TH 03 32
TH 03 32
ABSTRACT
Landslides are one of the severe natural hazards induced by heavy rainfall, deforestation, slope failure and urban
expansion. It can lead to significant loss of life and property in hilly and gully regions. Field studies that identify and
map landslides are expensive and time-consuming as it includes the cost of the survey, travelling, workforce, and
instrument. Although progression in technology and availability of high-resolution remote sensing data has now
made it possibleto identify landslides (satellite images and aerial photographs), accessibility to high-resolution
satellite data is still an expensive and tedious procedure. Several studies have conducted in a GIS environment to
map landslide zones, but the resolution of the open source data is commonly coarse (30 m), which adds to the
uncertainty of the outcome. In this study, application of the appropriate rule set with object-based image analysis
(OBIA) technique has been used to identify landslides zones, through a combination of spectral, textural and
geometrical properties of imagery and topographic data. It overcomes the shortcomings induced by pixel-based
classification. For the current study, High spatial resolution data such asGoogle earth imagery and CartoDEM (30 m)
has been used. This approach shows an excellent prospect for quick and near-to-actual assessment of landslides
zones which are generally induced by extreme rainfall events in the hilly regions of India. The methodology used has
the potential to facilitate more reliable disaster management strategies. This study shows the potential of open
source data and emerging technology in the field of landslide assessment.
1
towards the mapping of the landslide by using remote particular properties of landslides (e.g. shape) and the
sensing the techniques over the conventional methods. enhanced resolution of available imagery, pixel-based
However, the field observations are still the main bases classification techniques tend to result in noteworthy
of soil mapping as it captures the three-dimensional commission and omission errors (Casagli et al.
properties of Pedon (the smallest unit of soil). Currently, 2016).Rau et al. 2012 have foundthe outcomeof OBIA
no satellite dataensures this level of detailed is more sensible compared to the pixel-based
information.Advantages and limitations of conventional classification process which generally producesthe
methods are exhaustively presented in previous pepper-salt effect. Apart from the mentioned problems,
reviews (Scaioni et al. 2014 and Guzzetti et al. 2012) total reliability on brightness value may reduce the
and are not reminded here. accuracy as two pixels with same spatial reflectance
Scaioni et al. 2014 have discussed the importance might be entirely different types of objects/ features
and role of remote sensing in landslides inventory (e.g. Building and roads) or two pixels with very
studies.The coming of satellite optical imagery opened different reflectance may actually be part of the same
up new potentials in the practice of visual interpretation object type (e.g. different rooftop materials of
for landslide investigation. In the beginning, sensors buildings).Arithmetic operations with image bands can
had a low spatial resolution, in the order of 10–30 m highlight specific object classes. Vegetation indices are
(Resouresat, Landsat ). In the middle of the 90s, often used to categorise vegetation and separate it from
starting with IKONOS images, High Resolution and other categories (e.g. bare earth) in satellite data. The
Very High-Resolution( HR and VHR) optical sensors Excess Green Vegetation Index (ExG; Woebbecke et
were born, characterised by more outstanding al. 1995) is one of the most comprehensive indices if
properties regarding spatial resolution, number of only bands in the visible rangeare presented.(Hölbling
multispectral channels, temporal resolution, image et al. 2016)Have used “panchromatic” brightness layer
quality, global coverage, and stereo mode (Santurriet than brightness value (R, G, B). It was calculated for the
al. 2010). Some papers have been published in recent multispectral images by dividing the sum of the three
years on the visual or analytical interpretation of spectral bands by three. For the classification of
satellite images, mainly based on HR/VHR data (Tsai et landslides, this panchromatic layer was
al. 2010). Comparison of landslide inventories obtained valuableexposure of bare ground cause landslides
from aerial photos vs HR/VHR satellite images showed appear brighter as compare to their
that the new data source could provide equivalent immediatesurroundings on the imagery (Hölbling at al.
results even in areas where landslides haveleft faint 2015).
signs only (Fiorucci et al. 2011).Unfortunately, In this paper, a methodology has been proposed to
considering the high price of data acquisition, these map shallow eroded areas with a high level of detail
high-resolution images are generally utilised in a and accuracy using Google earth imageries and
specific small region and are rarely applied in large CartoDEM respectively. An object-based image
ones(Guo et al. 2016). analysis approach is used to classify eroded surfaces
Google Earth(GE) offers a welcome solution to the from the surroundings and also a new technique of
above-discussed issues;itprovides open, high spatial accuracy assessment is applied which judges the
resolution images suitable for landslide mapping. location as well as shapes and size of the extracted
However, Google Earth images are restricted to a objects with reference objects.
three-band colour code (R, G, and B), which is
expected to inferior the classification performance due
to its poor spectral resolution (Yu et al. 2011). The 2 STUDY AREA AND DATA
potential for the classification of spatial characteristics
by Google Maps has been underestimated(Drǎgutet al. Kalsi is an excellent tourist destination as it not only
2010). By analysing the tone, texture, and geometric provides one with beautiful views but also provides a
features in a GE image, experts can recognise eroded wide range of pastimes for the visitor coming here. It is
surfaces with high assurance. one of the most visited tourist destinations of India
Image analysis for landslide study is an important today. This small town is known both for its magnificent
process and wildly used in the field of earth science beauty and its historical values as well. Its one of the
because visual inspection cannot be avoided. In most preferred tourist destinations of north India.The
contrast to pixel-based image classification, the object- Kalsi is a beautiful village at the junction of Yamuna and
based image analysis (OBIA) partition land-cover parcel Tons. The 79.28-hectare study area is located
into image objects and classified by expert rules. Image approximately 49 kilometres from Dehradun city, India.
segmentation algorithms such as the multiresolution It comprises pastoral hill country on moderately
segmentation can be applied to create image objects, indurated Tertiary sandstone and mudstone, with relief
which serve as the basis for the classification process.A in the order of 780 m above sea level. Most of the
probable solution to the difficulties associated with area’s falls in the zone of heavy rainfall as a
pixel-based classification could be the need to consequence, rain-triggered shallow landslide erosion
functionat the spatial scale of the things ofconcern is common.
themselves,moderately than depending on the extent of
image pixels (Rapoza et al. 2008) hence the methods of
object-based image analysis (OBIA) for generating and
updating geographical information are becoming more
critical. Despite the advantages of OBIA, the visual
interpretation of ortho-photos or pixel-based
classification approaches are still the predominant
methods used for mapping landslides. Due to the
2
3.1 Segmentation
3
3.3 Classification spectral bands by three, which is termed as
‘Panchromatic”. For the classification of landslides, this
Automated thresholding for the classification with panchromatic layer was significantsince landslides
spectral features is implemented to the characteristics regions appear lighter in tone than their immediate
of each scene individuallyto classify the imageries. surroundings on the photographs due to the removal of
existing features which causes the exposure of bare
3.3.1 Excess Green Vegetation Index (ExG): ground (Hölbling et al. 2012).PAN was used to identify
the landslides whichwere left due to positing ExG value.
Excess Green Vegetation Index (ExG; Woebbecke et By referring to imagery, recent landslides appear as a
al. 1995,Mayr et al., 2016) is used to classify eroded bright objectwhich shows that they have higher PAN
surface from the surroundings. Google earth image value.
consists ofthe red (R), green (G) and blue (B) colour
bands which are used in the estimation of ExG. The 3.3.4 Other Characteristics
mathematical expression of ExG as follows in equation
[1]. Object-based image analysis has various advantageous
over pixel-based classification; it allows the user to work
with other properties such as geometric and texture
ExG = 2g – b – r [1] than spectral properties of images. Digital elevation
model is used as an ancillary data for landslide
Where, detection as landslide generally occurs at steep
b= slopes,so the integration or digital elevation models
(DEM) and its derivatives (e.g. slope, drainage pattern)
g= add to the accuracy of the extraction.
r= Geometrical properties such as length/width ratio
are proven to be very useful in extraction of road
feature. In the image, some places it is difficult to
The data was acquired on 22nd October 2017, so distinguish between the road and eroded surface as
according to thephenological cycle of the grass and they both have high reflectance properties which lead to
trees, theircoloursappear as dark green, while landslide the misclassification of the way into the eroded
areas appear as shades of brown. By observing the surfaces, but Road is a linear feature,so its L/B ration is
equation, it can understand that more weight is higher than the other classes. By training the
assigned to a greencolour,so the object reflects more segmentation algorithm according to the mentioned
green band will have higher ExG value and vice versa. properties, roads can be classified differently from the
As eroded surface has low reflectivity in green band, eroded surfaces,
the extracted ExG value is less than zero. These
indices work better when there is an enhanced 3.4 Accuracy Assessment
separability between the grass and eroded surface.
The assessment was carried out by a few parameters
3.3.2 Green Red Vegetation Indices (GRVI) of the discrepancy methods. Assessment has done by
the reference training object which was digitised
This index is beneficialto extract vegetation from manually by visual interpretation of imagery. This
Google earth images. As these images don’t contain method focused on the area parameters of the both
NIR band (Vegetation tends to have high reflectance in segmented and reference objects.Area-based method
NIR bands), these indices use only Red and Green assesses the accuracy of objects based on both
band for extraction of vegetation (Rau et al. 2011). The location and geometry. Several parameters havebeen
mathematical expression for GRVI is as follows in used in the study: (1) the relative area of an overlapped
equation [2]. region to a reference object (RAor), (2) the relative area
of an overlapped region to a segmented object (RAos),
(3) the quality rate (qr), (4) the SimSize, given as below,
GRVI = [2] (Taylor et al. 2013). Each factor judges the efficiency of
segmented objected concerning the reference object.
Its value ranges from -1 to +1, which is useful for The mathematical expression of each elementis shown
differentiating forest area from the sparse vegetation as in equation [3], [4], [5] and [6] respectively.
forest area has a higher GRVI value then sparse
vegetation. As in image, sparse vegetation appears [3]
brownish; it is very tedious to differentiate sparse
vegetation from the eroded surface. But, GRVI has a
different value for each class as green vegetation [4]
(GRVI > 0), soils (GRVI < 0), and water/snow (GRVI
close to 0) (Motohka et al., 2010). They are used to [5]
remove the sparse vegetation regions from the eroded
surface regions.
[6]
3.3.3 Panchromatic (PAN)
Where n is the number of segmented objects created
Anarithmetic average was calculated for the by an algorithm.Ar is the area of the reference object, As
multispectral images by dividing the sum of the three
4
is the area of the ith segmented object, Ao (i) is the area similarity in reflection with the eroded surface. That
of the ith overlapped region associated with the pixel can be removed by considering geometrical
reference object, and Au (i) is the area of the union properties like length/ width ratio. As the road is a linear
between the references object and the ith segmented feature, so objects were trained such that all object
object. RAorandRAor asses the accuracy by measuring which is having a length/width ratio higher than 3.5
the overlay region between the reference and should be classified as roads. To extract and
segmented objects. When objects are well-segmented, discriminate both forest and sparse vegetation classes,
the overlappingarea will be more so both RAor and GRVI indices are used. As forest will have high GRVI
RAosvalues will be close to 100. The quality rate value (Greater than 0.4) then sparse vegetation (0.1
parameters focus on both the geometry and location of ~0.2) and for rest, the average GRVI value was
the segmented objects.qr parameter (Weidner 2008) approximately 0.01~ 0.03. To detect freshly detached
ranges between 0 and 1. The values close to zero landslide, PAN is used as they tend to have a higher
indicate a perfect match while values close to one reflection,but they consist positive ExG value due to low
indicate an over or under segmentation. The SimSize red band reflection. So all the objects which are having
(Zhan et al. 2005) measures the similarity regarding the PAN value higher than 175, are classified as a freshly
size of the segmented object. It ranges between 0 and detached landslide, they were later on merged with the
1, with one being ideal.It assesseshowaccurately landslide class. The detailed imagery is shown in Figure
segmented objects have preserved the shape of the 4 (b). As built up is in a regular shape,so by using the
feature. The more alike object will have a higher value geometrical properties of shape index, they are
(close to 1) and vice versa. classified.A detached portion of land is represented as
a yellowcolour. Total 13 landslides were observed from
the imagery. While keeping the previous year imagery
4 RESULTS AND DISCUSSIONS (Figure 4(a)), there wasa minimal trace of landslides.
These Figures help in understanding the magnitude of
The outcome of the study is divided into two portions as the disaster.
follows: Landslides mapping and Factors affecting and
Accuracy assessment.
5
Figure 5. The topography of the study area
When comparing the total area of mapped landslides Overall 79.77 75.44 0.29 0.82
between the semi-automated object-based mapping
and the manual mapping, only minor differences are
detected (Figure 7). A slight trend towards
overestimating the landslide area with OBIA compared The reason could be understood by observing Figure 6
to the manual mapping can be recognised. as extracted objects are very irregular in shape and
continuity as compared to reference objects that leads
to a smaller the overlay region which lowers the
accuracy of these parameters.The qr parameter value
lies between 0 and 1. The values close to zero indicate
a perfect match,i.e. overlay and union area is equal
while values close to one indicate an over or under-
segmentation,i.e. overlay region is very small as
compare to the union because object lies away from
6
each other. The value obtained is 0.29 which indicatesa data such as IKONOS, Cartosat but they carry a
good similarity between overlay and union region.The limitation of cost and availability within.
SimSize processes the similarityregarding the size of Object-based image analysis was used for
the ith segmented object and ranges between 0 and 1, automatic extraction of various land use/ land cover; it
with 1, being ideal parameters considering both doesn’t promise to replicate the real scenario it just
geometric and spatial properties of the feature into helps in portraying a crude representation of
consideration.Simsize value is 0.82, which indicates the reality.Main drawbacks of visual interpretation are the
geometrical similarities of objects are quite high. uncertainty of outputs, the subjectivity and the strict
The manual approach demonstrationsgains for dependency on human expertise.While reference layer
delineatingsingle landslides or splitting up multiple is prepared through manual mapping, while generally
landslides into smallerlandslide regions. This is a accurate within the limits of image quality, is a very
challenge in OBIA since objectsproduced through time-consuming process and also its level of accuracy
segmentation rarely correspond to single landslides due depends on the efficacy of the user. This has to be
toscale issues which correspond to over- or under considered when reading accuracy values. Reference
segmentation.Under segmentation occurs whentwo or layer (Manually digitised) cannot be viewed as a true
more segments may represent a single object and over layer as landslide depends on the various geological,
segmentation occurs when a single segment anthropogenic and hydrological factors which cannot be
maycontain several objects respectively (Clinton et al., identified through visual interpretation. In this study, all
2008). Advanced split and merge algorithms could be of the elementswhich trigger the landslides are not
used to refine the delineation of image objects,or extracted. So field survey is must to testify the
manual editing can be done,but it is a tedious and time- outcomes.
consuming process. However, the creation of
“meaningful” objects about a particular context or aim
can be very complex (Blaschke et al., 2014). Thus, 6 CONCLUSION
instead of associating the absolute number of mapped
landslides, the overlapping area was used for The use of OBIA for landslides identification and
calculating the mapping accuracy in this study. classification has been proven as a promising method,
Although results from manual mapping performed by an advantage of the OBIA approach, compared to a
local experts are often the only reference available, they pixel-based classification, is that it copes with the noise
cannot constitute anutterlyaccurate reference as their of the high-resolutiondata. Low pass filtering such as
generation depends on various factors. Above smoothing of the spatial data is avoided prior to
mentioned factors have to be well-thought-out when segmentation of the image to keep the eroded area
interpreting accuracy values. boundaries as crisp and spatially accurate as possible.
Aggregation of features on a segment basis (averaging
pixel-based features oversimilarobjects), however,
5 UNCERTAINTY IN OUTCOME avoids a salt-and-pepper effect in the final classification.
Avoiding the need for manual threshold selection or
Remote sensing methods work on the basic principle of training areas, the information extraction workflow has
optics such as reflection, refraction,and absorption of advantages in terms of repeatability, objectiveness, and
radiation. How any object will respond to a transferability to other study areas, but the results
specificwavelength could be understood by its spectral obtained are still far from receivinga reasonable and
reflectance curve. The reflected radiation is further scientifically reliable landslide inventory. More attention
converted to a brightnessvalue but is a necessary input should be given to the classification rules and the data
to automatic extraction of features. If the reflection in the analysis. The lack of the DEM with a spatial
curve of soil is seen then few observation can be noted: resolution equal with the Google Earth and the lack of a
(1) Higher reflection for longer wavelength (Till 2.2 µm) more detailed lithological and geological map have
(2) Absorption band (1.4, 1.9 and 2.2 µm) (Fabre et al., brought to many uncertainties in the classification
2015). It implies that all the signification outcome lies in process.
the longer wavelength. Now coming to the study, Inaccuracy assessment, the developed landslide
Google earth imagery is used for feature extraction that detection algorithm achieved79.77, 77.54, 0.29 and
comprises only visible bands. Which indicates loss of 82percent for parameters RAor, RAos, qr and
valuable information as for soil reflection increase SimSizerespectively. While current accuracy parameter,
steadily towards longer wavelength in the visible region, overall accuracy reflects 96.7 percent accuracy. The
but so does construction material such as concrete and massive difference between the outcomes is due to
asbestos. Due to this problem, a significant amount of consideration of different factors in the analysis. Area-
misclassification can be observed between roads, built based accuracy methods give more reliable information
up and possible landslides. It was the main reason as it captures both geometry and location into account.
behind the inclusion of barren land into a potential The study demonstrates that landslide mapping can
landslide. This caused uncertainty in the outcome. be done by using open source data with considerable
As ancillaryinformation, digital elevation model is accuracy.
used to map possible landslides. But CartoDEM has a
spatial resolution of 30 m which is larger than the
spatial resolution of google earth 0.46 m. It dilutes the 7 REFERENCES
information for each pixel.The significant difference in
spatial resolution of both layers makes outcome from Blaschke T., Hay G. J., Kelly M., Lang S., Hofmann P.
overlay operation unreliable. This error can be reduced and Adding E., 2014. Geographic Object-Based
to a great extent by using high-resolution stereo pair Image Analysis – Towards a new paradigm. ISPRS
7
Journal of Photogrammetry and Remote Sensing, Laliberte A.S., Browning D.M. and Rango A. 2012. A
87, 180-191 comparison of three feature selection methods for
Campbell J.B. And Wynne R.H., 2011. Introduction to object-based classification of sub-decimeter
Remote Sensing”. New York: The Guilford Press, resolution Ultracam-L imagery. Int. J. Appl. Earth
5th Ed., 667 Pp. Obs., 15, 70–78.
Casagli N., Cigna F., Bianchini S., Hölbling D., Füreder Malamud B.D., Turcotte D. L., Guzzetti F. and
P., Righini G., Del Conte S., Friedl B., Reichenbach P. 2004. Landslide inventories and
Schneiderbauer S., Iasio C., Vlckoh J., Greif V., their statistical properties. Earth Surf. Process.
Proske H., GranicaK.,Falco S., Lozzi S, Mora O., Landforms 29, 687–711.
Arnaud A., Novali F. and Bianchi M. 2016. Landslide Mayr A., Rutzinger M., Bremer M. and Geitner C. 2016.
mapping and monitoring by using radar and optical Mapping eroded areas on mountain grassland with
remote sensing: Examples from the EC-FP7 project terrestrial photogrammetry and object-based image
SAFER. Remote Sensing Applications: Society and analysis. ISPRS Annals of the Photogrammetry,
Environment 4: 92–108. Remote Sensing and Spatial Information Sciences,
Clinton N., Holt A., Yan L., Gong P. 2008. An Accuracy Volume III-5, XXIII ISPRS Congress, Prague, Czech
Assessment Measure for Object-Based Image Republic-5:137-144.
Segmentation. The International Archives of the Motohka T., Nasahara K.N., Oguma H. and Tsuchida,
Photogrammetry, Remote Sensing and Spatial S.
Information Sciences. Vol. XXXVII. Part B4. Beijing. 2010. Applicability of Green-Red Vegetation Index
Cruden D.M. and Varnes D.J. 1996. Landslide Types for Remote Sensing of Vegetation Phenology.
and Processes, Special Report, Transportation Remote Sensing 2, pp.2369- 2387.
Research Board, National Academy of Sciences, Mulders M.A. 1987. Remote Sensing in Soil Sciences,
247:36-75. ELSEVIER SCIENCE PUBLISHERS B.V. The
DrǎgutL.andEisank C. 2010. ESP: a tool to estimate Netherlands
scale parameters for multi-resolution image Platt, R.V. and Rapoza, L., 2008.An evaluation of an
segmentation of remotely sensed data. Int. J. the object-oriented paradigm for land use/land cover
Geogr. Inf. Sci. 24 (6), 859–871. classification, Professional Geographer, 60, pp. 87-
Dragut, L. and Blascke, T. 2006. 100.
Automatedclassification of landform elements using Rau J. Y., Jhan J. P., Lob C. F., Linb Y. S 2012.
object-based image analysis, Geomorphology 81, Landslide mapping using imagery acquired by a
pp. 330-344. fixed-wing UAV, International Archives of the
Fabio Vittorio De Blasio 2011. Introduction to the Photogrammetry, Remote Sensing and Spatial
Physics of Landslides, Springer Dordrecht Information Sciences, Volume XXXVIII-1/C22, 2011
Heidelberg London New York. ISPRS Zurich Workshop, Zurich, Switzerland : 195
Fabre, S., Briottet X. and Lesaignoux A. 2015. -200.
Estimation Rau J.Y., Jhan J.P., Lo C.F. and Lin Y.S. 2011.
of Soil Moisture Content from the Spectral Landslide
Reflectance of Bare Soils in the 0.4–2.5 µm Domain, mapping using imagery acquired by a fixed-wing
Sensors, 15: 3262-3281. UAV. Int. Arch. Photogramm. Remote Sens. Spat.
Fiorucci F.,CardinaliM.,CarlàR., Rossi M.,Mondini Inf. Sci., 38/1-C22, 195–200.
A.C., SanturriL.,ArdizzoneF.,Guzzetti F. 2011. Sharma N. Jul 17, 2017. Frequent landslides in
Seasonal landslide mapping and estimation of Uttarakhand worries scientists. Hindustan Times.
landslide mobilization rates using aerial and satellite Taylor P., Montaghi A., Larsen R., and Greve M.
images, Geomorphology, 129: 59–70. H.2015.
Guo Z., Shao X., Xu Y., Miyazaki H., Ohira W. and Accuracy assessment measures for image
Shibasaki R. 2016. Identification of Village Building segmentation goodness of the Land Parcel
via Google Earth Images and Supervised Machine Identification System (LPIS) in Denmark. Remote
Learning Methods, Remote Sens. 2016, 8, 271. Sensing Letters, 4:10, pp-946-955.
Guzzetti F., Mondini A., Mondini A., Cardinali M., Tsai F., Hwang J.H., Chen L.C., Lin T.H. 2010. Post-
Fiorucci disaster assessment of landslides in southern
F., Santangelo M., Chang k. 2012. Landslide Taiwan after 2009 Typhoon Morakot using remote
inventory maps: New tools for an old problem. sensing and spatial analysis. Nat. Hazard. Earth
Earth-Science Reviews 112:42-66. Syst., 10, 2179–2190.
Hölbling D., Betts H., Spiekermann R., Phillips C. 2016. Santurri L., Carlà R., Fiorucci F., Aiazzi B., Baronti S.,
Semi-automated landslide mapping from historical Cardinali M. and Mondini A. 2010. Assessment of
and recent aerial photography, AGILE 2016 – very high-resolution satellite data fusion techniques
Helsinki, June 14-17, 2016. for landslide recognition, Int. Arch. Photogramm.
Hölbling D., FriedlB.andEisank C. 2015. An object- Remote Sens. Spat. Inf. Sci., 38/B7, 492–496.
based approach for semi-automated landslide Scaioni M., Longoni L., Valentina Melillo V. and Monica
change detection and attribution of changes to Papini M. 2014.Remote Sensing for Landslide
landslide classes in northern Taiwan. Earth Science Investigations: An Overview of Recent
Informatics Volume 8, Issue 2, pp 327–335. Achievements and Perspectives. Remote Sens., 6,
Hölbling D., Füreder P., Antolini F., Cigna F., Casagli N. 1-x manuscripts; doi: 10.3390/rs60x000x
and Lang S. 2012. A semi-automated object-based Selby, M.J. 1993. Hillslope materials and processes,
approach for landslide detection validated by Oxford University Press, Oxford.
persistent scatterer interferometry measures and Weidner U. 2008. Contribution to the assessment of
landslide inventories. Remote Sens., 4, 1310–1336.
8
segmentation quality for remote sensing
applications, Proceedings of the 21stCongress for
the International Society for Photogrammetry and
Remote Sensing, 03–11 July, Beijing, China.
Woebbecke, D. M., Meyer, G. E., Von Bargen, K. and
Mortensen, D. A. 1995. Colour indices for weed
identification under various soil, residue, and lighting
conditions. Transactions of the ASAE, 38(1), 259-
269.
Yu, L. and Gong, P. 2011. Google Earth as a virtual
globe
tool for earth science applications at the global
scale: Progress and perspectives. Int. J. Remote
Sens. 33: 366–3986.
Zhan Q., Molenaar M.,Tempfli K., and Shi W. 2005.
Quality assessment for geospatial objects derived
from remotely sensed data, International Journal of
Remote Sensing, 26(14):2953–2974.