Monitoring Land Cover Change by Leveraging a Dynamic Service-Oriented Computing Model
Round 1
Reviewer 1 Report
This paper proposed an online interoperable dynamic service computing model, and developed a prototype system based on the model, which demonstrates and validates the online land cover change monitoring. I think this model and this system are somewhat innovative and very practical， and they will be popular with users.
What I want to know is how does the author consider the accuracy of data processing? Has the author compared the results obtained using this model and system with those obtained by other methods?
Author Response
Please see the attachment.
Author Response File: 
Author Response.docx
Reviewer 2 Report
The authors proposed a Dynamic Service Computing Model (DSCM) for monitoring land cover change (LCC) and I think the work is interesting for the users who need to detect the change of LCC. The paper is overall well written. Some points of my confusion are as follows:
(1) I think too much for the description of the service and data preparation, making the paper to be like an instruction manual rather than an article of science and technology. I think the authors should focus more on algorithms about the dynamic detection model.
(2) The authors give a Walk-Through example for the DSCM, however, there is no validation of the result and we do not know what the accuracy of the detection results is.
(3) Maybe the paper will be more useful if why the LCC happens is conserved in the detection process.
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Reviewer 3 Report
1. In the abstract, it is necessary to state which satellite data is used
2. The abstract needs to be conveyed, "Does the Dynamic Service Computing Model (DSCM) have the same ability to detect changes in land cover from images with different spatial resolutions?"
3. Qualitatively stated that "MDCM is more effective" for detecting changes in land cover, what are this model's drawbacks? Beyond this, there needs to be a reason why this MDCM needs to be developed, bearing in mind that there are already many other similar webs (cloud) processing.
4. In lines 82-90, it is better to present it in a complete paragraph, with options (1)-(3) that blend with the previous sentence.
5. The author must state why the three questions in lines 82-90 must be asked.
6. Does this MDCM not require training samples? Because in Table 1, it is not written that "training samples" is one of the required parameters.
7. Behind the importance of presenting the change detection scheme in Figures 2 and 3, the author seems to have also conveyed the type of classification algorithm used or available in MDCM. This is also related to number 6, does it only use unsupervised techniques and or maybe it does not have the option of using various machine learning algorithms
8. The author needs to provide examples of ComplexData, LiteralData, and BoundingBoxData (Line 162)
9. What is BPEL? Please explain (Line 198).
10. What is OWL? Please explain (Line 218).
11. DFin consoles, CSinconRules, ReinconRules, and then use SWRL (227). What is the difference between these three restrictions? Describe them in table form, and under what conditions can these rules be used?
12. What is Rad data? Is it the same as radiance data? (253)
13. Do these WebApps have the ability to download data like Google Earth Engine?
14. Why is this sentence so important, "Black box" is an approach that can be used without the user knowing how its internal algorithms work. The user only needs to know the input and output characteristics (283)"?.
15. "After cropping, each piece of data is 300*300 pixels in WGS84 coordinates". This means 25-26 grids from Landsat 8 or 36-37 grids from Sentinel 2 satellite imagery for each data processing. Is not dividing 300*300 too big, considering that there will be many land cover classes in an area as large as that number of pixels? Why not smaller or greater than 300?
16. Several other satellite images use CRS outside WGS84, e.g. UTM. Does MDCM also have the ability to reproduce image data with a different CRS? What if the image used has a different CRS? Does MDCM accommodate this? Geometric correction of both images?
17. Does MDCM also have the capability to perform image pre-processing? This is important because the image correction level also determines the quality of the image classification results.
18. Was the data validation of the classification results carried out? If yes, in what way?
Author Response
Please see the attachment.
Author Response File: Author Response.docx
