Search

The rapid growth of solar power generation and the variable nature of the solar resource pose challenges for our electricity grids. Forecasting future changes in the irradiance might help to cost-efficiently manage this variability both for photovoltaic and concentration solar plants as well as grids with high solar penetrations. So far, for shortest-term forecasts with lead times of a few minutes, all-sky imager based nowcasting systems are used. However, due to the complexity of dynamically changing 3d cloud shapes as well as certain geometrical effects such as self-occlusion or near-horizon saturation, all-sky imager based nowcasting systems exhibit inherent weaknesses. Here, we present a novel system to generate shortest-term solar forecasts, which is located at Plataforma Solar de Almería in southern Spain. This approach is based on downward-facing cameras (shadow cameras), taking images of the ground. From these images, spatially resolved irradiance maps are derived. By tracking cloud shadows, future irradiances are predicted. A demonstration system is achieved, which provides shortest-term forecasts for the next 2 min. To the best of our knowledge, this is the first time such a system is developed. We benchmark several possible algorithmic approaches on 16 days and compare the deviations to a state-of-the-art all-sky imager based nowcasting system on 22 days. The root-mean-squared deviation (RMSD) of this shadow camera based nowcasting system for direct normal irradiance (DNI) and 1‑min temporal averages is 15.6 % for lead times of 2 min (MAD, DNI: 9.6 %). In comparison to an all-sky imager system, this is an improvement as the all-sky imager system only reaches 22.0 % RMSD and 14.8 % MAD (both DNI). This demonstrates the feasibility and attractiveness in terms of accuracy of the proposed concept.

International audience; Downward-facing shadow cameras might play a major role in future energy meteorology. Shadow cameras directly image shadows on the ground from an elevated position. They are used to validate other systems (e.g. all-sky imager based nowcasting systems, cloud speed sensors or satellite forecasts) and can potentially provide short term forecasts for solar power plants. Such forecasts are needed for electricity grids with high penetrations of renewable energy and can help to optimize plant operations. In this publication, two key applications of shadow cameras are briefly presented.

The incoming downward shortwave solar irradiance is harvested to an increasing extent by solar power plants. However, the variable nature of this energy source poses an operational challenge for solar power plants and electrical grids. Intra hour solar irradiance nowcasts with a high temporal and spatial resolution could be used to tackle this challenge. All sky imager (ASI) based nowcasting systems fulfill the requirements in terms of temporal and spatial resolution. However, ASI nowcasts can only be used if the required accuracies for applications in solar power plants and electrical grids are fulfilled. Scalar error metrics, such as mean absolute deviation, root mean square deviation, and skill score are commonly used to estimate the accuracy of nowcasting systems. However, these overall error metrics represented by a single number per metric are neither suitable to determine the real time accuracy of a nowcasting system in the actual weather situation, nor suitable to describe any spatially resolved nowcast accuracy. The performance of ASI-based nowcasting systems is strongly related to the prevailing weather conditions. Depending on weather conditions, large discrepancies between the overall and current system uncertainties are conceivable. Furthermore, the nowcast accuracy varies strongly within the irradiance map as higher errors may occur at transient zones close to cloud shadow edges. In this paper, we present a novel approach for the spatially resolved real-time uncertainty specification of ASI-based nowcasting systems. The current irradiance conditions are classified in one of eight distinct temporal direct normal irradiance (DNI) variability classes. For each class and lead-time, an upper and lower uncertainty value is derived from historical data, which describes a coverage probability of 68.3%. This database of uncertainty values is based on deviations of the irradiance maps, compared to three reference pyrheliometers in Tabernas, Andalucia over two years (2016 and 2017). Increased uncertainties due to transient effects are considered by detecting transient zones close to cloud shadow edges within the DNI map. The width of the transient zones is estimated by the current average cloud height, cloud speed, lead-time, and Sun position. The final spatially resolved uncertainties are validated with three reference pyrheliometers, using a data set consisting of the entire year 2018. Furthermore, we developed a procedure based on the DNI temporal variability classes to estimate the expected average uncertainties of the nowcasting system at any geographical location. The novel method can also be applied for global tilted or horizontal irradiance and is assumed to improve the applicability of the ASI nowcasts.

The magnitude of the direct normal irradiance (DNI) depends on seasonal differences due to the sun-earth geometry, and atmospheric extinction processes caused by aerosols and clouds. The most relevant source for intra-hour variability in the DNI are clouds. Suitable sites for concentrating solar power (CSP) plants have an annual average DNI above approx. 2000 kWh/m². Not only the average DNI but also the temporal and spatial DNI variability on the field has a considerable impact on the dispatched electricity [1]. A classification method based on eight distinct variability classes for 1 minute resolved DNI on an hourly basis is introduced by [2]. In this work, the approach presented by [2] is adapted for an intra-hour application, which is more suitable for CSP applications. Here, we discuss the impact of the considered time window on the classification. Furthermore, the impact of the spatial solar field DNI on the variability classification was studied, using spatially resolved DNI maps of a camera based nowcasting system [3].Moreover, we present the results of a variability classification study, including data of a complete year (2017) for a NETRA facility near New Delhi and the Plataforma Solar de Almería (PSA)

Kurzzeitprognosen der Solarstrahlung für die nächsten 15 Minuten bergen ein großes Potenzial zur Betriebsoptimierung von solarthermischen Kraftwerken, PV-Systemen als auch des Netzbetriebs. Kameragestützte Prognosesysteme erfuhren in den letzten Jahren einen signifikanten Entwicklungsschub, begünstigt durch die Verfügbarkeit von günstigen hochauflösenden Kameras, der Fortschritte bei den verfügbaren Speicher- und Rechenkapazitäten und dem stetigen Wachstum des solaren Anteils im weltweiten Energiemix. Das Potenzial für räumlich aufgelöste Kurzzeitprognosen der Globalstrahlung wurde u.a. in [1] untersucht. [2] und [3] beschreiben Methoden für Direktstrahlungsprognosen, welche bis zu vier unabhängige 2D Wolkenschichten berücksichtigen. In [4] werden zwei Kameras mit Fischaugenobjektiven für eine detaillierte 3D Wolken-Photogrammmetrie eingesetzt. Im Rahmen dieser Arbeit wurde ein Echtzeit- kurzzeitprognose-System für räumlich hochaufgelöste (5mx5m) DNI Karten basierend auf zwei bis vier Standard Überwachungskameras (Mobotix Q24 oder Q25) entwickelt [5]. Im vorgestellten System wird jede sichtbare Wolke als individuelles 3D Objekt in einem virtuellen Raum modelliert. Jedem Wolkenobjekt im Raum werden die Attribute Höhe, Position, Dicke und Transmissionsgrad zugewiesen. Ein individueller Bewegungsvektor wird für jedes Wolkenobjekt ermittelt. Mittels dieses Systems können Kurzzeitprognosen der Direktstrahlung (DNI) auch bei komplexen mehrschichtigen Wolkenzugsbedingungen erstellt werden.

International audience; Aerosols and clouds affect strongly the incoming solar radiation. Under cloud-free skies, aerosols become the dominant factor that affect the direct normal and the global horizontal irradiance (DNI and GHI respectively). For cloudy skies, the spatial and temporal variability of clouds is a challenging issue on solar irradiance resource and forecasting. In recent years, all-sky imagers are used for the detection of aerosol optical properties, cloud coverage, type and velocity in numerous applications including meteorological observations and solar energy control systems. In this presentation, a detailed overview of the methodologies developed at the Plataforma Solar de Almeria (PSA), in the framework of the EU project " Direct Normal Irradiance Nowcasting methods for optimized operation of concentrating solar technologies " (DNICast, http://www.dnicast-project.net/) is presented. In the framework of DNICast, we develop and provide aerosol optical depth (AOD) as intermediate product of cloud camera images as a basis for an AOD/DNI nowcasting scheme. Our main purpose is to analyze sky images being created and produced by a low-cost Mobotix Q24M off-the-shelf surveillance camera in order to estimate the AOD at different wavelengths from RGB intensities (Red, Green, Blue channels) of the photo. We use the RGB intensities/radiances from the zenith point and the size of the saturated area around the sun as input for the AOD determination. These data are taken into account in a multi-linear approach to estimate the AOD values at 440, 500 and 675nm and compared with the measurements of a CIMEL sun photometer at the same wavelengths. According to results, the mean/median difference and the standard deviation are less than 0.01 and 0.03 for all wavelengths. For the estimation of cloud coverage and type, we used several spectral and textural metrics mainly based on the method proposed by Kazantzidis et al. (2012). Notably, the spectral metrics are now only applied on cloudy pixels instead of on the full image. The classification algorithm provided two possible implementations of cloud type classification: i) Global cloud classification: classification of dominant cloud type and ii) Grid-based cloud detection: a classification of the cloud type is provided per grid element in the image. Based on visual human observation, the accuracy of the global cloud classifier ranges between 76 % (for cirrus cloud) and 84 % (for cumulus).