Mapping of Urban Vegetation with High-Resolution Remote Sensing: A Review

Round 1

Reviewer 1 Report

Review report for “Mapping of urban vegetation with high-resolution remote sensing: a review”

 

The manuscript seems not well organized, and the contents are a little messy. After reading the whole manuscript, I did not feel a clear understanding about the status of urban vegetation mapping.

 

The manuscript contents and the topic seem inconsistent. The title says “mapping of urban vegetation with high-resolution remote sensing”. However, the major contents of the manuscript are talking about tree classification. Is urban vegetation just trees? How about other urban vegetation such as lawns and gardens, and the mixture of various plants as parks and gardens?

 

While tree classification and mapping are important, what are the uniques of urban trees compared with non-urban trees? If they are similar, does the mapping of urban trees have any differences from the mapping of non-urban trees? What is the value to review the mapping of urban trees rather than simply trees? Just because the study areas of the reviewed articles are urban areas? Almost the whole manuscript is talking about tree mapping. It seems that the term “urban” does not hold any important meaning here. Such a review does not have extra value compared with a review purely for tree mapping.

 

I think what is truly important for urban area is Urban Green Space, because urban green spaces are important to urban residents and visitors and they provide the ecological services to urban population and activities. Urban green spaces may include urban forests, parks, lawns, gardens, and street greenery such as trees and bushes. Purely tree classification and mapping from high resolution RS imagery is another topic, which is not unique to Urban.  

 

The authors probably should focus on the current general situation of urban green mapping, what problems have been solved, and what problems remain, and finally what further studies should be done to tackle the remaining issues in the future. What are the major issues with current urban green mapping and how they are coped with? What progresses have been made?

 

In addition, the reference selection seems not well designed. Why only consider publications after 2000? Wasn’t there such kind of research before 2000? If there was, the publications should be important because they are the pioneer studies that led to the recent situation in the study field. “Terrestrial and laser scanning” may not be a proper keyword to use for searching for the related articles. It may largely limit the search scope of articles related with urban green mapping. On the contrary, “urban green space” probably should be included. “Google street view” should be very useful for finding studies in street greenery classification and mapping.

 

1. In Introduction, the authors talked about urban green/green space. However, in the “Results” section, almost all contents are about tree classification.
2. The publication search scope seems problematic. The article collection is too concentrated on tree mapping, and ignores urban green space mapping.
3. In “3.1.1. Functional vegetation types” subsection, the authors only introduced a few of specific study cases, not the general status and trend. That is not what a review article should do.  
4. The following contents are about tree mapping, completely deviated from the article topic.
5. “3.2.4. Importance of phenology in vegetation mapping”. This is a common issue to all land cover land use mapping, not unique to urban green.
6. If urban green has no unique features, it has no difference from non-urban green. Then the review should not be limited to urban area.
7. Method review also did not particularly targets urban area and urban greenery.

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 2 Report

This article consists of a good bibliographic review on the works of urban vegetation: "The objective of this review is to give an overview of the different  approaches used by scholars to map and classify vegetation in an urban environment at a high level of detail. The paper is structured according to the main decisions that need to be made throughout the mapping process: the choice of a suitable vegetation typology, the remote sensing data to be used, and the mapping approach to be applied".  In this sense, it would be useful to include works of C. Weber, C. Petropoulou, J. Hirsch, A. C. Broneur and others. 

It is also important to distinguish between these works that deal only with public green (works that include issues of accessibility) and works that deal with the differentiation of areas of the city in relation to the presence / absence of green or its quality.

Finally, in addition to the literature review, it would be good to tested and apply different approaches in an area with the different methods and resolutions proposed.

Author Response

Please see attachment

Author Response File: Author Response.pdf

Reviewer 3 Report

The authors present a review focused on the mapping of urban vegetation using high resolution remote sensing data.

The authors clearly define their objective, methodology and material they used to perform their review.  Thy provide details for the three different aspects related to the mapping of urban vegetation that they examine. Their discussion provides information about the existing bottlenecks and achievements during the last years. Furthermore, their conclusions are supported by the analysis they performed.

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 4 Report

Mapping of urban vegetation with high-resolution remote sensing: a review

Most larger cities already monitor vegetation through extensive field surveys, however this only provides information concerning the public green space. Monitoring urban vegetation is also costly and time consuming, hence the increasing interest in automated mapping techniques.

Nowadays, large research already exists concerning the mapping of tree species and crop types in a rural environment. Nonetheless, research on mapping of urban green has its own challenges that are related to the spatial and spectral heterogeneity of the urban landscape and the complex three-dimensional structure of urban areas resulting in large shadowed areas, multiple scattering and issues of geometric mismatch in combining different data sources.

In this article, the authors conduct an overview of the main approaches proposed to map urban vegetation from high resolution remotely sensed data. Previous studies are reviewed from three perspectives: the vegetation typology, remote sensing data used, and the mapping approach applied. The interest of this study is to facilitate sustainable urban planning, being possible to establish a detailed inventory of urban green to adequately manage and to understand the ecological services rendered by vegetation.

The work conducted is rigorous, exhaustive, and well structured. Thus I can recommend for publication just after few suggestions are considered. These are the next ones:

• The authors argue in the introductory section: “With regards to vegetation typology, a distinction is made between studies focusing on the mapping of functional vegetation types and studies performing mapping of lower-level taxonomic ranks, with the latter mainly focusing on urban trees. Researchers for both types of mapping have used a wide variety of high-resolution imagery. The fusion of various types of remote sensing data as well as the inclusion of phenological information through the use of multi-temporal imagery prove to be the most promising avenues to improve mapping accuracy.” They focus their approach here, but this maybe must be combined with the use of LiDAR systems (ALS) which allow achieving accurate (geometric) data. It is important related with the argument shown just few lines after that: “Interest in mapping of non-tree species in urban environments is still limited. The same holds for mapping of understory species. Most studies focus on mapping of public green spaces while interest in the mapping of private green is less common”. Just some few lines later, they argue “However, traditionally land cover mapping in an urban context often concerned only two vegetation classes: high vegetation and low vegetation [19,20]”. Some relevant studies work in the combinatory interface between the use of aerial imagery and LiDAR. For example, Hermosilla et al. (2014) introduced a methodology for estimating street based urban metrics that allowed discriminating complex urban typologies. In their method, they differentiated the height and volume of green spaces within urban areas. The same author had applied before (2012) a study for land-use Mapping from Aerial Images and LiDAR Data in the city of Valencia (Spain) achieving a differentiation of around 20 land uses/covers in a consolidated urban environment and including more than 5 related to vegetation (such citrus orchads, arable lands, garden/parks, forest, irrigated crops, carob-tree orchads, rice fields). I was reviewing the sub-section “Fusion of LiDAR data and spectral imagery” and I missed references to findings related to the previously presented studies. I can recommend considering these studies and some other similar ones.
• In line 447, the authors refers to “terrestrial sensors”, which gather information through a sensor mounted on a moving vehicle, usually an automotive system (…). Data captured by terrestrial spectral sensors is gaining popularity for the mapping of road side vegetation.” Do the authors include in these systems the so-called Mobile Laser Systems (MLS)?
• In line 465, the authors state “Terrestrial laser scanning is another type of data acquisition used for vegetation map ping. Puttonen et al. [94] and Chen et al. [95] found this type of data useful for the mapping of tree species, however a higher accuracy may be obtained when this type of data is merged with higher resolution spectral data [94].” I think could be quite convenient for the authors refer to the combinative use of laser scanning data (TLS) and close-range photogrammetry data? What I am missing is to introduce briefly why they must be combined (gaps in laser systems, low radiometric quality, limitations for data capturing from the laser system, etc). For this, close-range photogrammetry can supplement laser scanning point clouds. Just after that, the authors argue, “In both studies, the point clouds associated with a vegetation object were manually delineated. The segmentation of various objects from terrestrial point clouds remains a significant challenge on par with the actual classification of the clouds due to the large volume of data and the irregularities in the point cloud caused by the complexity of the urban environment [95].” Basically, the method implemented is the usual one. The successive point clouds obtained from laser scanning and photogrammetry are generated and merged (This process is called registration). After that, the most relevant features are traced by using a CAD modelling process (This is called segmentation), which allows to thin the dataset and to reduce the computational costs. This methodology is common applied in digital preservation, where the reconstruction of vegetation features can be part of the context. You must check and include some relevant studies related to Generation of 3D models by using laser scanning and digital photogrammetry. The authors refer to the importance of the distribution of laser points for extracting the structure of the crown of trees, and also to the intensity of the laser beam (ref. Liu et al [59]). I agree, but what is the importance of the laser density in the results? I would suggest to add half paragraph related to this and the importance of this factor in laser systems.

 

Two more recommendations/suggestions:

• Due to this article comprises a large and exhaustive literature review, the authors tend to summarize/classify/organize these in lists. My recommendation is pretty simple: When a list of different options is shown, the authors should use a numbered list, which would help to the readers to the mentally organize the structure. For example, in the abstract the authors argue “Studies are reviewed from three perspectives: the vegetation typology, remote sensing data used, and the mapping approach applied”. I would suggest using “Previous studies are reviewed from three perspectives: (a) the vegetation typology, (b) remote sensing data used, and (c) the mapping approach applied.”
• After I read the paper, I was the feeling that the discussion section was too large and repetitive with the ideas, arguments and studies shown along the paper. I would suggest to reduce it and to summarize more the discussion.

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

No comments