“…The only exception takes place at high values of Ñ j j and Ñc j j where there is insufficient data in the DNS dataset to accurately predict the value of the PDF and the regions with high probability of finding Ñc j j % 0, which log-normal distribution fails to predict. The discrepancy between the log-normal distribution and the tail of the scalar gradient PDFs is consistent with several experimental (Geyer et al, 2005;Karpetis and Barlow, 2002;Markides and Mastorakos, 2006;Sreenivasan and Antonia, 1997;Su and Clemens, 2003) and computational (Hawkes et al, 2007;Knaus et al, 2012;Malkeson et al, 2013;Pantano, 2004;Ruan et al, 2012) findings in the context of passive scalar mixing, non-premixed, and partially-premixed combustion. Despite some inaccuracies, the log-normal distribution appears to do a reasonable job for the purpose of parameterising Ñ j j and Ñc j j, which in turn could be used for modeling the scalar dissipation rates and cc .…”