Pregled bibliografske jedinice broj: 756991
How to overcome persistent biases in the composite analyses: an example with cosmic rays and cloud cover
How to overcome persistent biases in the composite analyses: an example with cosmic rays and cloud cover // The XIIth Scientific Assembly - IAGA 2013
Mérida, Meksiko, 2013. (predavanje, međunarodna recenzija, sažetak, znanstveni)
CROSBI ID: 756991 Za ispravke kontaktirajte CROSBI podršku putem web obrasca
Naslov
How to overcome persistent biases in the composite analyses: an example with cosmic rays and cloud cover
Autori
Čalogović, Jaša ; Laken, Benjamin ; Dunne, Eimear M. ; Pierce, Jeffrey R.
Vrsta, podvrsta i kategorija rada
Sažeci sa skupova, sažetak, znanstveni
Skup
The XIIth Scientific Assembly - IAGA 2013
Mjesto i datum
Mérida, Meksiko, 26.08.2013. - 31.08.2013
Vrsta sudjelovanja
Predavanje
Vrsta recenzije
Međunarodna recenzija
Ključne riječi
superposed epoch analysis; Forbush decreases; cosmic rays; clouds; Monte Carlo; statistics
Sažetak
The main purpose of superposed epoch (composite) analysis is to resolve significant signal to noise problems, identifying responses to particular events that may otherwise be obscured by unrelated variations at similar time scales. This technique has been frequently employed to examine a hypothesized link between solar activity and the Earth’s climate following Forbush decrease events, during which strong reductions in the background cosmic ray flux occur. However, numerous studies using composites to test a link between cosmic rays and cloud properties arrived at a range of conflicting results. In this work, we argue that minor methodological differences in the manner in which composites have been both constructed and analyzed could provide an explanation for different results. Using extensive Monte Carlo simulation techniques and the two most widely used satellite cloud datasets (ISCCP, MODIS) we provide details on how a composite may be objectively constructed to maximize signal detection, to robustly identify statistical significance, and to quantify the lower-limit uncertainty related to hypothesis testing. Additionally, we also demonstrate how convincing false-positive results may be obtained from non- significant data, e.g. by using a small number of events in a composite, and by calculating anomalies against a base period, and/or by using traditional statistical tests (e.g. the Student’s T- test) which are based on wrong assumptions in the case of cloud data.
Izvorni jezik
Engleski
Znanstvena područja
Fizika, Geologija
