Evaluating Elevation Change Thresholds between Structure-from-Motion DEMs Derived from Historical Aerial Photos and 3DEP LiDAR Data

Round 1

Reviewer 1 Report

This paper provided a method to reconstruct DTMs from historical aerial photographs that is comparable to LiDAR data and evaluated elevation change thresholds over multi-decadal scales. The result highlighted that a relationship between aerial photo scales and predicted vertical accuracy can be used to define geomorphic change thresholds over time. This paper is clearly stated and well organized to answer research questions described in the introduction section. However, there are minor points in which the authors should clarify the ideas of the paper:

Terminology

The authors use terminology digital terrain model (DTM), digital elevation model (DEM), and digital surface model (DSM) in this paper. Among three, DSM is differentiated from the others. However, the terminology, DTM and DEM, needs to be clarified in the first place since DTM and DEM can be interchangeable in some countries.

Details about filtering procedures

The authors use a filtering method to produce a close approximation of a DTM for each historical DSM. It would be better if the authors stated how filtering procedures can eliminate the effect of surface features such as vegetation and buildings, and what the thresholds of filtering used in this study. This is because this filtering stage can affect the quality of output.

Detail description for Line 247-248

In the conclusion section, the authors stated that smaller scale photography may be adequate for detection elevation change depending on the scale of analysis. However, it is not clear how this conclusion derived from the result. It would be better if further details are addressed to support.

Editing

Line 198-201: left arrangement needed

Line 217 (Table 4): SFM -> SfM

Line 222: Thresholding. -> Thresholding

Author Response

Please see the attachment. 

Author Response File: Author Response.docx

Reviewer 2 Report

Please see the attached

Comments for author File: Comments.pdf

Author Response

Please see attached document.

Author Response File: Author Response.docx

Reviewer 3 Report

No suggestions; the paper is really sounding and well written. Conclusions are coherent with the analisys.

Author Response

The authors would like to thank the reviewer for the review and the comments.  We appreciate that the conclusions were considered supported by the analysis.

Reviewer 4 Report

Thank you very much for this manuscript.

This is a nice little paper which could be published with only minor changes.
However, I would like to encourage the authors to expand a little bit particularly to make the results work as a guide for others to actually apply the methods to the vast archive of historical imagery around the world .
Below some comments with additional ones in the marked up pdf.
What needs to be supplied are the parameters used in the Agisoft software as these will have an influence on the results. You mention 60% forlap, but what about the side lap?
A map showing for example the photos as frames over the terrain would be useful.
Please clarify why you have gone for the Adobe Photoshop method and what you actually did to obtain consistently sized photos with the nadir point in the centre.
The Adobe photoshop step seems to be rather simple.e have you found /looked at/ considered this publication software? Works brilliantly!
Salach, A., 2017. SAPC – application for adapting scanned analogue photographs to use them in structure from motion technology, in: ISPRS - International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Presented at the ISPRS Hannover Workshop: HRIGI 17 – CMRT 17 – ISA 17 – EuroCOW 17 (Volume XLII-1/W1) - 6–9 June 2017, Hannover, Germany, Copernicus GmbH, pp. 197–204. https://0-doi-org.brum.beds.ac.uk/10.5194/isprs-archives-XLII-1-W1-197-2017

A few years ago I tested different sfm software with the aim to be able to extract digital elevation models from single strips of aerial photography flown along the coast. While all software could work with blocks of aerial photography only agisoft managed single strips! So what would be interesting in your paper would be to compare the results you got from blocks with those for just a strip for example. Or discuss the applicability of your methods and result to just a single strip of photography.
I personally have found accuracy to be even better than what you quote but I have the benefit of using manhole covers and other access slabs to underground infrastruture for example. This saying, please provide some description of the type of GCPs you used as this is not clear and will have an impact on the accuracy results.

Generally, I think the paper mixes the accuracy question of SFM processed historical aerial photography with the method to tease out whether a DOD shows an actual change. I would really drop this second aspect (just use the RSIQ) as this has been dealt with in other papers and would concentrate on making a really good case for using SFM with historical aerial photography as this could make the paper highly citable in future papers that use this method and just want to reference the accuracy issue. To this end you just need to expand a bit more in detail what you did and then provide examples showing the results of the comparisons.

For example, are the changes gradual or abrupt, ie do the photos show this gradual process in a location? You could have a time series of elevation changes for some locations with and explanation of the process that created it.

Many thanks!

Comments for author File: Comments.pdf

Author Response

Please see attached document.

Author Response File: Author Response.docx