Comment on 'The local isotropy hypothesis and the turbulent kinetic energy dissipation rate in the atmospheric surface layer' by M. Chamecki and N. L. Dias (October B, 2004(October B, , 130, 2733(October B, -2752
SUMMARYThe velocity-velocity-increment correlation is shown to be an incorrect method for determining the interaction between the energy-containing and inertial-range scales of a turbulent flow. As a consequence, the correlation cannot be used as a means for testing the local isotropy hypothesis.
KEYWORDS: Correlation function Inertial range Kolmogorov Structure functionOne main objective of Chamecki and Dias (2004) (hereafter CD04) was to assess the applicability of the local isotropy hypothesis (LIH) to atmospheric surface layer turbulence. Their analysis used three methods to test the applicability of the LIH. Two are based on the dimensional arguments of Kolmogorov (1941a,b), hereafter K41. Each of these relies on the notion that turbulent velocity increments are locally isotropic when spatial scales are sufficiently small, and therefore are independent of largescale forcing. It should be noted that the notion of local does not extend to the large scales. LIH requires that the Reynolds number (typically the Taylor microscale Reynolds number R λ ≡ σ u 1 λ/ν) is sufficiently large. Here, σ u 1 = u 2 1 1/2 and λ = σ u 1 / (∂u 1 /∂x 1 ) 2 1/2 ; the subscript corresponds to the longitudinal component of fluctuating velocity, (u 1 ), position, (x 1 ) or separation, (r 1 ). Time averaging is denoted by · .Implicit within the framework of K41 and LIH is the assumption that the inertial-range (IR) scales are statistically independent of the energy-containing scales. This assumption underpins the second method used by CD04 for testing the LIH. This method was essentially introduced by Praskovsky et al. (1993), who suggested using a longitudinal correlation function ρ u k 1 ,( u 1 ) l as a measure of the energy-inertial scale interaction. The general definition of the correlation function is given bywhereand ( u 1 ) = u 1 (x 1 + r 1 ) − u 1 (x 1 ). Equation (1) has been used by numerous authors with different power exponent (k, l) combinations; all have purported that the correlation is a suitable indicator of the level of energy-inertial scale interaction. The simplest (with k, l = 1, 1) has been used by CD04, Katul et al. (1995aKatul et al. ( , 1997 and Xu et al. (2001). Other higher-order correlations, with (k, l) > (1, 1), have been used by Katul et al. (1995aKatul et al. ( ,b, 1997, Praskovsky et al. (1993) and Xu et al. (2001).Smalley and Antonia (2001) and Hill and Wilczak (2001) have argued that ρ u 1 , u 1 is an inaccurate measure of the energy-inertial scale correlation, since u 1 must contain a non-zero correlation with u 1 . Although CD04 acknowledged that Hill and Wilczak (2001) commented on the applicability of ρ u 1 , u 1 , this was misinterpreted by CD04 and they suggested that ρ u 1 , u 1 is a better means of testing for homogeneity rather than isotropy. However, this statement is in err...