DLR-Logo -> http://www.dlr.de
DLR Portal Home | Imprint | Privacy Policy | Contact | Deutsch
Fontsize: [-] Text [+]

Benchmarking of cloud Segmentation algorithms for ground-based all-sky imagers

Hasenbalg, Marcel (2018) Benchmarking of cloud Segmentation algorithms for ground-based all-sky imagers. Bachelor's, Hochschule de Medien Stuttgart.

Full text not available from this repository.


Automatic cloud detection and analysis is a key component for many applications. With the imminent depletion of fossil fuels, alternative ways of generating electricity play a larger role in the electrical power supply. The optimal operation of large-scale Commercial solar power plants involves some challenges, including the adaption to electricity grid rules. The detection of clouds by camera based nowcasting systems is an important but difficult task, which contributes to mastering these challenges. These nowcasting systems are able to provide shorttime solar irradiance forecasts, optimizing the operation of solar power plants. The variability of the available sunlight is influenced by weather conditions, mainly clouds. Clouds shadowing solar power plants cause a sudden reduction in electricity generation. This variability affects the stability of electrical grids with high penetration of solar power plants. Nowcasting Systems are able to predict these sudden decreases in solar irradiance, giving the solar power plant time to prepare for these situtations. Large-scale data in low spatial and temporal resolutions about the cloud coverage is available from satellites. But to perform short term forecasting of solar irradiance for solar power plants, high temporal and spatial data is needed. All sky imagers, which are ground based cameras capturing the whole sky are able to provide this data in a sufficient resolution. Solar power plants respecting these short time solar irradiance forecasts are able to improve their usage of backup systems like batteries or diesel generators. In regions with high penetrations of solar power plants, especially island grids, network operators legally enforce limitations on the fluctuation of the power output of solar power plants, to ensure grid stability [Lave et al., (2013)] . By standing to these rules the solar power plant can avoid financial penalties from network operators and optimize the durability of the technical components. Therefore, using a cloud segmentation algorithm with a high accuracy under all weather conditions is a crucial step in the development of nowcasting systems. Because clouds can attenuate optical space-toground communication, detailled and highly resolved information on cloud coverage might be beneficial for such applications as well. Classic satellite communcation is using radio signals, which are able to pass through cloud covers. Laser communication on the other hand enables far better uplink and downlink speeds as well as enhanced security. But optical communication requires the two communicating parties to be in sight of each other. That is why clouds cause problems in laser communication approaches [Schimmel et al., (2018)] . By ground stations being able to automatically predict cloud free moments, optimal timeframes for data transfers could be identified. Additionally, precise automatic cloud detection and analysis is a crucial information for aviation. Especially pilots performing flights under the visual flight rules (VFR), where a clear sight on the runway is required, could benefit from high resolution realtime cloud coverage data from airports or other ground stations. Over 10% of fatal accidents in general aviation occur when pilots transition from a VFR flight into instrument meteorological conditions (IMC) [Coyne et al., (2008)] . These transitions often are involuntarily and due to bad estimations of visibility and cloud height. There are already commercial softwares available (e.g. EasyVFR, PocketFMS) for purchase, which provide benefitial data to VFR pilots. Including these realtime cloud data could contribute to the overall safety of future VFR flights. Finally, automatically retrieved highly resolved spatial and temporal cloud data is an interesting and valuable input for further research. This could include the validation of cloudiness forecasts based on satellite images, numerical weather prediction or other types of weather forecasting. For these different types of applications, an all-sky-imager in form of a digital camera is an adequate measuring instrument. These all-sky-imagers can be off-the-shelf surveillance cameras.

Item URL in elib:https://elib.dlr.de/123770/
Document Type:Thesis (Bachelor's)
Title:Benchmarking of cloud Segmentation algorithms for ground-based all-sky imagers
AuthorsInstitution or Email of AuthorsAuthors ORCID iD
Date:30 October 2018
Refereed publication:No
Open Access:No
Gold Open Access:No
In ISI Web of Science:No
Number of Pages:56
Keywords:all-sky imager, forecasts
Institution:Hochschule de Medien Stuttgart
HGF - Research field:Energy
HGF - Program:Renewable Energies
HGF - Program Themes:Concentrating Solar Thermal Technology
DLR - Research area:Energy
DLR - Program:E SW - Solar and Wind Energy
DLR - Research theme (Project):E - Impact of Desert Environment
Location: Köln-Porz
Institutes and Institutions:Institute of Solar Research > Qualification
Deposited By: Kruschinski, Anja
Deposited On:21 Dec 2018 13:42
Last Modified:29 Jan 2019 10:03

Repository Staff Only: item control page

Help & Contact
electronic library is running on EPrints 3.3.12
Copyright © 2008-2017 German Aerospace Center (DLR). All rights reserved.