Open AccessOn the Bayesian estimation of cloud fraction from lidar transects
Author(s) -
Settle Jeffery J.,
van de Poll H. Martine
Publication year - 2007
Publication title -
journal of geophysical research: atmospheres
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.67
H-Index - 298
eISSN - 2156-2202
pISSN - 0148-0227
DOI - 10.1029/2006jd007251
Subject(s) - cloud fraction , quantile , exponential distribution , cloud computing , lidar , prior probability , bayesian probability , statistics , cloud cover , remote sensing , meteorology , geology , mathematics , computer science , physics , operating system
The problem of estimating cloud cover at different heights from spaceborne lidar transects has recently been tackled with a new statistical model. Exponential distributions are assumed for the cloud length and for the intercloud gaps and a Jeffreys prior distribution is assumed for the parameters of these distributions. These allow us to generate a posterior probability distribution for the actual cloud cover, given the observations (a set of cloud and intercloud lengths). The resulting model enables us to establish confidence intervals, also known as credible intervals, for estimates taken from this distribution. The estimates of cloud cover and credible intervals have so far been undertaken by numerical integration. In this paper we present approximations to the mean, variance, and quantile points of the posterior distribution. These were initially developed to handle the case where the number of clouds in a transect becomes very large, the distribution becomes very peaked, and the numerical integration is prone to error. However, by developing the asymptotic forms a little we find that the results apply with acceptable accuracy to problems with small numbers of whole clouds.