar Ultraviolet Advanced Variability Experiment), an optimized telescope for FUV (Lyman-Alpha) and MUV (200–220 nm Herzberg continuum) imaging (sources of variability); UPR (Ultraviolet Passband Radiometers), with 64 UV filter radiometers; a vector magnetometer; thermal plasma measurements and Langmuir probes; and a total and spectral solar irradiance and Earth radiative budget ensemble (SERB, Solar irradiance & Earth Radiative Budget). SWUSV is proposed as a small mission to CNES and to ESA for a possible flight as early as 2017–2018.

Computing diffuse fraction of global horizontal solar radiation: A model comparison

For simulation-based prediction of buildings' energy use or expected gains from building-integrated solar energy systems, information on both direct and diffuse component of solar radiation is necessary. Available measured data are, however, typically restricted to global horizontal irradiance. There have been thus many efforts in the past to develop algorithms for the derivation of the diffuse fraction of solar irradiance. In this context, the present paper compares eight models for estimating diffuse fraction of irradiance based on a database of measured irradiance from Vienna, Austria. These models generally involve mathematical formulations with multiple coefficients whose values are typically valid for a specific location. Subsequent to a first comparison of these eight models, three better performing models were selected for a more detailed analysis. Thereby, the coefficients of the models were modified to account for Vienna data. The results suggest that some models can provide relatively reliable estimations of the diffuse fractions of the global irradiance. The calibration procedure could only slightly improve the models' performance.