Accuracy Evaluation for Plan-Reliefs and Historical Maps Created during WWI in Northern Italy

Round 1

Reviewer 1 Report

This paper explores a procedure to reconstruct plan-reliefs created during World War I using Structure from Motion techniques; the accuracy/precision evaluation of the reconstruction was conducted by comparing the results with a current DTM; in addition, different historical maps were georeferenced and a detailed analysis was conducted to identify the best transformation considering map scale, map origin, conservation status and GCP number. This study is very interesting and important for application in the field of historic heritage preservation.

The abstract is well organised and highlights the significance of this research and the introduction clearly outlines the research problem.

Data used in the study are clearly presented and results are presented and explained with detailed and proper statements

Finally, the color scheme for the figures representing the difference between digital models is described only in the text, I suggest adding the legend in the figures as well (figures 4 and 5).

In my opinion, the paper can already be published.

Best regards.

Author Response

Thanks.

Reviewer 2 Report

This is an excellent work on the georeferencing and evaluation of historical relief plains and maps. A professional work on conception, documentation, methods, processing and evaluation. I like the fact that the georeferencing process is assessed with both on-ground (m) and on-map (mm) errors, a method I promote too. Minor corrections and additions should be made to better objectify and visually present the results. I also added some minor suggestions for pre-processing map scans and DTMs - please consider if these can be easily implemented. Detailed suggestions listed as comments in PDF file (attached).

Comments for author File: Comments.pdf

Some minor linguistic errors or inaccurate wording: table 2; line 263;  line 296, see comments listed in PDF file

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

Some minor problems. I use to use the word(s) in form of „georeference” and „georeferencing”. Instead of „Austria-Hungarian”, I suggest the form „Austrian-Hungarian”. In Table 1, the Map 37-07-02 is indicated as IT, however in Fig. 2(b), it is indicated as Austrian.

My main concern is about the obvious mis-use of the projections/coordinate systems of the historical maps. The information summarized in Lines 193-199 are at least partially false. The Genoa1902 datum might be used, but the Reference 24 does not contain any hint of „Sanson-Flamsteed projection” (I know, I wrote it :) ). I suggest to check the old Gauss-Boaga first (and I offer my experience to the Authors, if needed, to identify it). The projection of the Austrian 75k series is the quite complex polyhedric solution (with differently centered Stereographic projections for all sheets). It can be firmly substituted by sinusiodal projection… aye, THIS is Flamsteed :) ). See Molnár & Timár (2009) „Mosaicking of the 1:75 000 sheets of the Third Military Survey of the Habsburg Empire” DOI: 10.1556/AGeod.44.2009.1.11 The selected references (Lines 193-199) refers to CADASTRAL map projections. May I also suggest to publish some example maps as Supplementary Data, e.g. in lower resolution scan, such as 150 dpi?

To the plan-relief georefence: I’ve never seen this kind of process anywhere before, so I acknowledge this as a kind of pioneering work. However, as these old plan-reliefs were supposedly made in approximate and hand-made manners, their accuracy analysis refers to this process. The only question of mine is concerning the Table 2. Are there any, even indirect connection between the horizontal and 3D errors? I mean, if the horizontal control provides e.g. 40 meter error, the elevation accuracy can be marred (1) by this and (2) by the building process of the plan-relief.

To the historical maps georeference: In case of any map, whose terrain extents exceed cca. 20 kilometers, it is always the best solution to to define the GCPs in the map’s native coordinate system. In case of the Italian map, I see map grids. Unfortunately, I cannot give a hint of the coordinate system of the map shown in Fig. 2(a) but I repeat, I offer my experience to the Authors if it is needed. So, if map grid is provided, no terrain points are needed for GCP, use just the coordinate line crossings, after the proper definition of the coordinate system (geodetic datum, map projection, with proper parameters). The terrain elements can be used then just for control points (CPs). I assume, the relatively high RMSE values (Table 3) are because of the use of non-native coordinate system (projection). I shall also ask: are these figures indeed expressed in meters? (some softwares give the errors in pixel units). Also, 0.2 mm of „reading error” (caption of Table 3) is really optimistic, I advise to use a value of 0.5 mm. And, 40-50 meters error is not so bad for a scanned 1:75000 scale map sheet of this age.

I think, the above thing (use the non-native grid coordinates in GCPs) made any Fisher-test-like thing. I suggest to use Helmert transformation in any case (with the map’s own projection). From this point, I suggest to remake the old map georeference according to this and re-evalute the Discussion and Conclusions accordingly. If the Authors insist on the current calculation, I'm not against publishing it, I'm just saying that it's not the correct way.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 4 Report

In general (e.g. in the introduction) you could refer to "digital editing of cartographic sources" with work you've done (you may find some references on the Web). Historical sources editing is a process of represesting historical artifacts, usually textual, but recently also iconographic and cartographic, such as maps, plans, etc.  Digital technologies significatly influenced historical sources editing. The paper would definitely benefit by introducing this concept: it would be more open towards humanities and show interdisciplinary approach.

please be concise with using "historic map". This is a proper term for an "old" map.

geo-referentiation --> "georeferencing" shoud be used

the vertical scale of plan-relief (table 2)

Could you elaborate why you didn't use geographic coordinates to georeference historic topographic maps, and focused on GCPs instead?

ln. 312: "The outcome of the geo-referencing process yields satisfactory results." Could you compare with other similar studies? What do you mean by "satisfactory"?

4.2 Surface comparison -> can you collect the results in a table, like you did with historic maps georeferencing?

Could you compare your results in plan-relief analysis with results obtained by previous scholars? In the introduction you cite some of them (e.g. Niederöst, Macher). It would be good to refer to them in the conclusions as well.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 3 Report

Dear Colleagues, however I still have some points to argue with (see below, for consideration in upcoming publications), I can accept the paper in present form.

1. I'm still quite convinced that in case of Italian topo maps of 1918 were not printed in Sanson-Flamsteed projection. This projection CAN BE USED AS AN APPROXIMATION MODEL of the native projection, of course, and in this small extent, the error will be quite low.
2. In Figure 2 there are some map projection kilometer grid(s). In case of Italian map (a), without further analysis I cannot guess the coordinate system type and parameters, albeit the last two km digites are shown well. In case of Austrian maps, only the grid is given without figures - however it is not so hard to guess the system. However, I assure that even using GCPs, the maps can be geo-referred with quite good results.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf