Dear authors,

Thanks for your work to revise the manuscript. The new algorithm is well described and the new outcrop dataset is valuable for a range of work. Therefore, I am happy to accept the manuscript with some technical corrections, suggested below.

P3L29: add a space between 30 and m, and between 15 and m.

P3L31: add a space between 100 and m, and between 30 and m.

P4L32: Pixel values were taken for the classification development (Fig. 6), but location of these images are not presented in this paper. Is it possible to add them to Figure 9?

P6L24 and Figure 5: add “EPSG 3031” to specify the projection.

P7L9: make the square of km to the superscript.

P7L13: define “SD” after standard deviation. Currently SD is not defined in the manuscript.

P7L29: Please define “overall accuracy” shown in Figure 11b. How is it related to “100% normalized accuracy” defined with Equation 3?

P8L16: Equation 3 can appear earlier, because the classification accuracy is presented in Figure 11, which is cited first time at P7L25.

P9L20: Add space after the number, 100m and 60m.

Figure 1: If possible, add Landsat8 scene number, like Fig. 3.

Figure 2: Spectral reflectance for Landsat 8’s bands 7-11 are not shown in this figure, though these bands are used in the algorithm. Can you briefly describe it in the text?

Figure 5: At the top beginning box, add version (version 7?) of ADD dataset. To a box after two datasets are merged, Add EPSG number to specify the projection (EPS3031, I believe). Correct a typo “Col(n)vert raster mosaic to polygon”

Figure 6: It’s difficult to see discarded and selected domains separated by the dashed bar. Is it possible to put different background color masks to indicate the discarded domain and the selected domain more clearly?

Figure 7: Is it better to show the graticule of 82 degrees 40 minutes S? South of this graticule is not covered with the Landsat 8 and thus with the new classification (P8L26 and P9L3).

Figure 8: Include a sentence in the caption so that readers know the location of 10 km x 10 km scenes are indicated in Figure 9. Label “I” should be non capital “i”.

Figure 10: It is hard to see some curves. Is it possible to use different colors together with labels left or right ends of the curves?

Figure 11: It’s probably better to simplify the legends so that it only shows “omission”, “commission”, and “correct”. Equations in three lines are hard to understand, and it is well explained in the caption anyway. Also, move the label a and b indicating the panels outside of the box. Add period between “Equation (3)” and “(b)”. Spell out QC (quality control).

Supplement: The attribute table associated with the shape file includes gridcode and grid, which are not self-explanatory. It is best if authors can include metadata in the zipped file, which briefly explain the new algorithms, limitations, statistics, and explanations of the attribute table.

Thank you for choosing TC/TCD to publish your valuable work.

Best regards,

Kenny Matsuoka
Editor, TC/TCD