Fig. 3: A framework of uncertainty quantification.
From: Global quantification of the dispersion effect with POLDER satellite data
a The flowchart of uncertainty quantification, where \({b}_{{s\_F}}\) and \({{{{\mathrm{SE}}}}}_{{s\_F}}\) represent the fitting parameter \(b\) and its standard error (\({{{\mathrm{SE}}}}\)) considering different sources of uncertainty (\(S=s1,{s}2,{s}3\), representing sources 1, 2, and 3) and using different fitting methods (\(F=p,{d}\), representing pre-binned and direct fitting methods). \(N({b}_{{s\_F}},\,{{{{{\mathrm{SE}}}}}_{{s\_F}}}^{2})\) represents a normal distribution with a mean of \({b}_{{s\_F}}\) and a standard deviation of \({{{{\mathrm{SE}}}}}_{{s\_F}}\). b The probability distribution functions of the parameter \(b\) in ocean, land, and global scales, where the point and errorbar represent the best estimate (i.e., the mean value) and its 5–95% confidence interval.
