Review Reports

Round 1

Reviewer 1 Report

General comments
The present manuscript is dealing with an environmental monitoring method based on thermal imaging to simply and rapidly identify litter pollution. For that purpose the authors installed air and soil surface temperature sensors to record temperature time series over three months. Furthermore, they acquired six thermal and RGB images at two days on a beach in Italy. From the abstract and the title I expected an interesting method in the field of intelligent image processing. Unfortunately, I was quite disappointed here. Apart from the very thin data situation, there are already numerous approaches using modern artificial intelligence techniques to assign structures like litter material in images. The few examples described here are not very convincing that the use of IR
cameras is justified.
I miss the scientific added value of the manuscript and therefore I recommend the rejection.

Specific comments
Page 2, line 51: There is only one sensor for air temperature (ok) and
also only one for soil temperature. I would recommend to add some soil
sensors and maybe also include some soil moisture sensors, too. So you
can derive a soil temperature distribution (e.g. transect) and may have
more information about soil temperature also in shady places.


Page 2, Fig. 1: There is no information on the polar plots in the  gure
caption, axis labels are missing, plot titles are not clear.

Page 4, Fig. 3: The beach part under consideration has a length less
than 100 m?


Page 5, lines 139 : I do not understand the sentence.


Page 6, lines 159: Here you mention, that also soil moisture may have
an influence. That's true and that's why I suggest also to include soil
moisture data to improve the prediction.


Page 8, Fig. 8: In Figure 8a it is clearly seen, that the described simple
method has it's problems with shaded areas, because they are significantly
cooler than the predicted soil temperature value. How large is the influence
by changes in emissivity (there are surely variations between dry and moist
sand as well as litter)? The thermal images are blurry and the litter is in
my opinion much better to see in the RGB images? So why to use the
thermal pics? What is the added value using your method?

Page 9/10, Tab. 3/4 I do not understand, how do you obtain the
detected items? Did you pick it manually at the randomly chosen points by
visual control of both picture types? That would make sense, if you want
to use the data as training data for a supervised learning algorithm. Your
problem seems to me easily solvable with modern computer vision algorithms and that would be an application I really miss in this manuscript / the added value of such an environmental monitoring method. However, much more data would than be required.

Comments for author File: Comments.pdf

Author Response

ANSWER TO REVIEWER 1

 

Thank you for reviewing our manuscript and for providing supportive comments. We appreciate the effort you made to improve this manuscript and are grateful for your insightful comments. Below you can find your comments and the lines where the suggested changes were made (in bold).

 

• Only one soil temperature sensor was installed because we limited the work in the area with the highest presence of beach litter. Furthermore, the constant anthropic presence on the beach even in the autumn months could damage/remove the distribution of the soil temperature sensors installed.

(Page 2, line 51: There is only one sensor for air temperature (ok) and also only one for soil temperature. I would recommend to add some soil sensors and maybe also include some soil moisture sensors, too. So you can derive a soil temperature distribution (e.g. transect) and may have more information about soil temperature also in shady places.)

 

• Polar plots information has been added to the figure caption, the chart title has been modified and the Figure 1 has been enlarged.

(Page 2, Fig. 1: There is no information on the polar plots in the figure caption, axis labels are missing, plot titles are not clear.)

 

• Yes, the length of the beach considered is less than 100 m. It is the area with the highest presence of beach litter.

(Page 4, Fig. 3: The beach part under consideration has a length less than 100 m?)

 

• The sentence has been rewritten and better explained (Lines 172-174).

(Page 5, lines 139: I do not understand the sentence.)

 

• We know that soil moisture could affect soil temperature just as air humidity could affect air temperature. Furthermore, the air humidity could be influenced by other atmospheric parameters (rain, wind, atmospheric pressure) and the sandy soil moisture by other parameters (vegetation, storm surges, rain fall). Nonetheless, since the thermal imaging camera acquires only the temperature data, we have standardized the experimental monitoring activity only to temperature data (air and soil). This was done to make the thermographic technique a methodology to support other new methodologies (spectral sensors, artificial intelligence algorithms) in emergency conditions of coastal pollution in which it is necessary to be quick in identifying and mapping.

 

(Page 6, lines 159: Here you mention, that also soil moisture may have an influence. That's true and that's why I suggest also to include soil moisture data to improve the prediction.)

 

 

• Considering the emissivity of dry sand and water, the emissivity difference between dry and wet sand could be about 3-5%.

This methodology was developed as an aid to other new methodologies (spectral sensors, artificial intelligence algorithms). In particular for the identification and rapid quantification of anthropogenic pollution along the coast. In this sense, this work would be useful for UAV surveillance. In fact, using UAV systems with larger scale surveys, the shaded areas would have less weight in the acquired thermograms than the really radiated beach areas. (inserted in the lines 283-287)

 

(Page 8, Fig. 8: In Figure 8a it is clearly seen, that the described simple method has it's problems with shaded areas, because they are significantly cooler than the predicted soil temperature value. How large is the influence by changes in emissivity (there are surely variations between dry and moist sand as well as litter)? The thermal images are blurry and the litter is in my opinion much better to see in the RGB images? So why to use the thermal pics? What is the added value using your method?)

 

• The points in the tables correspond to the total of random points (morning and afternoon acquisitions) created with the GIS software. The confusion matrix was then created by visually checking the RGB images.

 

(Page 9/10, Tab. 3/4 I do not understand, how do you obtain the detected items? Did you pick it manually at the randomly chosen points by visual control of both picture types? That would make sense, if you want
to use the data as training data for a supervised learning algorithm. Your problem seems to me easily solvable with modern computer vision algorithms and that would be an application I really miss in this manuscript / the added value of such an environmental monitoring method. However, much more data would than be required.)

Reviewer 2 Report

The authors propose a manuscript titled “Autumnal Beach Litter Identification Using Ground-based IR Thermography in the Coastal Dunes Park (Apulia Region, Italy)”

The article is well structured with original data. However it is deficient in some aspects which can be solved based by personal suggestions. For example, there is no map studied area, no specific references in studied area. I appreciate the original idea of the work which with the following revisions will convince me and the editor to publish it on Journal.

Introduction

Row 41-43. This point is incomplete and incorrect. The authors declare: “In the environmental monitoring field, thermography had a remarkable development, especially in wildlife census [5] and wildfires detection [6]. The flora monitoring cannot be detected with this system, because not identify for example all plants species, in particular the small ones. So change the period in this way: “In the environmental monitoring field, thermography had a remarkable development, especially in wildfires detection [6], while is only a first approach for flora and vegetation habitats montoring, and sometimes also for wildfauna, which necessarily require observations and confirmations directly in the field”.

Rows 79-83. Please pay attention for this sentence: “The whole coastal sector is exposed to a wind regime characterized by winds coming from the north-western quadrants (Figure 1) therefore storm surges are very frequent in  winter, contributing to define a strong erosion of the dune belt, and also causing the stranding of a large amount of natural and anthropogenic material”.

NOTE: The erosion of system erosion and the relative vegetation are not caused by the wind but are mainly due to manwork with the sand leveling in the summer season, especially in the investigated site and in general on the whole psamophilous coastal area that extends from the site study and far as Brindisi. The work completely lacks the expertise of an ecologist or botanist, as the wind from the north-west has always been there for decades, and in nearby sites there is a well-preserved psammophilous vegetation with all dune vegetation series, with the embryonic dunes “Cakiletum maritimae”, which is the first type of vegetation starting from the shoreline. The authors know that there is a dune series with habitats protected by the Directive 92/43 EEC? There is no mention of this in the text. The object of the work speaks of coastal dune for which it is necessary to give two words on protected habitat, in order to give a complete picture at the reader.

The manuscript is poor in reference. Here I attached 3 references concerning the natural habitat and biodiversity in the studied area. Please add at least these references in the introduction or in materials and methods that you will have to reformulate.

• Ciola, G.; Maiorano, F.; Massari, M.A. Il Parco Naturale Regionale delle Dune Costiere da Torre Canne a Torre San Leonardo: il valore della biodiversità per ricostruire comunità solidali. Territori e comunità. In: Gisotti, M.R.; Rossi, M. Le sfide dell'autogoverno comunitario. Atti dei Laboratori del VI Convegno della Società dei Territorialisti Castel del Monte (BA), 15-17 November 2018. Sdt Edizioni, Bari, Italy.
• Perrino, E.V.; Ladisa, G.; Calabrese, G. Flora and plant genetic resources of ancient olive groves of Apulia (southern Italy). Resour. Crop Evol. 2014, 61, 23-53. Doi: 10.1007/s10722-013-0013-1
• Tomaselli, V.; Tenerelli, P.; Sciandrello, S. Mapping and quantifying habitat fragmentation in small coastal areas: a case study of three protected wetlands in Apulia (Italy). Monit. Assess. 2012, 184, 693–713 Doi: 10.1007/s10661-011-1995-9

Materials and Methods

Figure 3. In the red grid grows, without anthropic disturbance (therefore sand leveling by mechanical means) of the Cakiletum vegetation. Please consider this crucial environmental aspect before your ratings.

Figure 4.

Do the authors know the vegetation where they installed the temperature sensors? I know it (Ammophiletum or Agropyretum?) and it would be appropriate at least to say which type it is, even without botanical comments.

These habitats grow exactly in the perimetred area, starting from the coastline (http://vnr.unipg.it/habitat/cerca.do?formato=stampa&idSegnalazione=8)

• Annual vegetation of drift lines (Habitat code 1210)
• Embryonic shifting dunes (Habitat code 2110)
• Shifting dunes along the shoreline with Ammophila arenaria (white dunes) (Habitat code 2120)

Rows 138-146. I apprecciated the descripton about the random points were created with a GIS software for every processed thermogram, and the calculation throught the kappa coefficient on a morning and afternoon scale. Well done.

3. Results and discussion

The tables and figures are clear.

Table 1. I noticed that there are temperature differences unevenly distributed between Air and Sandy-soil surface. The differences in Maximum daily average temperature (° C) is 1.3 ° C, while for the Minimum daily average temperature is 2.3 ° C. Please explain better these question or try to give an explanation.

Conclusion

Well done, but please two words of study perspective

Author Response

ANSWER TO REVIEWER 2

 

Thank you for reviewing our manuscript and for providing supportive comments. We appreciate the effort you made to improve this manuscript and are grateful for your insightful comments. Below you can find your comments and the lines where the suggested changes were made (in bold).

 

 

• Period changed (Lines 45-49)

(Row 41-43. This point is incomplete and incorrect. The authors declare: “In the environmental monitoring field, thermography had a remarkable development, especially in wildlife census [5] and wildfires detection [6]. The flora monitoring cannot be detected with this system, because not identify for example all plants species, in particular the small ones. So change the period in this way: “In the environmental monitoring field, thermography had a remarkable development, especially in wildfires detection [6], while is only a first approach for flora and vegetation habitats montoring, and sometimes also for wildfauna, which necessarily require observations and confirmations directly in the field”.)

 

• I added the references and reformulated the introduction (Lines 75-85).

(Rows 79-83. Please pay attention for this sentence: “The whole coastal sector is exposed to a wind regime characterized by winds coming from the north-western quadrants (Figure 1) therefore storm surges are very frequent in  winter, contributing to define a strong erosion of the dune belt, and also causing the stranding of a large amount of natural and anthropogenic material”.

NOTE: The erosion of system erosion and the relative vegetation are not caused by the wind but are mainly due to manwork with the sand leveling in the summer season, especially in the investigated site and in general on the whole psamophilous coastal area that extends from the site study and far as Brindisi. The work completely lacks the expertise of an ecologist or botanist, as the wind from the north-west has always been there for decades, and in nearby sites there is a well-preserved psammophilous vegetation with all dune vegetation series, with the embryonic dunes “Cakiletum maritimae”, which is the first type of vegetation starting from the shoreline. The authors know that there is a dune series with habitats protected by the Directive 92/43 EEC? There is no mention of this in the text. The object of the work speaks of coastal dune for which it is necessary to give two words on protected habitat, in order to give a complete picture at the reader.

The manuscript is poor in reference. Here I attached 3 references concerning the natural habitat and biodiversity in the studied area. Please add at least these references in the introduction or in materials and methods that you will have to reformulate.)

• Ciola, G.; Maiorano, F.; Massari, M.A. Il Parco Naturale Regionale delle Dune Costiere da Torre Canne a Torre San Leonardo: il valore della biodiversità per ricostruire comunità solidali. Territori e comunità. In: Gisotti, M.R.; Rossi, M. Le sfide dell'autogoverno comunitario. Atti dei Laboratori del VI Convegno della Società dei Territorialisti Castel del Monte (BA), 15-17 November 2018. Sdt Edizioni, Bari, Italy.
• Perrino, E.V.; Ladisa, G.; Calabrese, G. Flora and plant genetic resources of ancient olive groves of Apulia (southern Italy).  Crop Evol.2014, 61, 23-53. Doi: 10.1007/s10722-013-0013-1
• Tomaselli, V.; Tenerelli, P.; Sciandrello, S. Mapping and quantifying habitat fragmentation in small coastal areas: a case study of three protected wetlands in Apulia (Italy).  Assess.2012, 184, 693–713 Doi: 10.1007/s10661-011-1995-9

 

• I added the presence of the vegetative species (suggested) in the caption of figure 4

Figure 3. In the red grid grows, without anthropic disturbance (therefore sand leveling by mechanical means) of the Cakiletum vegetation. Please consider this crucial environmental aspect before your ratings.

Figure 4.

Do the authors know the vegetation where they installed the temperature sensors? I know it (Ammophiletum or Agropyretum?) and it would be appropriate at least to say which type it is, even without botanical comments.

These habitats grow exactly in the perimetred area, starting from the coastline (http://vnr.unipg.it/habitat/cerca.do?formato=stampa&idSegnalazione=8)

• Annual vegetation of drift lines (Habitat code 1210)
• Embryonic shifting dunes (Habitat code 2110)
• Shifting dunes along the shoreline with Ammophila arenaria (white dunes) (Habitat code 2120)

 

 

• Considering Tab. 1, from the average daily data of the air temperature and the sandy soil temperature of the monitored period, it is possible to see a greater thermal excursion in the sandy soil than in the air. This can be explained by a greater capacity of dispersion of thermal energy by the sandy soil in the coldest hours of the day (usually at night).

Table 1. I noticed that there are temperature differences unevenly distributed between Air and Sandy-soil surface. The differences in Maximum daily average temperature (° C) is 1.3 ° C, while for the Minimum daily average temperature is 2.3 ° C. Please explain better these question or try to give an explanation.

• Study perspective added (lines 306-308)

(Conclusion: Well done, but please two words of study perspective)

Round 2

Reviewer 1 Report

I still miss the methodical novelty of the of the presented work and would like to recommend for further work in this research area to strengthen the data base and to apply recent computer vision algorithms for analysing image data. However, the authors provided some more insights with their new version and highlighted the need for such investigations in coastal environments. Therefore, I would like to encourage the authors to continue their efforts in future and to use mobile sensor technology (e.g., drones) and modern algorithms to expand their studies.

Author Response

Thanks again for the suggestions for future research work. Surely in the next months we will plan thermal and multispetrical surveys from UAV in the study area. We are just waiting to receive the drone with the sensors and the authorization to fly in the study area.

Reviewer 2 Report

I congratulate the authors for their efforts to deep revise the manuscript based on the suggestions given. I have no further changes to request. Only one mistake at line 350 in reference where the Authors must be add Genet. in this way: Genet. Resour. Crop Evol.

Author Response

The mistake has been corrected. Word “Genet.” added. Thanks again for the suggestions.