A closure study of extinction apportionment by multiple regression

Autor:	Peter H. McMurry, Barbara J. Turpin, Edward S. Macias, L.A. de P. Vasconcelos, Warren H. White
Rok vydání:	2001
Předmět:	Constraint (information theory) Estimation Atmospheric Science Offset (computer science) Meteorology Extinction (optical mineralogy) Linear regression Statistics Sample (statistics) Light scattering Regression General Environmental Science Mathematics
Zdroj:	ResearcherID
ISSN:	1352-2310
DOI:	10.1016/s1352-2310(00)00273-9
Popis:	Multiple regression has been widely used to apportion particle light scattering among distinct chemical species. The resulting scattering budgets are shown here to be unbiased estimates under certain theoretical conditions. The theory allows species’ particle size distributions and water uptakes to vary from sample to sample, as they are known to do in reality. The sole constraint is that variations in each species’ characteristics be statistically independent of all species’ concentrations. Individual violations of this condition cause identifiable biases, and multiple violations can offset each other to yield regression estimates that are accurate by accident. Detailed and summary accountings of statistical errors are illustrated using a mechanistic model derived from measurements in southern California.
Databáze:	OpenAIRE
Externí odkaz:	https://explore.openaire.eu/search/publication?articleId=doi_dedup___::e6a31e157ac6d6d5b2f30dc64be40c46 https://doi.org/10.1016/s1352-2310(00)00273-9 Zobrazit plný text záznamu Full Text from ScienceDirect