Review Reports - Burnt-Area Quick Mapping Method with Synthetic Aperture Radar Data

Round 1

Reviewer 1 Report (New Reviewer)

The main contribution of this paper is the assessment of forest and field fire detection method and suitable method selection for Indonesia. The paper is well organized. However, all the figures show no geo-reference. Please add longitude and latitude for all the figures. Besides, please add study area map for better present the location and basic information of the study area. The authors select RF as the best method for their study area. The existing analysis and explanation are not reasonable enough. Please provide more details.

Author Response

Thank you for the comments, it is really helpful to make our manuscript better. We made the revisions according to points provided as follows:

Point 1: The main contribution of this paper is the assessment of forest and field fire detection method and suitable method selection for Indonesia. The paper is well organized. However, all the figures show no geo-reference. Please add longitude and latitude for all the figures. Besides, please add study area map for better present the location and basic information of the study area. The authors select RF as the best method for their study area. The existing analysis and explanation are not reasonable enough. Please provide more details.

Response 1:

The geo-reference longitude and latitude has been added to Figures 1, 2, 5, 7, 8 in the revised order
The location and study area map is shown in Figure 1 (formerly Figure 7) and the description is added on Page 3 Section 2.1 Location and Data
The analysis and explanation of the best result from SVM (not RF) is added on Page 15:

As our research aims for quick mapping method, the second parameter to test is the processing time for each methods. Burned area mapping using the SVM and RF classification method each requires the process time for 1 hour, while CART takes 2 hours. This process time represents the temporal resolution of the map produced or in other words the period of updating the map. When compared to the minimum updating period of the forest fire mapping process which is currently valid, which is 12 hours [33], the quick mapping method can speed up the process so that this information can more quickly reach the fire mitigation team in the field.

Author Response File: Author Response.pdf

Reviewer 2 Report (New Reviewer)

Authors present Burned-Area Quick Mapping Method with Synthetic Aperture Radar and VIRS Data in this manuscript. And comments are listed as follows.

1 As presented in this title, authors state that they propose a quick mapping method. But the efficiency of the proposed method is not presented and described in this manuscript.

2 Description of novelty of this manuscript should be strengthened in the abstract and introduction.

3 Authors should present more formulas to describe the proposed method.

4 This manuscript has no comparison in the part of experiment. Authors should present enough comparison.

Author Response

We appreciate the comments, it contribute to improving our manuscript. We made the revisions according to the points as suggested

Point 1: As presented in this title, authors state that they propose a quick mapping method. But the efficiency of the proposed method is not presented and described in this manuscript.

Response 1: The proposed method is measured for its accuracy and process time between SVM, RF, and CART methods as the name “Quick Mapping” suggest, it is described on Page 15:

Point 2: Description of novelty of this manuscript should be strengthened in the abstract and introduction

Response 2: Added the novelty as “The Quick-Mapping employed in this research provides faster mapping time compared to currently employed method based on field report data to enable better and more efficient firefighting effort” in the abstract and paragraph 6 of Introduction as:

Finding the best algorithm for GEE to detect burned area in Indonesia have never been done before so that in this study we develop a rapid assessment of burned area detection method using GEE and evaluate the performance of available machine learning algorithm in GEE for detecting the post-fire burned area in Indonesia using radar sensors. This research aims to quickly map burned area due to forest fire using cloud computing of SAR image data with SVM, RF, and CART classification methods being considered and the best result is selected to provide forest fire impact map for effective disaster mitigation efforts. The proposed method provides faster mapping time compared to currently employed method based on field report data to enable better and more efficient firefighting effort.

Point 3: Authors should present more formulas to describe the proposed method.

Response 3: We added Burned Area Index (BAI) in Equation (3) as the basis for the Normalized Burn Ratio (NBR) in Equation (4) from which the equations (5) and (6) is modified and derived.

Point 4: This manuscript has no comparison in the part of experiment. Authors should present enough comparison.

Response 4: As in Response 1, the experiment is compared to the current existing method of field data collecting as descibed on Page 15

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report (New Reviewer)

I think this paper has been revised carefully according to the comments from reviewers. This manuscript can be accepted for publication in this version.

This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.

Round 1

Reviewer 1 Report

Mapping fire scar with either optical remote sensing or SAR is actually quite mature. The current study lacks novelty, with no significant new methods or scientific insights. Furthermore, errors are here and there throughout the entire manuscript.

Reviewer 2 Report

The paper addresses an interesting and relevant topic but requires major modifications. It is well written and structured. It evaluates the applicability of different machine learning methods implemented in a cloud computing environment for a fast mapping of burned areas.

My major concerns are the limited novelty and the missing description and discussion of the SAR data. The different machine learning techniques as well as their applicability for forest applications have been demonstrated and published already. Here, a new facet is the fast assessment of burned areas which require a data source being available quickly and reliably (SAR data) and sufficient computing power (cloud computing).

I suggest to add a description of the SAR data, their potential and limitations. The quality assessment and discussion should be extended accordingly. More classes then “burned” and “unburned” would be appreciated.

Generally, the machine learning part is well written but the paper misses an adequate treatment of the SAR data which are the main source of information.