Deterministic Global 3D Fractal Cloud Model for Synthetic Scene Generation

Round 1

Reviewer 1 Report

The present work tries to develop a tool for rendering 3d cloud scenes, in which meteorological data were used to make these simulated clouds to match real weather patterns. This tool seems to be useful in many applications, and the several techniques employed are very likely sophisticated or reliable. Although I cannot fully understand the technical details, it is clear that this manuscript reports a substantial work that has notable potentials in relevant disciplines. However, some obvious problems exist in this manuscript and should be corrected to make it more readable.

Major
1. For rending clouds, 3D seems to be an important new feature of this work, compared to the previous ones. But this was not clearly described in the method of this manuscript. Given the current Section 2.1, it is hard to capture the essential idea for generating 3D cloudy scene. In addition, the information of Figure 1 and Figure 2 is not adequately explained in the text. I strongly suggest that this part should be rewritten, since 3D function was declared as the key contribution of this study.

2. In the part of Section 3, the simulated results were compared with those from ABI/GOES, shown in Figure 8-11. However, such a comparison is just superficial. In fact these figures and this Section as a whole are both rather rough. The authors should provide more solid and rigorous evidences to demonstrate the enhanced capability of this new tool.

Author Response

Major
1. For rending clouds, 3D seems to be an important new feature of this work, compared to the previous ones. But this was not clearly described in the method of this manuscript. Given the current Section 2.1, it is hard to capture the essential idea for generating 3D cloudy scene. In addition, the information of Figure 1 and Figure 2 is not adequately explained in the text. I strongly suggest that this part should be rewritten, since 3D function was declared as the key contribution of this study.

Reply: I have rewritten section 2.1 to supply additional detail about how my level-set functions work. Raymarching is described in section 2 prior to 2.1. I've also added more captions to Figure 2, and more description to figure 1.

2. In the part of Section 3, the simulated results were compared with those from ABI/GOES, shown in Figure 8-11. However, such a comparison is just superficial. In fact these figures and this Section as a whole are both rather rough. The authors should provide more solid and rigorous evidences to demonstrate the enhanced capability of this new tool.

Reply: I have rewritten the comparison between the GOES ABI images and the cloud model spectral radiance. I have simulated full disc images with atmospheric transmission and path radiance. I have used GOES cloudtop altitudes to select cloudtop pixels from the GOES imagery, and have provided comparison histograms of spectral radiance for the imaged and simulated cloudtops for bands 1, 3, and 14 (VIS, NIR, and LWIR).

Author Response File: Author Response.docx

Reviewer 2 Report

1

Comments on the ‘Remote Sensing’ manuscript (remotesensing-2965404) entitled “

Deterministic Global 3d Fractal Cloud Model for Synthetic Scene

Generation” by Aaron M. Schinder, Shannon R. Young, Bryan J. Steward,

Michael Dexter, Andrew Kondrath, Stephen Hinton, and Ricardo Davila.

The authors present the generation of three-dimensional model clouds. However, there are

several points in the manuscript that need improvement. The authors need to revise their paper to address

my following specific comments.

1. List of authors: ‘1-6’ should be ‘1’, and ‘7’ should be ‘2’. Also, ‘U.S.A.’ should be added at the end

of the address.

2. Lines 68 and 268: ‘Haggstrom’ should be ‘Häggström’.

3. Line 79: ‘(Fig. 1)’ should be added at the end of the sentence.

4. Figure 2: Explain what the two panels on the left side mean.

5. Lines 106-107: Reword “vertical air motion lifting and cooling that humidity to condensation”.

6. Figure 3: This figure is not cited in the text.

7. Table 1: This table is not cited in the text. ‘High Stratus’ in the first column should be ‘Altostratus’.

Also, what is ‘Param Altitude’ in the first row?

8. Line 133: ‘(Fig. 4)’ should be added at the end of the sentence.

9. Figure 5: This figure is not cited in the text. Also, a), b), and c) should be mentioned in the figure

caption.

10. Below Eq. (2b): ‘Bext’ should be ‘Bext’. ‘m2’ should be ‘m2’. ‘Itrans’ should be ‘Itrans’. ‘Iref’ should

be ‘Iref’.

11. Figure 6: When log10 Cloud Thickness*Extinction is 1, i.e., Bext*L is 10, Cloud Reflectance is ~0.71

for single scattering albedo  = 0.9 in Fig. 6. However, Iref/I0 is 0.909 when Bext*L is 10 according

to Eq. (2b), and Iref/I0 is 0.513 when  is 0.9 according to Eq. (3a). Why are these values (0.71 and

0.513) so different?

12. Legend of Fig. 6: ‘w’ should be ‘’.

13. Line 170: ‘Gang Hong’ should be ‘Hong’.

2

14. Line 182: ‘Miles, et. al.’ should be ‘Miles et al.’.

15. Line 183: ‘Mitchell et. al.’ should be ‘Mitchell et al.’.

16. Line 185: ‘g/m3’ should be ‘g/m3’.

17. Table 2: ‘cm3’ should be ‘cm3’. ‘um’ should be ‘m’. ‘g/m3’ should be ‘g/m3’.

18. Line 187: ‘(Fig. 7)’ should be added at the end of the sentence. Also, ‘g/m3’ should be ‘g/m3’.

19. Figure 7: Although the LWC is comparable between stratus and cirrus (0.35 vs. 0.1), Bext for stratus

is two orders of magnitude larger than Bext for cirrus. Why is Bext so much different between stratus

and cirrus in Fig. 7?

20. Figure 7: In the caption, ‘[5], [7]’ should read ‘[9], [11]’. ‘[6], [8]’ should read ‘[10], [12]’.

21. Table 3: This table is not cited in the text.

22. Table 3: In the third row, ‘um’ should be ‘m’. ‘m2’ should be ‘m2’. In the 11th row, ‘vapr’ should

be ‘vapor’.

23. Line 202: ‘(Figures. 8 and 9)’ should be added at the end of the sentence.

24. Figure 8: Is the yellow line in each panel located in the same geographic location? Explain it in the

figure caption.

25. Figure 10: Is Figure 10 the same as Figure 9 but only for the full disk?

26. Figure 11: The comparison between the satellite measurement and the model simulation is not clear

due to the lack of ground emission in the simulated image.

27. Figures 10 and 11: These figures are not cited in the text. Also, ‘um’ should be ‘m’.

28. Line 223: ‘the table’ should read ‘Table 4’.

29. Line 276: ‘1992’ should be ‘1993’.

Author Response

Comments on the ‘Remote Sensing’ manuscript (remotesensing-2965404) entitled “

Deterministic Global 3d Fractal Cloud Model for Synthetic Scene

Generation” by Aaron M. Schinder, Shannon R. Young, Bryan J. Steward,

Michael Dexter, Andrew Kondrath, Stephen Hinton, and Ricardo Davila.

The authors present the generation of three-dimensional model clouds. However, there are

several points in the manuscript that need improvement. The authors need to revise their paper to address

my following specific comments.

1. List of authors: ‘1-6’ should be ‘1’, and ‘7’ should be ‘2’. Also, ‘U.S.A.’ should be added at the end of the address.

Reply: I've updated the author block.

2. Lines 68 and 268: ‘Haggstrom’ should be ‘Häggström’.

Reply: Updated.

3. Line 79: ‘(Fig. 1)’ should be added at the end of the sentence.

Reply: Added "as shown in Figure 1"

4. Figure 2: Explain what the two panels on the left side mean.

Reply: I've added labels and changed the caption. a) is a coarser Perlin fractal. b) is a finer Perlin fractal.

5. Lines 106-107: Reword “vertical air motion lifting and cooling that humidity to condensation”.

Reply: Changed the wording.

6. Figure 3: This figure is not cited in the text.

Reply: Changed to cite this figure in the paragraph following it.

7. Table 1: This table is not cited in the text. ‘High Stratus’ in the first column should be ‘Altostratus’.

Also, what is ‘Param Altitude’ in the first row?

Reply: I've added some explanation and a citation in the paragraph preceeding the table. The parameter altitude is an altitude used in calculating certain features of the weather-masks for each cloud type.. Renamed "High Stratus" to Altostratus.

8. Line 133: ‘(Fig. 4)’ should be added at the end of the sentence.

Reply: Referenced figure 4.

9. Figure 5: This figure is not cited in the text. Also, a), b), and c) should be mentioned in the figure caption.

Reply: Figure 5 referenced in line 148. Caption adjusted.

10. Below Eq. (2b): ‘Bext’ should be ‘Bext’. ‘m2’ should be ‘m2’. ‘Itrans’ should be ‘Itrans’. ‘Iref’ should be ‘Iref’.

Reply: Updated tex subscripts and superscripts.

11. Figure 6: When log10 Cloud Thickness*Extinction is 1, i.e., Bext*L is 10, Cloud Reflectance is ~0.71 for single scattering albedo  = 0.9 in Fig. 6. However, Iref/I0 is 0.909 when Bext*L is 10 according to Eq. (2b), and Iref/I0 is 0.513 when  is 0.9 according to Eq. (3a). Why are these values (0.71 and 0.513) so different?

Reply: Equation 2b (now 3b) is for the case when single-scattering albedo is 1. (This is the case where light scatters within a cloud, but is not absorbed. In this case (in one dimension), all light entering the cloud must exit at either end. (For 3d, all light entering a cloud will exit it somewhere - the reason why overcast days are not pitch-black in the visible.) For thick clouds, the light can scatter many times within the cloud, but will likely exit the illuminated surface. When even a small absorption cross section is present, the situation changes significantly - the longer the light's path-length in the cloud, (the greater the number of times it scatters) the higher the probability it is absorbed. For the case where the cross section is entirely absorptive, no light would reflect from the illuminated surface, but some small Beer's law limit amount may transmit through the opposite surface.

While I arrived at equation (3a) and (3b) (now 4a and 4b) from the numerical behavior of system (2), other authors have arrived at a similar factor in their modeling of cloud surface reflection. Shanks arrives at it in equation 4 of Shanks, Joseph G., and David K. Lynch. “Specular Scattering in Cirrus Clouds.” edited by David K. Lynch and Eric P. Shettle, 227. Paris, France, 1995. https://doi.org/10.1117/12.228943.

12. Legend of Fig. 6: ‘w’ should be ‘’.

Reply: Adjusted.

13. Line 170: ‘Gang Hong’ should be ‘Hong’.

Reply: Adjusted

14. Line 182: ‘Miles, et. al.’ should be ‘Miles et al.’.

Reply: Adjusted

15. Line 183: ‘Mitchell et. al.’ should be ‘Mitchell et al.’.

Reply: Adjusted

16. Line 185: ‘g/m3’ should be ‘g/m3’.

17. Table 2: ‘cm3’ should be ‘cm3’. ‘um’ should be ‘m’. ‘g/m3’ should be ‘g/m3’.

Adjusted.

18. Line 187: ‘(Fig. 7)’ should be added at the end of the sentence. Also, ‘g/m3’ should be ‘g/m3’.

Adjusted

19. Figure 7: Although the LWC is comparable between stratus and cirrus (0.35 vs. 0.1), Bext for stratus is two orders of magnitude larger than Bext for cirrus. Why is Bext so much different between stratus and cirrus in Fig. 7?

Reply: I had a typo when supplying the ice-water content that I had been taking as a tentative "typical value", from Mitchell's figure 5. His ice-water contents were 0.01 g/m3 (10 mg/m3) (plus or minus an order of magnitude) for the cirrus clouds he measured. 

The droplet sizes for the cirrus clouds are ~10x larger, and the total number of droplets (and therefore surfaces from which light scatters) is ~ mass_density/(density_of_water * droplet_volume). Droplet radius^3 There are ~3000x as many water droplets in a volume of the stratus cloud as ice particles in the thin cirrus cloud. There is about a factor of 200-300 difference in the extinction coefficients. The size of the ice "droplets" is also much larger than, as opposed to within the neighborhood, of the light wavelength.

While we are currently using the models from these two sets of papers, if better data is available we are more than willing to adjust our extinction coefficients. Especially for thin cirrus clouds. For water clouds, a mean free path for scattering of ~10-20 meters in the visible band should be about right. Our goal is to use whatever data we can obtain to produce radiometrically accurate synthetic scenes.

I've reworded the paragraph preceeding the extinction plots to try to include some of this explanation:

Combining this information, one particular example is plotted in Figure \ref{fig:Figure7} for stratus (10 µm, 0.35 $g/m^3$) and cirrus clouds (90µm, 0.01 $g/m^3$). Note that for these two droplet sizes and mass densities, the number of droplets/volume for the stratus cloud is 3,000 times higher than for an optically thin cirrus cloud. (8.4E7 $\#/m^{3}$ and 2.62E4 $\#/m^{3}$ respectively.)

20. Figure 7: In the caption, ‘[5], [7]’ should read ‘[9], [11]’. ‘[6], [8]’ should read ‘[10], [12]’.

Adjusted.

21. Table 3: This table is not cited in the text.

22. Table 3: In the third row, ‘um’ should be ‘m’. ‘m2’ should be ‘m2’. In the 11th row, ‘vapr’ should be ‘vapor’.

23. Line 202: ‘(Figures. 8 and 9)’ should be added at the end of the sentence.

24. Figure 8: Is the yellow line in each panel located in the same geographic location? Explain it in the figure caption.

Reply: Added some explanations to the caption.

25. Figure 10: Is Figure 10 the same as Figure 9 but only for the full disk?

Reply: I think so, but I'm going to change some of the full disk images to address the other reviewer's comments.

26. Figure 11: The comparison between the satellite measurement and the model simulation is not clear due to the lack of ground emission in the simulated image.

27. Figures 10 and 11: These figures are not cited in the text. Also, ‘um’ should be ‘m’.

28. Line 223: ‘the table’ should read ‘Table 4’.

Changed

29. Line 276: ‘1992’ should be ‘1993’.

Changed

Author Response File: Author Response.docx