The Results of Digitizing Historical Maps: Comparison of Lithuanian Land-Use Structure in the 19th and 21st Centuries

Round 1

Reviewer 1 Report

I would recommend also mentioning the aspects of land ownership. For example, in parts 341-343, consequences of political decisions – whether the private land was collectivized, or nationalized..., and how it changed later, for example, whether there was a reprivatization,...   

Author Response

Response for reviewer is given in the file. Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

Dear authors,

This research paper describes actual topic – Long-term Land-Use Changes as a Tool for Indication of Political and Economic Changes: the Results of Digitizing Historical Maps. In their article authors notice, that historical maps are a unique source of information for studies on land use/land cover (LULC). Since there are no georeferenced and digitized LULC national level maps in Lithuania, authors georeferenced 27 selected sheets from 1846-1872 topographical maps of the Russian Empire covering the territory of the modern Republic of Lithuania. Thus, the authors present the results of processing the aforementioned set of analogue historical map sheets into a digital map of land use in 19th century Lithuania.

And I would like to share with authors some doubts and remarks too: it seems important to notice, that it would be needed to concentrate on the abstract of the study, presenting more clearly the results and conclusions of the study. As well, when developing "Conclusions" section it would be needed to include to the debate more future oriented theoretical implications, thus accessing deeper concluding insights.

 

Author Response

Response for reviewer is given in the file. Please see the attachment.

Author Response File: Author Response.docx

Reviewer 3 Report

The authors describe an effort for the digitization of historical land use from historical maps in Lithuania in the 18th century. The paper is interesting to read, as it describes the technical part of georeferencing and digitization, and also provides quantitative insight into historical land use changes in Lithuania which are quite impressive to look at. The paper is potentially interesting to readers of Land. However, prior to publication, the following issues should be fixed: 1) the framing of the paper (what is the contribution?) should be formalized; 2) the quantitative results should be expanded (i.e., a land use transition matrix would be very interesting), and a qualitative comparison to HYDE could be added; 3) if possible, the created historical data should be made available to the public. Thus, I recommend a major revision.

Paper title: it is not clear from the title, nor from reading the paper what is the actual contribution of this manuscript. In my opinion, there are two contributions (a) a georeferencing process for multiple map quadranges, and (b) quantitative insight into Lithuania’s historical LULC changes. Please ensure that the contributions of the paper are clearly laid out in title, abstract, and the paper itself.

Line 46,47: “ Nevertheless, there are many historical maps that are not available on paper but only digital;”  I don’t understand this sentence. Isn’t is the opposite? Histocal maps are available on paper, but not digitally?

Line 51: The next step, as noted by Podobnikar and Kokalj [11], is transformation of the scanned maps into the common coordinate system.  “into a common coordinate system”sounds better. However, I don’t think this is very accurate. Do you mean the georeferencing process here? I would recommend to be cartographically precise, and calling it “spatial reference system” (SRS) and not “coordinate system”. Even if maps are georeferenced but in different SRS, they still can be used for further processing.

The introduction should also state that there are automated methods for (a) georeferencing, and (b) digitization, please mention this, and possibly cite 1-2 examples for each.

Reference list should be expanded. Several important contributions are not cited. E.g., work from Dominik Kaim et al. should be cited, as well as important review papers on topographic map processing and information extraction (e.g. Leyk et al. 2014), as well as recent work on deep-learning based efforts from Hussaini & McDonough et al., Heitzler et al., as well as data fusion approaches, e.g., Uhl et al., should be cited.

Also, a sentence should be included in the introduction stating that there are several global/regioal LULC change models (e.g. HYDE). This will place your work in a broader context of data-driven historical geospatial analysis.

Line 61: I recommend to remove the “researcher skills” – this is obvious, any scientific work depends on researcher skills -this statement does not add anything to the paper.

Line 160: coordinate system  “spatial reference system”

Line 162: “The older the map, the greater the difficulties in georeferencing”  this is probably true in most of the cases, but please give a justification for this claim, e.g. the lower spatial accuracy due to the topographic measurement techniques used, paper distortions due to light / humidity, … You mention these issues correctly in lines 215, but a quick explanation here would be good.

 

Line 165: “Studies show that people working with old maps lack these skills and choose incorrect methods of georeferencing in their studies [2] “  you cannot write this. This is not scientific at all. It overly generalized and is certainly not true, and may be interpreted as offensive. There are many researchers that have these skills. Needs to be toned down considerably.

Line 187,188: I cannot make sense of this sentence. What do you mean by “we chose not to change” ??

Line 193: please provide a definition of LKS, and some descriptions – datum, ellipsoid, coordinate system?

Fig. 5 is quite impressive. However, Fig. 4 as a standalone figure is not very telling. I recommend to make Fig. 4 a full-page figure and put the contemporary land use map below, or even a third map, showing the areas of change, vs areas of stability.

Paragraph about forest change (lines 346-353): where does the information reported in [48] come from? Also from historical maps? I find it very interesting that forest cover has increased by 6%, but the paragraph does not provide an explanation for that. Please explain why you consider this plausible. Also, please elude on how inaccuracies in the digitized areas, and inconsistencies between CORINE and the historical extractions could affect the reported changes.

In line 325, the authors state: “In this study, we created a product not previously available to researchers.” – However, in the data availability section they state “data available upon request”. – What prevents you from publishing the data in a dedicated repository (zenodo, figshare, Harvard dataverse….)? I would urge the authors to publish the data. There are three possible framings for this paper: 1) Authors describe a novel method for information extraction from historical maps, 2) Authors focus on the historical land use changes, or 3) Authors describe the production of a dataset, and publish the dataset according to the FAIR principles. In my impression, this paper would be much more valuable (and cited!) if the data was available. Of course, I understand if you want to keep the data unpublished for future analyses, to avoid getting “scooped”. In that case, the data should be published at a later stage.

Results section: The trends in Table 3 are interesting, but the results section could easily be made even more interesting, if a full land use transition matrix would be provided, i.e., indicating the % area that converted from wetland to forest, from other to built-up, etc.

Lastly, some words on the quality of the outputs are needed. How accurate are the results? Why is it difficult to validate the results? What could be done easily, is a comparison to the historical land uses reported in HYDE. This could be done qualitatively, by visualizing the available corresponding LU classes from HYDE, next to the results obtained by the effort presented herein.

Line 433: “Since there is no general method for georeferencing historical maps,…”  I don’t understand this statement. The general method is manual georeferencing using GCPs created in a GIS, and using some affine transformation (or other) to warp the scanned image. The authors describe an interesting method for georeferencing, but they do not compare the method to other georeferencing methods. Thus, the framing of the paper should be adjusted. Is the purpose of the paper to provide a method for georeferencing, or is the purpose to highlight land use change in Lithuania? In my understanding it is the latter. Please clarify, and make the contribution of the paper clear in abstract, intro, and conclusion.

 

 

 

 

Author Response

Response for reviewer is given in the file. Please see the attachment.

Author Response File: Author Response.docx

Reviewer 4 Report

Accurate digital reconstruction of historical maps requires skills and high professionalism in the correct binding and interpretation of data on the individual spatial features. In addition, historical maps are valuable information for understanding the long-term processes of anthropogenic impact on landscape changes, evaluating the economic, social and ecological effects in changing landscape territorial structures. The authors well present the research methodology and features of the digital reconstruction of historical maps. The results of the interpretation of comparisons of the historical and modern chrono-slices in the change of landscape territorial structures are also well presented. Therefore, the presented article has a high technical and scientific value. Recommended for publication.

Author Response

Response for reviewer is given in the file. Please see the attachment.

Author Response File: Author Response.docx

Round 2

Reviewer 3 Report

All my comments have been addressed. However, there is an issue regarding the new Figs. 4 and 5. The authors chose a very complex way to illustrate a simple land use transition matrix. I would ask the authors to provide a simple transition matrix (or two, one for the relative % and one for the absolute areas), which can even be in the form of a table. The current figure is very difficult to interpret.

 

Regarding the new Fig. 4, the map (B) should be the same size as (A), and should only contain the changed areas. Showing both changed and unchanged areas makes it hard to see any patterns. Also, the legend for maps 4A and 4B should be separate, as the green color appears twice.

Author Response

Please check the attachment. 

Author Response File: Author Response.docx