Comparison of Multi-Temporal PlanetScope Data with Landsat 8 and Sentinel-2 Data for Estimating Airborne LiDAR Derived Canopy Height in Temperate Forests

Round 1

Reviewer 1 Report

The manuscript compares PlanetScope, Sentinel-2 and Landsat 8 images for estimating canopy height in temperate forests of Japan. The analysis focuses on the multi-temporal aspect by comparing single, multi-seasonal and time-series features for the three sensors compared, used within random forest models. Although the manuscript follows a simple line, it is of interest for a wide community dealing with scaling-up localized measurements using optical satellite images.

Major comments:

- It would be of interest to show which features had the highest importance in the RF models. Are textural features worth the effort? Maybe for PlanetScope but not for Landsat? Are time series features contributing significantly?

- In Table 2, you used 11 bands from Sentinel-2; I think this should be 10 (2,3,4,5,6,7,8,8A,11,12). Better mention for each sensor which bands were considered.

- Figure 2,3,4: maybe you can integrate all three sensors for each of the figures in a combined graph; this would help in comparing the performances better.

Minor comments:

- Be more specific and avoid using vague statements. For example, “slightly more accurate” on L25, “was almost the same” on L113-114 etc.

- L51-53, citations needed.

- L85, “which was” should be “which has”

- Figure 1 – scale bar for the big map needed.

- Table 1. Need to explain what “average valid pixels” is.

- Figure 6 caption is confusing. Maybe split this figure into 2 figures for clarity.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

General

The authors explore PlanetScope data for forest height estimations. The paper is relevant as it assesses the useful applications of a new generation of sensors. Although generally the paper is well-written, it could benefit from being checked by a native speaker. I do have some comments that can be found below.

Major remarks

L 146: why did the authors choose NDVI? It is well-known that NDVI easily saturates. Indices such as NDMI are much more suitable. I am aware due to the limited bands in Planet this is not possible but it would be interesting to see the model prediction of Sentinel and Landsat using the indices that make use of the red-edge bands (i.e. SWIR1 and 2) as the authors also mention in the discussion. Furthermore, I am missing the comparison with radar (SAR) data as at the end you want to be able to select the best sensor for larger scale monitoring tree height.

L110: As mentioned by the authors, many studies (e.g. Houborg et al., 2019; Mulatu et al., 2019) looked into low radiometric quality of PlanetScope images. Can you please provide more information on the claim that this did not or only marginally affected the results of this study?

Minor remarks

L70: Other studies also looked into height and AGB with multiple images from the Planet platform (Mulatu et al. 2019). This is a relevant study that should be mentioned here and further discussed in the discussion.

L74: what do you mean by time-series variables and how are they different from seasonal variables?

L85: please correct the sentence “which was” to “which had...coverage”

L117 & L123: please remove “distinct” here and in the rest of the manuscript. Furthermore it would be more informative to see a figure which presents the useable data points for each sensor during the year(s) in addition to table 1.

L179: I recommend to add a reference for the Rand index.

L221: Why is R2 not in percentage?

L268: Could you elaborate on the seasonal data being so effective in comparison to time-series and single observations? What could be the reason since you have evergreen trees...noise in the satellite data?

L275: From the results I find the accuracies using Planet data only marginally better. Only if you want very detailed spatial information this could be preferred

L304: This sentence is not clear to me. do you mean Planet data combined with a lower resolution sensor?

References

Houborg, R.; McCabe, M. Daily Retrieval of NDVI and LAI at 3 m Resolution via the Fusion of CubeSat, Landsat, and MODIS Data. Remote Sens. 2018, 10, 890.

Mulatu, K.A.; Decuyper, M.; Brede, B.; Kooistra, L.; Reiche, J.; Mora, B.; Herold, M. Linking Terrestrial LiDAR Scanner and Conventional Forest Structure Measurements with Multi-Modal Satellite Data. Forests 2019, 10, 291.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 3 Report

Please see review comments. 

Comments for author File: Comments.pdf

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

The authors improved the manuscript significantly. I have no further major comments.

Author Response

To reviewer 1:

The authors improved the manuscript significantly. I have no further major comments.

Response to the comment:

Thank you very much for reviewing our manuscript again.

Reviewer 2 Report

The authors have made significant improvements to the manuscript and addressed the concerns I raised. I have only a small remark about:

L357-359: “Considering the comparable accuracies achieved by Sentinel-2, Sentinel-2 might be used for finer spatial resolution and PlanetScope might be limited for mapping detailed spatial information.”

This should be the other way around. Sentinel has a less fine resolution. I would write something like  “ Sentinel -2 might be more applicable for larger scale monitoring, while PlanetScope might be used for near real-time and detailed monitoring.”

 

 

Author Response

To reviewer 2:

Thank you very much for reviewing our manuscript again. We revised the manuscript based on your comment.

 

The authors have made significant improvements to the manuscript and addressed the concerns I raised. I have only a small remark about:

L357-359: “Considering the comparable accuracies achieved by Sentinel-2, Sentinel-2 might be used for finer spatial resolution and PlanetScope might be limited for mapping detailed spatial information.”

This should be the other way around. Sentinel has a less fine resolution. I would write something like  “ Sentinel -2 might be more applicable for larger scale monitoring, while PlanetScope might be used for near real-time and detailed monitoring.”

Response to the comment:

We have revised the sentence based on your suggestion. The following sentence was also revised to match the content of the suggested sentence.

L348-354: “Sentinel-2 might be more applicable for large scale monitoring, while PlanetScope might be used for near-real-time and detailed monitoring. In addition, because of the high spatial and temporal resolution of the PlanetScope data, robust forest monitoring using PlanetScope data, such as the near-real-time monitoring of forest attributes with a finer spatial resolution, may be possible.”