Cloud shade by dynamic logistic modeling
Marek Brabec,
Viorel Badescu and
Marius Paulescu
Journal of Applied Statistics, 2014, vol. 41, issue 6, 1174-1188
Abstract:
During the daytime, the sun is shining or not at ground level depending on clouds motion. Two binary variables may be used to quantify this process: the sunshine number (SSN) and the sunshine stability number (SSSN). The sequential features of SSN are treated in this paper by using Markovian Logistic Regression models, which avoid usual weaknesses of autoregressive integrated moving average modeling. The theory is illustrated with results obtained by using measurements performed in 2010 at Timisoara (southern Europe). Simple modeling taking into account internal dynamics with one lag history brings substantial reduction of misclassification compared with the persistence approach (to less than 57%). When longer history is considered, all the lags up to at least 8 are important. The seasonal changes are rather concentrated to low lags. Better performance is associated with a more stable radiative regime. More involved models add external influences (such as sun elevation angle or astronomic declination as well as taking into account morning and afternoon effects separately). Models including sun elevation effects are significantly better than those ignoring them. Clearly, during the winter months, the effect of declination is much more pronounced compared with the rest of the year. SSSN is important in long-term considerations and it also plays a role in retrospective assessment of the SSN. However, it is not easy to use SSSN for predicting future SSN. Using more complicated past beam clearness models does not necessarily provide better results than more simple models with SSN past.
Date: 2014
References: View references in EconPapers View complete reference list from CitEc
Citations: View citations in EconPapers (2)
Downloads: (external link)
http://hdl.handle.net/10.1080/02664763.2013.862221 (text/html)
Access to full text is restricted to subscribers.
Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.
Export reference: BibTeX
RIS (EndNote, ProCite, RefMan)
HTML/Text
Persistent link: https://EconPapers.repec.org/RePEc:taf:japsta:v:41:y:2014:i:6:p:1174-1188
Ordering information: This journal article can be ordered from
http://www.tandfonline.com/pricing/journal/CJAS20
DOI: 10.1080/02664763.2013.862221
Access Statistics for this article
Journal of Applied Statistics is currently edited by Robert Aykroyd
More articles in Journal of Applied Statistics from Taylor & Francis Journals
Bibliographic data for series maintained by Chris Longhurst ().