Feasibility Study Using UAV Aerial Photogrammetry for a Boundary Verification Survey of a Digitalized Cadastral Area in an Urban City of Taiwan

Round 1

Reviewer 1 Report

The paper suggests that UAV imaging can be used to support ground-based surveying of parcel boundaries in urban areas. A standard photogrammetric procedure to capture the images and to generate a map from them is followed. The map from UAV is registered with a cadastral map. A few corner points of buildings or road markings are measured from the UAV imagery and used as supplementary control points in ground-based measurement of parcel corner points with a total station. The approach reduces labor costs in future parcel boundary surveys within the coverage of the UAV imagery, since it is cheaper to measure control points from the UAV imagery already acquired than surveying each local area with ground-based techniques.

The structure of the paper is logical and the paper is mostly written with sufficient detail. It is positive that each step of the photogrammetric procedure is verified with reference data. The problem is that there is nothing scientifically new in the paper. All the methods used are well-known while no new methods or algorithms are presented. The paper can be considered to describe work in progress while it would be interesting see if all the parcel boundary lines could be extracted from the UAV imagery or from a dense DSM produced from the imagery and compared against the cadastral map.

Detailed comments:

Line 13: It is not clear what do you mean by "overlap analysis". Two maps overlap? A point overlaps with a map, i.e., the point is within the coverage of the map?
Line 24: It is well-known that supplementary control points can be measured from UAV aerial images.
Lines 104-106: It is confusing that "registration" will increase "the reliability and accuracy of registration", or do you mean that "overlap analysis" will increase "the reliability and accuracy of registration"? Or do you mean that the maps match better after registration than before registration? Registration usually means solving a coordinate transformation between two data sets. It seems that you are using "map registration" to mean something else?
Line 107: It is obvious that "aerial photogrammetry can preserve aerial images".
Lines 173-174: Do you mean the differences between the two observations which were averaged?
Line 182: What are observations that are not observations?
Lines 223-225: The sentence "UAV aerial photogrammetry ..." is not clear.
Section 2.2.1 describes what is done before registration, but the explanation of how the registration is done is missing. What features are matched in ArcMap? Points? Lines?
Line 269: Why are the distances between the detail points and cadastral lines computed before registration? Shouldn't they be computed after registration?
Line 275: Are you really staking out points in the real world? Why don't you compute the differences using the map and images? Or can't you measure the locations of parcel corner points from the imagery?
Line 307: Difference between which?
Line 357: Is it necessary to resample the images? Why don't you use the original images and calculated orientation parameters in space intersection? Does Pix4DMapper require that images are rectified to the normal case of stereo photogrammetry?
Lines 414-417: If the distance is less than 0.3 mm, then the supplementary control point and parcel points are very close to each other. This contradicts what is seen in Fig. 18. Do you mean that the difference between the surveyed and computed distance should be less than 0.3 mm in map coordinates and less than 0.15 m in object coordinates?
Table 5: Are the "computed" figures those which have been calculated according to the map?
Lines 428-429: The sentence "The locations after..." is not clear.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

The research focuses on the feasibility of UAV data for cadastral land surveying and boundary verification.

The study merges spatial data from both remote sensing techniques (UAV orthophoto) as well as ground survey equipment (i.e. Total station, Survey GPS) and ground control points. The authors conclude that the UAV aerial photogrammetry is consistent with current methods but includes much more detailed information than traditional surveying.

The manuscript is written in easy to follow manner, very neatly and in correct English with very few grammatical errors.

Language:

One to improve is on line 207 'measured'

470 something missing [, ; or ; and]

What would be very nice to see is the discussion how the results would hold if the UAV flight altitude was below 150m AGL (sometimes 120 m AGL) as this is the common limit for UAV survey in many countries around the World (due to the fact that the general aviation uses airspace higher than 150 m AGL). In such cases, the angles are different and shading of roofs, walls might play a role.

Can you also comment why most the control points and check points (i.e. figure 8) are chosen so close to each other? If they were spaced by larger distance then the results might have been worse?

If possible, please elaborate more on y parallax (244 - 258) possibly add some graphic / more explanation. I rarely see a paragraph with explanations and it is often omitted topic in the available documentation for UAV surveyors.

Author Response

Please see the attachment.

Reviewer 3 Report

The paper that authors propose turns out to be interesting from a technical application point of view. The proposed method of using a photogrammetric proximity survey to improve result of digitalization of cadastral maps is innovative and important. With this methodology it is possible to deduce many more control points easily and at relatively low costs, with accuracy necessary to be able to rectify the digitization of analogic maps that could have undergone deformations over time.

However, the paper presents some gaps and inaccuracies.

Need to go into more detail when:

- describes the choice of control points and the calculation method chosen for determining the accuracy of the results of the e-GNSS surveys (least-squares collocation adjustment)

- talks about the definition of the parameters in the AT and the use of the two software. It is necessary to motivate the proposed choices

- Specify and deepen the part on the generation of digitized map, on definition of control points and on final verification methods.

In all figures and tables the captions must be written in a broader and more exhaustive way, so that each individual image can be understood independently.

The bibliography must be increased, especially as regards the methods for defining the parameters due to AT.

Details:

Line 159 to 160: Motivate this choice

Line 163 to 165: Specify which colors have been used and the shape of the targets. Better to insert a photo. With 3cm of GSD, a target of 20cm is still at the limit. It would be better to use a double size target with a circular shape inside

Line 168: specify the meaning of e-GNSS

Line 172: Perhaps he meant HIGHER

Line 175 to 182: Specify better which methodology has been used for survey of control points and to which dataset least squares collocation method has been applied. It would be clearer if a table with coordinates of points in two reference systems and the errors resulting from them were inserted

Line 187 – 188: Briefly state what these procedures are.

Line 194 – 198: It would be good to insert a table with the results of the two approaches. Also report the parameters that have been obtained and justify and highlight those that are chosen.

Line 204 – 206: Insert a table with the parameters used

Line 207 – 210: Explain better. Insert a table with parameters determined by ORIMA

Line 210 – 212: It is not clear whether the same control points were directly measured also with topographic instruments or not. Also enter here a table with values for comparison

Line 216: Figure 4: Insert the image with the writing of the software in English or put the calls with the writing in English.

Line 227: it is better to use the term "point consistency" instead of "local consistency"

Line 230 – 233: It is not clear from where derive the 3D cadastral line. Explain better.

Line 235 – 243: It would be useful to insert some sample images of some typologies of remarkable points that have been listed

Line 247 – 252: Report a table with parallax values and highlight the best ones and report which AT parameters correspond. Comment on why other stereoscopic models may have high parallax.

Line 252: Bibliography references of PRO600 software for MicroStation

Line 264 – 267: Define the reason for the choice of affine transformation and specify the type of affine transformation: uniform or different scale factor for each axis?

Line 275: Maybe you meant to write “world”

Line 300: What is the difference between red and blue marker? Write it here too

Line 306 – 310: It is necessary to report more detailed specifications for the collocation last squares method, inserting the scheme of agorithms used and inserting results of the coordinates before and after the transformation on the 5 control points. Comment more widely and results.

Line 321: Increasing the caption so that the table is read independently

Table 2: Enter "Horizontal distance (m) with Total Station"

Line 351 – 353: It would be good to report the same diagram also for the other 5 models. However, better explain why you wanted to show the parallax errors in y for each single point of model 2 only.

Line 356 – 359: Figure 4 shows the overall ortho-image and not the individual undistorted images. Specify the method of creating the ortho-image.

Line 378 – 380: It is interesting to see a map showing all the points divided into the 4 error classes, to see if there can be areas where the error is greater and areas where the error is less and understand why.

Line 381-382: Explain this statement. Does it depend on the graphic error?

Figure 13, 14: Increase the specifications in the caption: the image must be able to understand independently. What are the lines? where do they come from? What are orange dots?

Line 404: The verification survey was performed with a total station using two methodologies: free station and radiation method. Specify the reason for this choice and comment on what is the difference in the results of the survey.

Line 425 – 427: These points, as shown in Figure 18b, are points 5 and 6. They do not appear in Table 5 and therefore the data cannot be compared.

Table 5: points 5 and 6 do NOT appear in the table. The table is not clear when reading. Specify better what Computed Distance and length means: on the registered map, on the original map, other?

Figure 19: In the images it is not clear what the single "real" points are. Indicate them with a small circle.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

The authors have responded to the comments of the reviewer well enough. The English language is mainly fine although proofreading is recommended.

Author Response

Thanks for the reviewer’s comments

Reviewer 3 Report

The paper has improved in its drafting and those issues that were initially unclear were specified.

From my point of view, the current version of paper can be published. It is only necessary to format Tables 3 and 4.

Author Response

Tables 3 and 4 were named as Tables 4 and 5.

They were revised according to the journal´s style.