Thank-you for your revised manuscript for which I now have some referee comments and some of my own. The referee wants “minor revisions” as copied below. I also have some comments (further below). Line numbers refer to your most recent manuscript.

“Review on “Spatial distribution of turbulent mixing in the upper ocean of the
South China Sea” (os-2016-80) by X.D. Shang, C.R. Liang and G.Y Chen.
The authors have improved the manuscript based on the previous comments. I can recommend publication after the following points are clarified:
1) Line 286-294: This should be removed since the same context is repeated in Summary.
2) Line 223: PDF estimator -> PDF estimator (histogram)
3) Fig.10: Theta is not a test statistic. The linear regression should be tested with t-test. Also R^2 needed to be mentioned. R^2 (coefficient of determination) is a standard variable defined in the most statistical book. See the following site as an example:
http://www.r-tutor.com/elementary-statistics/simple-linear-regression/coefficient-determination
Also this regression equation should be used to predict accurate dissipation rate from the model.
4) Line 385: What parameterization?”

My Editor comments.
Firstly, regarding the referee comments
1) I agree. But in the Summary, about line 437, you also need to mention region 2.
3) I agree. The point is that the variance of the data from the line fitted by regression, i.e. “residual variance”, should be significantly less than the original data variance. R^2 is the proportion of variance “explained” by the regression. 1-R^2 is the reduction factor {(residual variance)/(original data variance)}. This is similar to but not the same as your θ; in both cases, the smaller the better. Statistical tests concern whether the residual variance is small enough that the coefficient of regression is significantly different from zero. An important factor in significance is how many independent data points there are in the regression. I am not an expert in all this and suggest that you look at the Web site suggestion by the referee.

Please also address referee comments 2) and 4). Regarding comment 4) you may wish to label the equations after lines 298 and 301 if you want to specify “parameterizations” later.

Line 15. Better “. . , induced by shear instability, . .” (add “,” twice)
Line 28. Better “. . concentrated over rough topography [ . .”
Line 42. Better “. . the Pacific.”

Lines 56. Better “. . we present direct microstructure measurements that cover . .”
Line 57 “In addition . .” to line 72 “. . mixing level.” I think this is mostly motivation for parameterization and would be better as a separate paragraph. Also the second sentence “In order to estimate . . better.” might best be at the end of this separate paragraph.
Line 71. “. . dependent on an ocean turbulence model . .”

Line 117. “. . and the length of the profiler itself.” After “itself” insert “as the profiler tends to follow the larger-scale flow”?
Line 121. Best to define the buoyancy frequency N here by N^2 = - g/ρ ∂ρ/∂z.
Lines 122-123. Better “. . are the respective zonal and meridional . . ADCP. The mean . .”

Figure 4. The grey curves are rather faint and thin; they do not show well on a printed version.
Lines 183-185. The sentence “Another . . in the water column.” does not seem to add to what was said in lines 173-175. Or do you need to replace “in the water column” (which is anywhere) with “below the surface mixed layer”, “in the thermocline” or “below the thermocline” (for example).

Line 231. “. . <ε>T displayed a decreased trend towards . .” This means that the trend changed. Do you mean that or just that <ε>T decreased, i.e. “. . <ε>T decreased towards . .”
Line 239. Better “. . different times and the measurement time . .”
Lines 242, 248, 443. “measure” -> “measurement”

Line 394. Not “predication”. Maybe “prediction”? Or “. . It is typically evaluated . .”

Dear Authors
Thank-you for your latest revisions. I will now send the manuscript to Copernicus for the final stages of publication. This will include copy-editing. I expect some use of English to change and you should check that the final form retains your intended meaning. Meanwhile here are a few “technical corrections” where the intended meaning may not be clear to a non-expert copy editor.

Lines 44-45 “the fine-scale parameterization”. If this was the same parameterization as in Tian et al. (2009), then this is OK. Otherwise, “a” not “the”.
Line 62. Omit “the”.
Line 107. “Figs. 2b-2g”
Line 119. “-g/ρ ∂ρ/∂z” Either brackets (. .) around g/ρ or no spaces in g/ρ and good space before ∂ρ/∂z as written here.
Line 183 “underestimated”. At present this sentence means “Richardson number . . might be underestimated”. However, I suspect you want to say shear might be underestimated, hence Richardson number overestimated and possibility of instability underestimated.
Line 253. “and” not “which” (which refers to the item immediately before; here you want to refer to the wind).
Line 276. “. . internal tides are candidates . .” (omit “the”).
Line 366. “larger” not “smaller” (larger is better!)
Lines 385-386. I think “. . stationary; the energy . .” is better. Otherwise, “maintain” -> “maintaining” on line 386.