Quantitative Assessment of Volcanic Thermal Activity from Space Using an Isolation Forest Machine Learning Algorithm
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsComments for author File: 
Comments.pdf
English language needs only a general check.
Author Response
REPLY to Reviewer #1
The Reviewer #1 positively wrote: “The method is adequately described and the Introduction provides the necessary background information, introducing the main satellite remote sensing systems employed today for volcanic monitoring, and providing proper references. The main features of the volcanic system selected in this study are also synthetically described. Results are well discussed and showed that the proposed method is able to identify pre-eruptive increases and, often, eruptive events.”
We report the changes made on the manuscript following the referee’s suggestions.
Reviewer #1: “Lines 120-145: I think that you could use a bulleted list to better separate the different volcanoes, in order to improve the readability. Only if you want, you could also add some further information for each volcano.”
Authors: We thank the reviewer for pointing out this. We used a bulleted list to better separate the different volcanoes.
Reviewer #1: “Figure 1: I suggest adding a window with a focus on Southern Italy, to better show the position of both Etna and Stromboli volcanoes.”
Authors: We tried to add a window focused on Southern Italy, but the figure lost clarity. Therefore, we prefer to preserve the original figure and improve its resolution.
Reviewer #1: “Figure 3: Please improve the resolution of the figures; I also advise you to rotate the values on the x-axis. “
Authors: We thank the reviewer for pointing out this. We improved the resolution of the figures as you suggested, but we preferred to keep the x-axis as in the previous figure.
Reviewer #1: “Figures 4 and 5: Please improve the resolution of the graphs and rotate the values on the x-axis. I suggest considering the use of different colors and plot style. I think that the blue arrows are not the best way to show your good result. “
Authors: We improved the resolution of the figures as you suggested, but we preferred to keep the x-axis and the colors as in the previous figure.
Reviewer #1: “Figure 6: Please improve the resolution of the figures. I suggest choosing different colors for LOW, MEDIUM and HIGH (maybe green, yellow and red) and increasing the font size of the labels. Furthermore, I think that you can draw a dotted line in the graphs for each threshold value.”
Authors: We thank the reviewer for pointing out this. We improved the resolution of the figures and chose different colors as you suggested.
Reviewer #1: “Section 9: Maybe you could split it into different subsections, in order to distinguish the discussion of results for each volcano.”
Authors: As you suggest we separated the chapter in different subsections: 6.1. Thermal Anomalies detection; 6.2. Thermal Activity levels.
Reviewer 2 Report
Comments and Suggestions for AuthorsThis paper presents an innovative approach to monitoring volcanic thermal activity using satellite imagery and machine learning, specifically employing an isolation forest algorithm. The authors, Claudia Corradino et al., effectively demonstrate the potential of environmental remote sensing technologies in assessing volcanic hazards. The use of MODIS data, coupled with the application of machine learning through Google Earth Engine, represents a significant advancement in the field of volcanic monitoring.
The methodology section is well-detailed, providing clear insights into the implementation of the isolation forest algorithm. However, while the results are promising, a more comprehensive analysis of the algorithm’s performance metrics across different volcanoes would enhance the robustness of the findings.
The discussion could expand on the potential implications of this technology for real-time monitoring and risk mitigation, including how this tool can be integrated into existing warning systems operated by volcano observatories. It would also be beneficial to address any limitations or challenges encountered during the study, such as data resolution issues or the handling of false positives/negatives by the algorithm.
Overall, the paper is a valuable contribution to the Special Issue on Technologies for Forecasting Volcanic Hazards. With some enhancements in comparative analysis and deeper exploration of practical applications, it could serve as a significant reference for researchers and practitioners in the field of environmental remote sensing.
Author Response
REPLY to Reviewer #2
The Reviewer #2 positively wrote: “Overall, the paper is a valuable contribution to the Special Issue on Technologies for Forecasting Volcanic Hazards. With some enhancements in comparative analysis and deeper exploration of practical applications, it could serve as a significant reference for researchers and practitioners in the field of environmental remote sensing.”
We report the changes made on the manuscript following the referee’s suggestions.
Reviewer #2: “The methodology section is well-detailed, providing clear insights into the implementation of the isolation forest algorithm. However, while the results are promising, a more comprehensive analysis of the algorithm’s performance metrics across different volcanoes would enhance the robustness of the findings.
The discussion could expand on the potential implications of this technology for real-time monitoring and risk mitigation, including how this tool can be integrated into existing warning systems operated by volcano observatories. It would also be beneficial to address any limitations or challenges encountered during the study, such as data resolution issues or the handling of false positives/negatives by the algorithm.”
Authors: We thank the reviewer for pointing out this. We completed the discussion by focusing on the potential implications of this technology for real-time monitoring; in addition, we also pointed out the limitations of this method.
