ArticlePDF Available

Remote sensing machine learning algorithms in environmental stress detection: Case study of Pan-European south section of Corridor 10 in Serbia

Authors:
  • Military Geographical Institute
UNIVERSITY THOUGHT doi: doi:10.5937/univtho7-14957
Publication in Natural Sciences, Vol. 7, No. 2, 2017. Original Scientific Paper
GEOGRAPHY
REMOTE SENSING MACHINE LEARNING ALGORITHMS IN
ENVIRONMENTAL STRESS DETECTION - CASE STUDY OF
PAN-EUROPEAN SOUTH SECTION OF CORRIDOR 10 IN SERBIA
IVAN POTIĆ1*, MILICA POTIĆ2
1Faculty of Geography, University of Belgrade, Belgrade, Serbia
3Independent researcher, Belgrade, Serbia
ABSTRACT
The construction of the Pan-European Corridor 10 is one of the major projects in the Republic of Serbia, and it
enters the final phase. A vast natural area suffered a significant change to complete the project and therefore is the
existence of a need to monitor those changes. Nature requires adequate and accurate detection of environmental
stresses which inevitably arise after implementation of such large construction projects. Conversely to traditional
field monitoring of the environment, this paper will present the remote sensing method which includes usage of
European Space Agency's Sentinel 2A optical satellite data processed with different Machine Learning algorithms.
An accuracy assessment is performed on land cover map results, and change detection carried out with best
resulting data.
Keywords: Environment Monitoring, Gaussian Mixture Model, Random Forest, K-Nearest Neighbors, Confusion
Matrix.
INTRODUCTION
One of the major projects of the Republic of Serbia funded
by the World Bank (WB), European Investments Bank (EIB),
Hellenic Plan for the Economic Reconstruction of the Balkans
(HiPERB) and the Republic of Serbia, is the construction of the
main branch of Pan-European Corridor 10. The corridor connects
Salzburg in Austria and Thessaloniki in Greece through
Ljubljana in Slovenia, Zagreb in Croatia, Belgrade, and Niš in
Serbia, Skopje, and Veles in Macedonia (Figure 1). In Serbia, the
south part of Corridor 10 is called the “Highway Е75 – project
SOUTH” and it is presented and constructed as the motor road at
this point (Koridori Srbije, 2017).
Figure 1. Pan-European corridors in Serbia.
Source: belgradenet.com
* Corresponding author: ipotic@gmail.com
The Highway Е75 – project SOUTH extends for 74 km,
from Grabovnica to Levosoje (Figure 2). There are five sections
to complete in this area: Grabovnica Grdelica (L=5.6 km),
Grdelica Caričina Dolina (L= 11.8 km), Caričina Dolina –
Vladičin Han (L= 14.3 km), Vladičin Han – Donji Neradovac
(L= 26.3 km), and Donji Neradovac - Levosoje (L= 16 km)
(Figure 2) (Koridori Srbije, 2017).
Figure 2. Corridor 10 South project in Serbia.
Source: www.koridor10.rs printscreen
The construction zone of this scale indubitably has a
significant impact on the environment. A proper monitoring is
crucial to conserve the nature and mitigate the environmental
stress. Considering that technology has advanced, we are going
to use the achievements of remote sensing and its methods to
monitor the changes that have occurred during the construction
of Corridor 10. Further, the change detection of the land cover
GEOGRAPHY
will be performed to present the changes for the monitored
period. Area of interest is selected within the area that is under
active construction and covers 1.095,4 sq. km (Figure 3).
Figure 3. Area of Interest - part of Corridor 10 working zone.
EXPERIMENTAL
Materials and methods
Remote sensing technology is employed to achieve the goal
of this paper with the contemporary methodology that employs
the Machine Learning (ML) algorithms (Canziani et al., 2008;
Mas & Flores 2008; Jensen et al., 2009; Duro et al., 2012; Lary
et al., 2016).
Sentinel 2 satellite imagery was obtained using Copernicus
Sci Hub (Copernicus Open Access Hub, 2017) as starting data
for the analysis. Sentinel 2 product consists of the granules that
represent the particular region. The granule comes with 13
different bands where three different ground resolution bands are
present: 10 m, 20 m, and 60 m. 10 m bands are: visible Blue (B),
Green (G), Red (R), and Near InfraRed (NIR). 20 m bands are
three Vegetation Red Edge bands, Narrow NIR and two Short
Wave InfraRed (SWIR) bands. 60 m bands are Coastal Aerosol,
Water, Vapour and SWIR Cirrus band (Sentinel 2 MSI, 2017).
Two different Sentinel 2 products Level-2A were
downloaded for 2017. Since there were cloudy parts in the
research area, the mosaic was made using two different granules
T34TEN date from 01.07. 31.07.2017. Remote sensing/ raster
processing plugin for QGis was applied to perform the
mosaicking tasks.
To perform the change detection for the research area, the
same images from August 2016 were downloaded from the
Copernicus Sci Hub, and sub-scene created. The image was
cloud-free, and there was no need for mosaicking. The product
was Level-1C, so the data was processed to Level-2A using
SNAP (Sentinel Application Platform) toolbox software (ESA
STEP, 2017), which took more than 13 hours to complete.
Sentinel 2 products have multiple processing phases:
- Level-0 and Level-1A&B products are in preprocessing
phase and not available to users;
- Level-1C processing uses the Level-1B product and
applies radiometric and geometric corrections
(including orthorectification and spatial registration);
- Atmospheric correction is applied to Top-Of-
Atmosphere (TOA) Level-1C orthoimage products, and
a scene classification is presented as the Level-2A
product. Bottom-Of-Atmosphere (BOA) corrected
reflectance product is Level-2A with main output as an
orthoimage. Additional outputs are Aerosol Optical
Thickness (AOT) map, a Water Vapour (WV) map and
a Scene Classification Map (SCM) together with
Quality Indicators (QI) for cloud and snow
probabilities at 60 m resolution (Sentinel 2 MSI, 2017).
Sentinel 2 bands used to complete the analysis are Red,
Green, Blue and Near Infra-Red bands with 10m ground
resolution.
Pixel-based Machine Learning (ML) algorithms were used
to produce the land cover map of the area of interest. The most
common three ML tasks are Regression, Classification, and
Clustering.
Regression is employed as supervised learning task for
modeling and predicting variables, where we have numeric true
ground values for the research area. There are different
regression algorithms, such as:
- Linear Regression (works when there are linear
relationships between dataset variables);
- Regression Tree or Decision Trees repeatedly splits the
dataset into separate branches and maximize the
information gain. This allows the algorithm to learn
nonlinear relationships;
- Deep Learning algorithm applies to multi-layer neural
networks to learn extremely complex patterns using
convulsions and drop-out mechanisms, and others;
- Honorable Mention (Nearest Neighbors) save each
training observation. Further, they make predictions for
new observations as they search for similar training
observations and join the values (Elite Data Science,
2017).
Classification, as supervised learning task, is used in this
paper to model and predict land cover categories as the ML
algorithms can predict a class. Different classifications were used
in this article to obtain the best possible accuracy of the data:
- Classification Trees is employed in Random Forest;
- Gaussian Mixture Model (GMM) take on that data
points are generated from a mixture of a limited
number of Gaussian distributions with unfamiliar
parameters (Scikit learn, 2014).
GEOGRAPHY
K-Neighbors Classifier where the learning is based on the k
nearest neighbors of each query point. k is an integer value
specified by the user (Scikit learn, 2014).
The creation of a land cover map from BOA processed
Sentinel 2 data required a ground training samples. To obtain
such areas and create necessary vector file as training material,
historical google maps were employed using different sources
and plugins for QGis. Seven different classes recognized for both
2016 and 2017 and consist of 175 and 164 polygons respectively.
Two attributes created, as integer and text. Further, prepared
subscene for each year was processed using dzetsaka ML plugin
for QGis.
The accuracy assessment was performed using training
sample polygons in dzetsaka and SCP plugin for QGis.
Confusion matrix was created and presents overall accuracy and
kappa hat.
The land cover change was performed using SCP plugin in
QGis.
NUMERICAL RESULTS
After applying the algorithms, three different land cover
maps for each year were created (Figure 4).
Accuracy assessment for created land cover maps is
presented in Tables 1-3. As it can be seen, ML algorithms gave
very decent results where Random Forests goes up to 100% of
accuracy.
Figure 4. Land cover maps for the area of interest created using different ML algorithms.
GEOGRAPHY
Table 1. Confusion matrix for K-Neighbors Classifier.
Table 2. Confusion matrix for Gaussian Mixture Model Classifier.
Table 3. Confusion matrix for Random Forest Classifier.
Class 2016
Forest
Artificial
bare soil
Bare
soil
Artificial
area
Pastures
Agriculture
Class 2017
Forest
Artificial
bare soil
Bare
soil
Artificial
area
Pastures
Agriculture
Forest
115307
0
5
39
263
1601
Forest
116280
0
3
11
153
515
Artificial
bare soil
0
2059
0
286
0
1
Artificial
bare soil
0
2548
0
296
0
0
Bare
soil
0
0
239
39
35
16
Bare
soil
0
0
308
56
0
92
Artificial
area
7
499
20
11002
9
313
Artificial
area
0
227
33
5135
0
162
Water
0
0
0
8
0
0
Water
0
0
0
1
0
0
Pastures
212
1
26
363
13726
621
Pastures
180
0
22
0
4922
515
Agriculture
1139
58
1147
2085
1902
56578
Agriculture
240
19
1078
2168
458
56406
Kappa
91.63%
Kappa
94,09%
Overall
94.95%
Overall
96,79%
Class 2016
Forest
Artificial
bare soil
Bare
soil
Artificial
area
Water
Pastures
Agriculture
Class 2017
Forest
Artificial
bare soil
Bare
soil
Artificial
area
Pastures
Agriculture
Forest
114235
2
5
17
51
489
2879
Forest
115280
0
3
1
748
511
Artificial
bare soil
0
2083
0
1059
57
0
0
Artificial
bare soil
0
2401
0
727
0
0
Bare
soil
0
3
0
102
0
23
99
Bare
soil
78
0
55
119
0
185
Artificial
area
42
498
24
10270
3
3
1533
Artificial
area
116
383
174
4509
3
665
Water
0
0
0
0
2145
0
0
Water
0
0
0
0
0
0
Pastures
478
0
232
368
0
13085
922
Pastures
227
0
24
2
11892
869
Agriculture
1910
31
1176
2006
0
2335
53697
Agriculture
999
10
1188
2309
2553
56905
Kappa
87.23%
92.28%
Kappa
89.93%
94.15%
Overall
Overall
Class 2016
Forest
Artificial
bare soil
Bare
soil
Artificial
area
Water
Pastures
Agriculture
Class 2017
Forest
Artificial
bare soil
Bare
soil
Artificial
area
Pastures
Agriculture
Forest
116665
0
0
0
0
0
2
Forest
116453
0
2
8
485
298
Artificial
bare soil
0
2617
0
0
0
0
0
Artificial
bare soil
0
2647
0
150
0
2
Bare
soil
0
0
1437
0
0
0
0
Bare soil
0
0
825
13
0
42
Artificial
area
0
0
0
13821
0
0
0
Artificial
area
0
138
59
6526
0
187
Water
0
0
0
0
2256
0
0
Water
0
0
0
0
0
0
Pastures
0
0
0
0
0
15933
0
Pastures
93
0
10
0
10611
213
Agriculture
0
0
0
1
0
2
59128
Agriculture
154
9
548
970
4100
58393
Kappa
100%
100%
Kappa
93.71%
96.35%
Overall
Overall
GEOGRAPHY
Accuracy assessment results demonstrate how those ML
algorithms execute the classification. The best result is given by
the Random Forest algorithm with perfect accuracy of 100% for
2016 and 96.35% for 2017. In next part of this research, Random
Forest land cover map will be used for the final analysis.
Classification results are presented in Table 4:
Table 4. Classification results for RF land cover maps.
2016 Class
Pixel Sum
Percentage (%)
Area (km2)
Forest
7209286
65.89
720.93
Artificial bare soil
10320
0.09
1.03
Bare soil
7955
0.07
0.80
Artificial area
267578
2.45
26.76
Water
9883
0.09
0.99
Pastures
428576
3.92
42.86
Agriculture
3008299
27.49
300.83
2017 Class
Pixel Sum
Percentage (%)
Area (km2)
Forest
7419839
67.85
741.98
Artificial bare soil
17226
0.16
1.72
Bare soil
4049
0.04
0.40
Artificial area
192010
1.76
19.20
Water
9860
0.09
0.99
Pastures
543802
4.97
54.38
Agriculture
2748450
25.13
274.85
The results show that two classes are dominant with more
than 90% of the research area: Forest with 65.9% in 2016 and
67.6% in 2017 and Agriculture with 27.9% and 25.1%
respectively. Percentage of change is presented in Table 5.
Table 5. Change in classes. The positive values represent the
increase of pixels in 2017 while negative values present decrease
in 2017.
Class
Area (km2)
Percentage (%)
Forest
21.06
2.92
Artificial bare soil
0.69
66.92
Bare soil
-0.39
-49.10
Artificial area
-7.56
-28.24
Water
0.00
-0.23
Pastures
11.52
26.89
Agriculture
-25.98
-8.64
Change detection data in table 5 confirms the table 4 data
and presents how much each class has changed. The highest
increase has the Artificial bare soil (where our primary goal of
this work belongs Corridor 10 under construction), and Pasture
classes versus the Bare Soil, Agriculture, and Artificial classes
which decrease in area percentage cover. Figure 5 shows the
difference in the northern part of the research area where the
construction of Corridor 10 is in its full swing.
Figure 5. Northern part of the research area - Corridor 10 ongoing construction site.
GEOGRAPHY
CONCLUSION
As table 5 is presenting, the class of interest in this research
is within Artificial bare soil which presents the construction area
of new Corridor 10. It can be seen that there is an increase of the
area covered by this class which indicates that in one year there
were changes in the environment. Since the land cover is still
presented with same class and did not change into an Artificial
area where constructed paved highway belongs, we can
conclude that the motorway is still under construction. This data
acquired using remote sensing analysis of Sentinel 2 satellite
imagery can be of great help in monitoring changes of the
environment and big construction projects. Since the satellite
data are widely accessible and have satisfying ground resolution
with low, or no cost, we cannot exclude the remote sensing
techniques from the environmental research, but we must expand
the knowledge and capabilities provided. Random Forest
machine learning algorithm used in this paper confirms that the
classifying algorithms have advanced to the level when they can
be of great help to the environment analysts. High accuracy of
classified data obtained using Classification Tree algorithm gives
new perspective to remote sensing. Furthermore, different
machine learning algorithms (Random Forest, Gaussian Mixture
Model, K-Neighbors Classifier, and other) along with the
Artificial Neural Networks and Object Based Image Analysis
(OBIA) classification are in the focus of remote sensing
professionals and researchers, while rapid development and
improvement of the algorithms is in progress.
With this methodology, it is possible to perform a broad
spectrum of analysis, such as environmental stress detection
(landslides, wildfires, flooding, etc.) or land cover map creation
and other, with the very high percentage of accuracy while we
save time and money in the process that used to last much
longer.
REFERENCES
-Belgrade Net. (2017).
Retrieved from http://www.belgradenet.com/business/
Canziani, G., Ferrati, R., Marinelli, C., & Dukatz, F. 2008.
Artificial Neural Networks and Remote Sensing in the
Analysis of the Highly Variable Pampean Shallow
Lakes. Mathematical Biosciences and Engineering, 5(4).
doi:10.3934/mbe.2008.5.691
-Copernicus Open Access Hub. (2017).
Retrieved from https://scihub.copernicus.eu/dhus/#/home
Duro, D.C., Franklin, S.E., & Dubé, M.G. 2012. A Comparison
of Pixel-Based and Object-Based Image Analysis with
Selected Machine Learning Algorithms for the Classification
of Agricultural Landscapes Using SPOT-5 HRG
Imagery. Remote Sensing of Environment, 118, pp. 259-272.
doi:10.1016/j.rse.2011.11.020
-Elite Data Science. (2017). Retrieved from
https://elitedatascience.com/machine-learning-algorithms
-ESA STEP. (2017).
Retrieved from http://step.esa.int/main/toolboxes/snap/
-ESA. 2017. Sentinel 2 MSI.
Retrieved from https://earth.esa.int/web/sentinel/user-guides
Jensen, R.R., Hardin, P.J., & Yu, G. 2009. Artificial Neural
Networks and Remote Sensing. Geography Compass, 3, pp.
630-646. doi:10.1111/j.1749-8198.2008.00215.x
-Koridori Srbije. (2017). Retrieved from http://www.koridor10.rs
Lary, D.J., Alavi, A.H., Gandomi, A.H., & Walker, A.L. 2016.
Machine Learning in Geosciences and Remote
Sensing. Geoscience Frontiers, 7(1), pp. 3-10.
doi:10.1016/j.gsf.2015.07.003
Mas, J.F., & Flores, J.J. 2008. The Application of Artificial
Neural Networks to the Analysis of Remotely Sensed
Data. International Journal of Remote Sensing, Iss., 29(3).
doi:10.1080/01431160701352154
-Scikit learn. (2014).
Retrieved from http://scikit-learn.org/0.15/modules/
Article
Full-text available
Soil erosion is a global environmental and economic problem that is significantly related to land-use changes. Over the last decades, several mountain areas in Serbia were exposed to strong human pressure caused by winter tourism development. The largest ski center in Serbia is situated on Kopaonik Mountain within the boundaries of Kopaonik National Park, where the conflict between economic and conservation goals is rapidly growing. In this study, we calculated the sedimentation and surface runoff in three sub-basins in the area of Kopaonik ski resort for two years (1984 and 2018) using the Soil and Water Assessment Tool (SWAT) and analyzed the changes that occurred during the observed period. The results show an increase in surface runoff and sediment yield in sub-basins 1 and 3 and a decrease in sub-basin 2. The analysis of land cover change shows an expansion of evergreen forests, appearance of barren soil and urban areas, reduction of mixed forests and pastures, and the appearance of deciduous forests. These findings indicate that in the area studied, the dominant processes are the development of tourism and natural revegetation of abandoned agricultural land. Application of remote sensing techniques and SWAT contributes to identifying and monitoring land degradation problems and improving conservation and management practices.
Article
Full-text available
Continuous environmental mapping has never been more necessary because of the fast-paced lifestyle, technology development, climate change. Monitoring changes in the environment has become one of the most important tasks today to understand those changes and take the necessary measures to coordinate sustainable development with the emerging natural and socio-economic events. Remote sensing combined with cartography gives us the opportunity to obtain current information on the state of the environment and react suitably in order to reduce damage to a minimum in accordance with sustainable development.
Thesis
Full-text available
Електрична енергија је у савременом свету један од основних потреба сваког друштва. Њена производња се дуги низ година заснива на фосилним горивима која у себи садрже угљеник, представљена су угљем, нафтом и природним гасом и спадају у необновљиве изворе енергије. Иако се фосилна горива стварају путем природних процеса, сврставају се у необновљиве изворе енергије због веома дугог процеса настанка који може трајати и милионима година, док се њихова експлоатација одвија неупоредиво брже. Поред исцрпљивања лежишта фосилних горива, човечанство се сусрело са још једним великим проблемом који проистиче из њиховог коришћења (сагоревања) и који доводи до загађења животне средине испуштањем велике количине CO2 у атмосферу. Из ових разлога је неопходно окренути се изворима енергије који у мањој мери загађују животну средину и који се количински у природи произведу више него што је потребно за потрошњу, односно производњу енергије. У ову групу извора енергије спадају Сунчева енергија, снага ветра, енергија водних снага (енергија водотокова, морских струја и таласа, плиме и осеке), енергија биомасе и биогорива и геотермална енергија. Хидроенергија је од наведених обновљивих извора енергије најбитнији и најконкурентнији извор фосилним и нуклеарним изворима енергије. Иако се енергија водног потенцијала користи за производњу електричне енергије, утицај који се врши на животну средину приликом изградње хидроелектрана није мали. Свака промена услова животне средине је истовремено и стрес који се мора прецизно анализирати како би се стање животне средине након изградње хидроенергетских објеката вратило у приближно пређашње стање. Како се мале хидроелектране граде на водотоцима који имају мале протоке, утицај на живи свет у њима је веома велики јер свако мало колебање нивоа воде доводи до великог стреса по читав екосистем. Из тог разлога су анализирани стресови животне средине који настају изградњом мале хидроелектране и током њеног рада. За постизање циљева рада који се огледају у детерминисању узајамних утицаја животне средине и мале хидроелектране и геосистемској анализи простора за потребе инсталисања мале хидроелектране коришћене су различите научне методе, међу којима су извршене напредне анализе простора коришћењем географских информационих система и даљинске детекције. У раду је, поред основних физичко – географских и друштвено – географских података дат приказ стања електроенергетског система Србије и законске регулативе која се бави обновљивим изворима енергије. Геосистемска анализа простора је извршена за четири локације: слив реке Расине узводно од језера Ћелије и хидролошке станице Равни (43° 22' 26.6575" N, 21° 10' 22.6967" E) и три притоке реке Расине у чијим сливовима су предложене локације за изградњу мале хидроелектране у Катастру малих хидроелектрана из 1987. године. Како би анализа могла бити спроведена, прикупљени су различити подаци за проучавани простор. Дигитални модел терена је коришћен за приказ рељефа, креирање потенцијалних водотока и одређивање углова нагиба, мултиспектрални сателитски снимци су коришћени за израду карте земљишног покривача, хидрометеоролошки подаци су коришћени за анализу температуре ваздуха, падавина, брзине ветра, релативне влажности ваздуха и сунчевог зрачења и педолошка карта је коришћена за анализу типова земљишта на проучаваном простору. Коришћени подаци пописа становништва из 2011. године показују број корисника електричне енергије на територији Србије. За проучаване три притоке реке Расине у чијим сливовима су предложене локације за изградњу мале хидроелектране, дигитализовани су објекти са сателитских снимака високе резолуције из 2018. године који су изграђени на том простору.
Article
Full-text available
Suspended organic and inorganic particles, resulting from the interactions among biological, physical, and chemical variables, modify the optical properties of water bodies and condition the trophic chain. The analysis of their optic properties through the spectral signatures obtained from satellite images allows us to infer the trophic state of the shallow lakes and generate a real time tool for studying the dynamics of shallow lakes. Field data (chlorophyll-a, total solids, and Secchi disk depth) allow us to define levels of turbidity and to characterize the shallow lakes under study. Using bands 2 and 4 of LandSat 5 TM and LandSat 7 ETM+ images and constructing adequate artificial neural network models (ANN), a classification of shallow lakes according to their turbidity is obtained. ANN models are also used to determine chlorophyll-a and total suspended solids concentrations from satellite image data. The results are statistically significant. The integration of field and remote sensors data makes it possible to retrieve information on shallow lake systems at broad spatial and temporal scales. This is necessary to understanding the mechanisms that affect the trophic structure of these ecosystems.
A Comparison of Pixel-Based and Object-Based Image Analysis with Selected Machine Learning Algorithms for the Classification of Agricultural Landscapes Using SPOT-5 HRG Imagery. Remote Sensing of Environment
  • D C Duro
  • S E Franklin
  • M G Dubé
Duro, D.C., Franklin, S.E., & Dubé, M.G. 2012. A Comparison of Pixel-Based and Object-Based Image Analysis with Selected Machine Learning Algorithms for the Classification of Agricultural Landscapes Using SPOT-5 HRG Imagery. Remote Sensing of Environment, 118, pp. 259-272. doi:10.1016/j.rse.2011.11.020 -Elite Data Science. (2017). Retrieved from https://elitedatascience.com/machine-learning-algorithms -ESA STEP. (2017). Retrieved from http://step.esa.int/main/toolboxes/snap/ -ESA. 2017. Sentinel 2 MSI. Retrieved from https://earth.esa.int/web/sentinel/user-guides
  • D J Lary
  • A H Alavi
  • A H Gandomi
  • A L Walker
Lary, D.J., Alavi, A.H., Gandomi, A.H., & Walker, A.L. 2016. Machine Learning in Geosciences and Remote Sensing. Geoscience Frontiers, 7(1), pp. 3-10. doi:10.1016/j.gsf.2015.07.003
The Application of Artificial Neural Networks to the Analysis of Remotely Sensed Data
  • J F Mas
  • J J Flores
Mas, J.F., & Flores, J.J. 2008. The Application of Artificial Neural Networks to the Analysis of Remotely Sensed Data. International Journal of Remote Sensing, Iss., 29(3). doi:10.1080/01431160701352154 -Scikit learn. (2014). Retrieved from http://scikit-learn.org/0.15/modules/