Application of Landsat-8 and ALOS-2 data for structural ...eprints.utm.my/id/eprint/74865/1/MazlanHashim2017... · 1286 A. B. Pour and M. Hashim: Application of Landsat-8 and ALOS-2

Nat Hazards Earth Syst Sci 17 1285ndash1303 2017httpsdoiorg105194nhess-17-1285-2017copy Author(s) 2017 This work is distributed underthe Creative Commons Attribution 30 License

Application of Landsat-8 and ALOS-2 data for structuraland landslide hazard mapping in Kelantan MalaysiaAmin Beiranvand Pour and Mazlan HashimGeoscience and Digital Earth Centre (INSTeG) Universiti Teknologi Malaysia Johor Bahru UTM Skudai 81310 Malaysia

Correspondence to Amin Beiranvand Pour (beiranvandamin80gmailcom abeiranvandutmmy) and Mazlan Hashim(mazlanhashimutmmy)

Received 28 May 2016 ndash Discussion started 4 July 2016Revised 18 December 2016 ndash Accepted 4 July 2017 ndash Published 28 July 2017

Abstract Identification of high potential risk and suscep-tible zones for natural hazards of geological origin is oneof the most important applications of advanced remotesensing technology Yearly several landslides occur duringheavy monsoon rainfall in Kelantan River basin Peninsu-lar Malaysia Flooding and subsequent landslide occurrencesgenerated significant damage to livestock agricultural pro-duce homes and businesses in the Kelantan River basin Inthis study remote sensing data from the recently launchedLandsat-8 and Phased Array type L-band Synthetic Aper-ture Radar-2 (PALSAR-2) on board the Advanced Land Ob-serving Satellite-2 (ALOS-2) were used to map geologicstructural and topographical features in the Kelantan Riverbasin for identification of high potential risk and suscepti-ble zones for landslides and flooding areas The data wereprocessed for a comprehensive analysis of major geologi-cal structures and detailed characterizations of lineamentsdrainage patterns and lithology at both regional and districtscales The analytical hierarchy process (AHP) approach wasused for landslide susceptibility mapping Several factorssuch as slope aspect soil lithology normalized differencevegetation index (NDVI) land cover distance to drainageprecipitation distance to fault and distance to the road wereextracted from remote sensing satellite data and fieldwork toapply the AHP approach Directional convolution filters wereapplied to ALOS-2 data for identifying linear features in par-ticular directions and edge enhancement in the spatial do-main Results indicate that lineament occurrence at regionalscale was mainly linked to the NndashS trending of the BentongndashRaub Suture Zone (BRSZ) in the west and Lebir Fault Zonein the east of the Kelantan state The combination of differ-ent polarization channels produced image maps that contain

important information related to water bodies wetlands andlithological units The NndashS NEndashSW and NNEndashSSW lin-eament trends and dendritic sub-dendritic and rectangulardrainage patterns were detected in the Kelantan River basinThe analysis of field investigation data indicates that many offlooded areas were associated with high potential risk zonesfor hydrogeological hazards such as wetlands urban areasfloodplain scroll meander bend dendritic and sub-dendriticdrainage patterns which are located in flat topographic re-gions Numerous landslide points were located in a rectan-gular drainage system that is associated with a topographicslope of metamorphic and quaternary rock units Conse-quently structural and topographical geology maps were pro-duced for Kelantan River basin using PALSAR-2 data whichcould be broadly applicable for landslide hazard mappingand identification of high potential risk zone for hydroge-ological hazards Geohazard mitigation programs could beconducted in the landslide recurrence regions and flooded ar-eas to reduce natural catastrophes leading to loss of life andfinancial investments in the Kelantan River basin In this in-vestigation Landsat-8 and ALOS-2 have proven to success-fully provide advanced Earth observation satellite data fordisaster monitoring in tropical environments

1 Introduction

The state of Kelantan is located in the northeastern corner ofPeninsular Malaysia (Fig 1) Yearly several landslides oc-cur during heavy monsoon rainfall in Kelantan River basinIn recent years in particular there have been many severeflooding events (in the year 2005 2006 2007 2008 2009

Published by Copernicus Publications on behalf of the European Geosciences Union

1286 A B Pour and M Hashim Application of Landsat-8 and ALOS-2 data for structural mapping in Kelantan

Figure 1 Location of the Kelantan state in Peninsular Malaysia

2014 and 2015) which have led to significant damage to live-stock agricultural production homes and businesses in theKelantan River basin (Pradhan 2009 Pradhan et al 2009Pradhan and Youssef 2011 Tehrany et al 2013 Nazarud-din et al 2014) The problem stems from the inappropriateuse of lands that are vulnerable to erosion quick water runoffand slope failure Two distinct wet seasons from Septem-ber to December and February to May are reported by theMalaysian Meteorological Department (MMD) in PeninsularMalaysia (Mardiana et al 2000) A flood episode in Decem-ber 2014 caused several landslides one of which occurred foreach kilometer of the area of Kelantan state Mapping land-slides in tropical environments is difficult because landscapesare generally covered by thick tropical rainforest (Plank etal 2016) In addition the cloudy weather and rain affectedthe quality of the optical satellite remote sensing data (Mel-gani 2006 Ju and Roy 2008 Ramli et al 2009 Hashim etal 2013 Razak et al 2013 Eckardt et al 2013 Shahabiand Hashim 2015) The main target of this investigation isto identify high potential risk and susceptible zones in theKelantan River basin for mapping landslide occurrence zoneusing satellite remote sensing technology and geographic in-formation system (GIS) techniques

Advances in remote sensing technology allow the applica-tion of synthetic aperture radar (SAR) data in the geological

structural analysis for tropical environments (Shimada andIsoguchi 2002 Morelli and Piana 2006 Ramli et al 2009Pour and Hashim 2014 2015a b Pour et al 2013 2014)Structural field mapping is often difficult in heavily vegetatedterrain This is the case with the study area where dense veg-etation cover deep weathering and scarcity of bedrock ex-posure hamper geological structure mapping over a long dis-tance In addition the use of optical remote sensing data islimited due to the persistent cloud coverage of the study areafor the most part of the year (Ramli et al 2009 Hashim etal 2013) SAR data contain the potential to penetrate cloudand vegetation and unlike the optical data are dependentupon the surface roughness of the materials ideally suitedto mapping lineaments (faults and fractures) in tropical en-vironments Lineaments are related to large structural frac-tures which represent zones of weakness in the brittle partof the lithosphere The presence of linear tectonic structuresis one of the most important factors in geological hazard oc-currences (Bannert 2000a b) Lineaments are representedby faults (linear features) lithological contacts between rockunits and drainage patterns in any area (van der Pluijm andMarshak 1997) Observation from satellite images (LandsatThematic Mapper) in the Himalayan Mountains of Nepal andChina revealed a clear connection between active faults as-sociated with earthquakes and the occurrence of large land-slides (Bannert 2000a b) Therefore delineation of faultsand fractures using advanced remote sensing technology inany region is a necessity to assess the potential for many nat-ural geological hazards Numerous investigations have usedfaults drainage patterns and lithology factors as importantfactors to measure and map the susceptibility of geologicalhazard (Guzzetti et al 1999 Dai and Lee 2002 Suzen andDoyuran 2004 Abdullah et al 2013)

Landsat-8 was launched on 4 February 2013 carrying theOperational Land Imager (OLI) and the Thermal InfraredSensor (TIRS) sensors These two instruments collect im-age data for nine visible near-infrared shortwave infraredbands and two longwave thermal bands (Irons et al 2012)They have a high signal-to-noise radiometer performanceenabling 12 bit quantization of data allowing for more bitsfor better land cover characterization by sharper color con-trast Landsat-8 provides moderate-resolution imagery from15 to 100 m of Earthrsquos surface and polar regions (Roy et al2014)

The Advanced Land Observing Satellite-2 (ALOS-2) waslaunched on 24 May 2014 as the successor of ALOS-1 (launched on 24 January 2006 and decommissioned inMay 2011) (httpglobaljaxajppress20140520140524_daichi2html Blau 2014) The ALOS-2 is exclusively in-stalled with the Phased Array type L-band Synthetic Aper-ture Radar-2 (PALSAR-2) using microwaves to maximizedits ability compare to the ALOS-1 on which three sensors(one microwave and two optical devices) were on board(Igarashi 2001 Suzuki et al 2012)

Nat Hazards Earth Syst Sci 17 1285ndash1303 2017 wwwnat-hazards-earth-syst-scinet1712852017

A B Pour and M Hashim Application of Landsat-8 and ALOS-2 data for structural mapping in Kelantan 1287

In particular L-band microwaves from PALSAR-2 havethe ability to penetrate vegetation due to relatively longwavelengths (about 24 cm) making the data particularly use-ful for geological structural mapping in tropical environ-ments (Igarashi 2001 Rosenqvist et al 2004 ERSDAC2006 Arikawa et al 2010 Yamamoto et al 2013 Pourand Hashim 2014 2015a b Pour et al 2013 2014 Shi-mada et al 2015) The wavelength of the L band is rela-tively long among microwaves (C-band is about 6 cm and X-band is about 3 cm) allowing it to travel all the way down tothe ground through vegetation (Woodhouse 2006) Not onlycan information be obtained about vegetation but informa-tion of the ground surface can be obtained as well Addition-ally the L band is less affected by the growth of vegetationwhich is useful for SAR interference analysis (interferome-try) Therefore the L band is capable of acquiring changeson the land more precisely compared to shorter wavelengthSAR when some diastrophism takes place due to an earth-quake or volcanic activity and floods or landslides causedby a natural disaster (Suzuki 2014) Accordingly a furtherincrease in the amount of information for geological struc-tural mapping could be derived from the recently launchedPALSAR-2 data Analysis of the data can provide completelynew insights into heavily vegetated areas threatened by natu-ral hazards of geological origin Consequently the advancedSAR remote sensing data are broadly applicable for geoenvi-ronmental research to identify the causes of natural disastersand point the way to rehabilitation measures especially intropical environments

In this investigation multitemporal Landsat-8 thePALSAR-2 data and GIS techniques were integrated to de-tect and map landslide occurrence zones in the Kelantanstate following massive rainfall in December 2014 The spe-cific objectives that drive this investigation are (i) to identifyhigh potential risk and susceptible zones for geological originhazards using the recently launched ALOS-2 PALSAR-2 re-mote sensing data in the Kelantan River basin at regional anddistrict scales (ii) to produce accurate geological structureand topographical maps for the Kelantan River basin usingLandsat-8 and PALSAR-2 data and (iii) to compare the de-tected high potential risk and susceptible zones with highlydamaged areas from recent flooding events in the KelantanRiver basin

The remainder of this paper is organized as follows Sec-tion 2 briefly reviews geology of the Kelantan state Section 3explains the characteristics of satellite remote sensing dataand their applications in landslide mapping and methodologyused in the study Section 4 describes results and discussionFinally Sect 5 concludes with some accomplishments

2 Geology of the study area

Peninsular Malaysia forms an integral part of the SoutheastAsian continental core of Sundaland and comprises two tec-

tonic blocksterranes the Sibumasu Terrane in the west andthe Sukhothai Arc in the east which were assembled by theLate Triassic (Metcalfe 2013a b) The state of Kelantan islocated in the northeastern corner of Peninsular Malaysia(Fig 1) Kelantan River is the major river in the region It ap-pears at the convergence of the Galas River and Lebir Rivernear Kuala Kari and meanders over the coastal plain untilit finally degrades into the South China Sea Kelantan Riverbasin covers 923 km2 which is about 85 of the Kelantanstatersquos surface area It is composed of the flat slope to mod-erately sloping areas in the northern part and steep scrapsand high slopes in the southern part of the river basin (Prad-han et al 2009) A wide variety of rocks consisting of ig-neous sedimentary and metamorphic rocks are distributedin a northndashsouth trend in the Kelantan state Typically fourtypes of rocks are classified in the region including graniticrocks sedimentarymetasedimentary rocks extrusive rocks(volcanic rocks) and unconsolidated sediments (Fig 2) Lo-calized geological features comprise faulting and jointingin the granitic rocks and folding faulting and jointing inthe sedimentary rocks Granitic rocks are distributed in thewest (the Main Range granite) and east borders (the Bound-ary Range granite) of the state of Kelantan (Rahman andMohamed 2001 Department of Minerals and GeoscienceMalaysia 2003 Heng et al 2006)

The Main Range Granite is located in the west of thestate which is stretched along western Kelantan up to theboundary of Perak and Pahang states and Thailand (Fig 2)The dominant structural trend in Kelantan is along the NndashSto NWndashSE direction that derived from post-orogenic phase(Ghani 2009) The Main Range Granite is located along thewestern margin of the BentongndashRaub Suture Zone (BRSZ)and extends north to Thailand (Schwartz et al 1995 Met-calfe 2000) The northndashsouth-trending BentongndashRaub Su-ture (approximately 13 km wide) extends from Thailandthrough Raub and Bentong to the east of Malacca PeninsularMalaysia The BRSZ is characterized by a series of paralleltopographic northndashsouth-trending lineaments and the pres-ence of small bodies of mafic to ultramafic rocks that arecommonly serpentinized (Tan 1996) The Lebir Fault Zoneis located in the eastern part of Kelantan state (Fig 2) whichis one of the major lineaments in Peninsular Malaysia andconsidered to be post-Cretaceous and a sinistral strike-slipfault (Tjia 1989 Harun 2002)

The landscape in the state of Kelantan has been dividedinto four types mountainous areas hilly areas plain areasand coastal areas (Unjah et al 2001 Raj 2009) MountChama (Gunung) is the highest point (2171 m) in the Kelan-tan state located in Gua Musang district in the western partof the state near the border of Perak state (Nazaruddin etal 2014) Quaternary deposits consist of alluvium depositsor unconsolidated sediments Triassic marine siliciclasticsvolcaniclastics sandstone and limestone are the other sedi-mentary rocks found in the Kelantan state (Nazaruddin et al2014)

wwwnat-hazards-earth-syst-scinet1712852017 Nat Hazards Earth Syst Sci 17 1285ndash1303 2017

Figure 2 Geologic map of the Kelantan state (modified from De-partment of Minerals and Geoscience Malaysia 2003)

3 Materials and methods

31 Satellite remote sensing data and characteristics

Landsat-8 and ALOS-2 datasets were used in this studyPALSAR-2 of the ALOS-2 has been significantly improvedfrom ALOS-1rsquos PALSAR in all aspects including resolutionobservation band and time lag for data provision (Arikawa etal 2010 Kankaku et al 2010 Suzuki et al 2012) ALOS-2 science capabilities include global environmental monitor-ing using the time series PALSAR-2 The research target alsocovers biospheric cryospheric and coastal ocean research aswell as disaster mitigation (Shimada 2013) PALSAR-2 isa microwave sensor that emits L-band radio waves and re-ceives their reflection from the ground to acquire information(Suzuki et al 2012) It has three observation modes includ-ing (i) Spotlight mode the most detailed observation modewith 1 m by 3 m resolution and observation width of 25 km(ii) Strip map mode a high-resolution mode with the choiceof 3 m (ultra fine) 6 m (high sensitivity) or 10 m (fine) resolu-tion and observation width of 50 or 70 km and (iii) ScanSAR

mode a broad area observation width of 350 km (nominal)or 490 km (wide) and resolution of 100 or 60 m (Yamamotoet al 2013 Shimada et al 2015) Selection of the mostsuitable observation mode and time acquisition of PALSAR-2 data for the research objectives will maximize the effi-ciency of geoenvironmental monitoring works Two distinctwet seasons from September to December and February toMay are reported by the MMD in Peninsular Malaysia (Mar-diana et al 2000) Precipitation and soil moisture variationshave an influence on detailed geologic lineament analysis us-ing microwave signal SAR data are highly sensitive to thewater content in the soil because of large contrast between di-electric properties of water and dry soil (Eagleman and Lin1976 Jackson and Schmugge 1989 Lacava et al 2005)Hence SAR data acquired during dry seasons contain moreuseful information for detailed geological structural mappingin tropical environments

The ScanSAR observation mode has 60 m resolution and490 km swath width (wide mode) and the fine observationmode contains data with 10 m resolution and 70 km swathwidth Dual polarization for the both data includes HH+HVpolarization images (HH is horizontally transmitted and hor-izontally received HV is horizontally transmitted and verti-cally received) HV channel is a channel that transmits a vi-brating wave in a horizontal direction against the ground (H)and receives it vertically (V) It is expected to acquire moredetailed observation data on the ground Particularly HVchannel (cross polarization) in the ScanSAR and fine obser-vation modes increase the amount of information extractionfor geological structural mapping at both regional and dis-tricts scales (Pour and Hashim 2015a b) HV polarizationis more suitable for lineament extraction and edge enhance-ment in tropical environments than other polarization chan-nels because cross polarization is more sensitive to linea-ment and also enhances penetration (Henderson and Lewis1998 Pour and Hashim 2015a b) Penetration is propor-tional to wavelength and cross polarization also enhancespenetration (Henderson and Lewis 1998) Therefore HVpolarization channel records more geological features thatare covered by dense vegetation In the level 31 productof PALSAR-2 image quality corrections (noise reductionand dynamic range compression) are performed to level 15PALSAR-2 data It should be noted that level 15 PALSAR-2 data contain the following characteristics (i) range andmulti-look azimuth compressed data are represented by am-plitude data (ii) range coordinate is converted from slantrange to ground range and (iii) map projection is also per-formed (Japan Aerospace Exploration Agency 2014) Thepositioning accuracy of PALSAR-2 data is excellent and itdoes not always require the field survey to modify the posi-tioning accuracy So it can reduce the map production timeand contribute to cost reduction as well (Shimada et al2015)

Two level 1T (terrain-corrected) Landsat-8 OLIimages were obtained through the US Geologi-

cal Survey Earth Resources Observation and Sci-ence Center (httpearthexplorerusgsgov) Theimage data (LC81270562014339LGN00 andLC81270562015070LGN00) (pathrow 12756) wereacquired on 5 December 2014 (before the flooding event)and 11 March 2015 (after the flooding event) with lowcloud cover The image data cover four river basins inKelantan including Sungai Golok (103 91837 ha) SungaiKelantan (1 308 69095 ha) Sungai Kemasin (30 92782 ha)and Sungai Semerak (59 07648 ha)

In this investigation a ScanSAR-mode dual polarization(level 31 acquired on 6 February 2015) and two fine-modedual polarization (level 31 acquired on 6 February 2015)PALSAR-2 scenes were obtained from ALOS-2 data dis-tribution consortium online system Remote Sensing Tech-nology Center of Japan (RESTEC) (httpwwwrestecorjpenglishindexhtml) and PASCO Corporation (httpenalos-pascocomhttpssatpfjp) for comprehensive analysisof major geological structures and detailed characterizationsof lineaments in the state of Kelantan The data were pro-cessed using the ENVI (Environment for Visualizing Images)version 52 and ArcGIS version 103 software packages

32 Methods

Atmospheric correction was applied by Fast Line-of-sight Atmospheric Analysis of the Spectral Hyper-cubes (FLAASH) algorithm on Landsat-8 OLI imagesThis algorithm contains the standard MODTRAN modelatmospheres which has standard column water vaporamounts for each model atmosphere Moreover scene andsensor information (solar zenith angle satellite view angleand relative azimuth angle between the satellite and thesun) and the aerosol model are considered for running thealgorithm (Thome et al 1998)

Lineaments in remote sensing images have topographicrelief and are often a surface expression of 3-D geologi-cal structures in the subsurface Lineaments associated withbrittle and ductile deformation zones appear as rectilinearand curvilinear patterns in remotely sensed images Detailedextraction of lineaments helps in reconstructing the tectonichistory of a region (Clark and Wilson 1994) Geological lin-eament features are attributed to paleotectonic andor neo-tectonic activity of a region The use of remote sensing datain delineation of tectonically significant lineaments has beendemonstrated in many geological settings (Raharimahefa andKusky 2007 2009 Amri et al 2011 Hashim et al 2013Hamimi et al 2014 Pour and Hashim 2014 2015a b Pouret al 2014)

Spatial transforms provide reliable and robust image pro-cessing techniques to extract the spatial information from re-mote sensing data The techniques help to maximize clar-ity sharpness and details of features of interest towards in-formation extraction and further analysis Spatial convolu-

tion filtering is based primarily on the use of convolutionmasks This flexibility makes convolution one of the mostuseful tools in image processing The procedure could beused to enhance low- and high-frequency details as well asedges in the imagery (Haralick et al 1987 Jensen 2005Schowengerdt 2007) Linear features are formed by edgesof remotely sensed images Some linear features occur asnarrow lines against a background of contrasting brightnessothers are the linear contact between adjacent areas of dif-ferent brightness Edge enhancement delineates these edgesand makes the shapes and details comprising the image moreconspicuous and easier to analyze (Jensen 2005) It can beused in geological applications to highlight rectilinear andcurvilinear patterns in remote sensing images

Systematic image processing techniques were imple-mented in the PALSAR-2 data for geological structures andlineament mapping at both regional and district scales in thestate of Kelantan It is necessary to treat the speckle in radarimages by filtering before it can be used in various applica-tions (Sheng and Xia 1996) The presence of speckle in radarimages reduces the detectability of ground targets obscuresthe spatial patterns of surface features and decreases the ac-curacy of automated image classification (Lee and Jurkevich1994 Sveinsson and Benediktsson 1996) However imagequality corrections (noise reduction) have been already ap-plied to level 31 product of PALSAR-2 but some speckles(salt and pepper noise) could still be seen in the images Inthis study the median spatial convolution filter was used fornoise removal and smoothing the PALSAR-2 images Themedian filter is a particularly useful statistical filter in thespatial domain which effectively remove speckle (salt andpepper noise) in radar images without eliminating fine details(Russ 2002 Research Systems Inc 2008) The median op-eration has the effect of excluding pixels that do not fit thetypical statistics of the local neighborhood ie outliers Iso-lated noise pixels can therefore be removed with a medianfilter (Schowengerdt 2007) The median filter is especiallyuseful for removing shot noise (pixel values with no rela-tion to the image scene) and has certain advantages such notshifting boundaries and minimal degradation to edges (Elia-son and McEwen 1990 Russ 2002 Jensen 2005) In thisstudy a 3times 3 neighborhood convolution mask (kernel) wasapplied to the PALSAR images Image Add Back value wasentered at 60 The Image Add Back value is the percent-age of the original image that is included in the final outputimage This part of the original image preserves the spatialcontext and is typically done to sharpen an image

The directional nature of geological lineaments accentu-ates the need for directional filtering to obtain maximumstructural mapping efficacy Edge enhancing filter highlightsany changes of gradient within the image features such asstructural lines (Tripathi and Gokhale 2000) A linear fil-ter is calculated in the spatial domain as a weighted sum ofpixels within the moving window This discrete convolutionbetween the input image f and the window response func-

tion w both of size NX timesNY is written mathematically forthe output pixel gij as

Nxminus1summ=0

Nyminus1sumn=0

fmnwiminusmjminusn (1)

and expressed symbolically in the convenient form of

g = f middotw (2)

Since the nonzero extent of the window is typically muchsmaller than the image the sum in Eq (1) does not have tobe over every pixel If the window is wx timeswy pixels we canwrite an alternate expression

i+Wy2summ=iminusWy2

j+Wx2sumn=jminusWx2

where w is centered at (0 0) and is nonzero over plusmnwx2and plusmnw2

x2 In this form we can clearly see that the out-put pixel is a weighted sum of pixels within a neighborhoodof the input pixel The distinguishing characteristic of a lin-ear filter is the principle of superposition which states thatthe output of the filter for a sum of two or more inputs isequal to the sum of the individual outputs that would beproduced by each input separately This is achieved with aconvolution because Eq (2) is a linear weighted sum of theinput pixels Furthermore the filter is shift-invariant if theweights do not change as the window moves across the im-age (Schowengerdt 2007)

Directional filters are used to enhance specific linear trendsin an image A typical directional filter consists of two ker-nels each of which is an array of three by three pixels Theleft kernel is multiplied by the cosA where A is the anglerelative to the north of the linear direction to be enhancedThe right kernel is multiplied by sinA Angles in the north-eastern quadrant are considered negative angles in the north-western quadrant are positive The filter can be demonstratedby applying it to a dataset in which a brighter area is sepa-rated from a darker area along a total lineament that trendsnortheast (A=minus45) (Carr 1995 Sabins 1996 Vincent1997 Research Systems Inc 2008)

In this study for identifying linear features in particulardirections and edge enhancement in the spatial domain di-rectional convolution filters were applied to the median re-sultant image Directional filtering technique is a straightfor-ward method for extracting edges in the spatial domain thatapproximate the first derivative between two adjacent pix-els The algorithm produces the first difference of the imageinput in the horizontal vertical and diagonal directions (Har-alick et al 1987 Carr 1995 Sabins 1996 Vincent 1997Jensen 2005) As a result many additional edges of diverseorientations are enhanced (Richard and Jia 1999) The edgesappear as a plastic shaded-relief format (embossing) in theimage because of the 3-D impression conveyed by the filter-ing (Schowengerdt 2007) The directional filter is used for

producing artificial effects suggesting tectonically controlledlinear features (Pour and Hashim 2015a b)

Directional filters were used to enhance specific lineartrends in the median resultant image Four principal direc-tional filters NndashS EndashW NEndashSW and NWndashSE with 5times 5and 7times 7 kernel sizes were applied to ScanSAR and finescenes respectively A 5times 5 kernel matrix was selected forScanSAR scene to enhance roughsmooth and semi-roughfeatures at a regional scale in the northern part of PeninsularMalaysia A 7times 7 kernel matrix was applied to fine scenesfor enhancing semi-smooth and smoothrough features at dis-trict scale in the Kelantan state (Chavez and Bauer 1982Jensen 2005) Directional filter angles were adjusted as NndashS(0) EndashW (90) NEndashSW (45) and NWndashSE (135) North(up) is 0 and the other angles are measured in a counter-clockwise direction An Image Add Back value of 60 wasentered

Landslide is triggered by two forces namely ldquoprecondi-tion factorsrdquo that govern the slope stability and ldquopreparatoryand triggering factorsrdquo (Komac 2006) In this study land-slides caused by heavy flooding episodes in the Kelantanstate in December 2014 were reported as (i) a large-scalelandslide (gt 300 m2) that could be detected by satellite re-mote sensing data and (ii) a small-scale landslide (lt 300 m2)that measured and determined during a scientific expeditionin the Kelantan River basin between 20 and 25 January 2015Data collection in landslide-affected zones and flooded ar-eas has been done by Global Positioning System (GPS) sur-veying GPS surveying was carried out in the Tanah MerahMachang Jeli Kuala Krai and Gua Musang districts inthe Kelantan River basin using a Garminreg MONTERRAreg

with an average accuracy 5 m in 453 landslide-affected loca-tion points These locations were recorded in a forest rub-ber trees bushes (degraded forest) mixed crops oil palmcleared land urban area and agriculture lands

In the Landsat-8 OLI images landslides were determinedby tracking changes in vegetation pixel data using Landsat-8images acquired before and after flooding The normalizeddifference vegetation index (NDVI) was calculated using in-frared and red bands of Landsat-8 datasets

NDVIOLI = (Band5OLIminusBand4OLI)(Band5OLI+Band4OLI) (4)

where Band5OLI is reflectance unit of near-infrared-bandLandsat-8 OLI and Band4OLI is reflectance unit of red-bandLandsat-8 OLI

In this investigation the analytical hierarchy pro-cess (AHP) approach was used for landslide susceptibilitymapping (LSM) The AHP is a multi-objective multi-criteriadecision-making approach that enables the user to arrive at ascale of preferences drawn from a set of alternatives (Saaty1980 Saaty and Vargas 1991 2001) AHP involves build-ing a hierarchy of decision elements (factors) and then mak-ing comparisons between possible pairs in a matrix to givea weight for each element and also a consistency ratio It is

Table 1 Pair-wise comparison matrix for each input of AHP

Factor Slope Aspect Soil Lithology NDVI Land Precipitation Distance to Distance to Distance tocover road river fault

Slope 0032 0031 0029 0018 0030 0034 0023 0060 0014 0014Aspect 0032 0031 0036 0070 0030 0021 0015 0050 0014 0014Soil 0161 0123 0145 0175 0075 0206 0226 0150 0170 0211Lithology 0065 0015 0029 0035 0050 0026 0015 0060 0019 0021NDVI 0161 0154 0291 0105 0150 0206 0090 0100 0226 0211Land cover 0097 0154 0073 0140 0075 0103 0135 0100 0170 0169Precipitation 0065 0092 0029 0105 0075 0034 0045 0060 0019 0021Distance to road 0161 0185 0291 0175 0499 0309 0226 0300 0283 0211Distance to river 0129 0123 0048 0105 0037 0034 0135 0060 0057 0085Distance to fault 0097 0092 0029 0070 0030 0026 0090 0060 0028 0042

Sum 1000 1000 1000 1000 1000 1000 1000 1000 1000 1000

based on three principles decomposition comparative judg-ment and synthesis of priorities (Malczewski 1999) To ap-ply this approach it is necessary to break a complex unstruc-tured problem down into its component factors arrange thesefactors in a hierarchic order assign numerical values to sub-jective judgments on the relative importance of each factorand synthesize the judgments to determine the priorities tobe assigned to these factors (Saaty and Vargas 2001 Yalcin2008 Yalcin et al 2011 Choi et al 2012)

In this investigation to apply the AHP approach 10 fac-tors ndash slope aspect soil lithology NDVI land cover dis-tance to drainage precipitation distance to fault and distanceto the road ndash were extracted from Landsat-8 images PAL-SAR data and fieldwork as shown in Fig 3 Comparisons be-tween the 10 factors show that slope land use type distanceto the fault line and precipitation intensity are more impor-tant than the others A scale of numbers is crucial to indicatehow many times more important or dominant one element isover another element with respect to the criterion or prop-erty to which they are compared The relationships betweenthe detected landslide locations and these 10 related factorswere calculated based on AHP procedure and prior knowl-edge (Tables 1 and 2) The distance between detailed faultline derived from PALSAR-2 data and all landslides pointwas computed in ArcMap 103 software using Euclidean dis-tance

Among selected 10 factors as the input of the AHP ap-proach it seems that the excessive rainfall during Decem-ber 2014 flood episode in the Kelantan is more significantthan than the other nine parameters Thus rainfall analy-sis was carried out based on daily precipitation from about80 stations before and during December 2014 flood episodein the Kelantan state Table 3 shows rainfall amount for eachdistrict in Kelantan state The rainfall analyses for 452 land-slide events were carried out to define the threshold rain-fall intensity triggers landslides to occur All data from raingauge were extracted to obtain the average rainfall intensity

Table 2 Factor weights for each input of AHP

Factors Weight

F1 precipitation 0141F2 slope 0123F3 soil 0121F4 aspect 0102F5 lithology 0097F6 land cover 0086F7 distance to road 0084F8 distance to drainage 0081F9 NDVI 0073F10 distance to fault 0062

of the whole year of 2014 and to compute the anomaly ofrainfall amount during the flood episode The pre-occurrenceof the next landslides could be predicted by determination ofa threshold value of the rainfall intensity in Kelantan that trig-gers the slope failure to occur Most landslide events reportedduring and after the flood episode occurred during the rainyseason The relationship between absolute antecedent rainfallprecipitation and landslide occurrence was studied by usinga general methodology for statistical analysis of landslide in-duced by rainfall based on the reconstruction of absolute an-tecedent precipitations Equation (2) was used to obtain thespatial distribution of the landslide susceptibility map of eachclass during and before flood episode in Kelantan 2014

LSM=nsumi=1

(RiWi) (5)

where Ri represents the rating class for each and every layerwhere Wi is the weight for the landslide factors After get-ting the result of LSM it is then be divided into five differentclasses labeled as low risk (LS) fair risk (FS) medium sus-ceptibility (MS) high risk (HS) and very high risk (VHS)based on the natural break method

Table 3 Rainfall amount for each district in Kelantan

District Width Rainfall Percentage Annual Percentage(km2) amount of rainfall rainfall of annual

during in flood amount rainfallflood episode (mm) ()

episode ()(mm)

Tumpat 16950 155870 1044 362128 640Kota Bharu 11564 305400 2045 176410 1075Pasir Puteh 43380 478350 3203 143472 1355Tanah Merah 88414 552730 3701 199658 1576Jeli 133048 513220 3437 34818 962Kuala Krai 232900 309110 2070 202365 964Gua Musang 821430 958410 6419 92203 2431Machang 54626 146900 984 160500 233Pasir Mas 13900 240220 1601 140786 942

4 Results and discussion

A wide-swath ScanSAR observation mode of PALSAR-2was used for comprehensive analysis of major geologicalstructures which shows mega-geomorphology and mega-lineaments in the Kelantan state Figure 4 shows an RGBcolor composite of HH polarization channel in red HV po-larization channel in green and HH+HV polarization chan-nel in blue for the ScanSAR resultant image The RGBcolor composite yields an image with great structural detailsand geomorphological information The different colors inthe image indicate different backscattering signals from theground The Main Range granite located in the western partof the image and the Boundary Range granite on the east-ern borders of Kelantan state appear as light green to greenin color which shows the regions with high altitude in thescene (Fig 4) Major transcrustal lineaments such the BRSZand Lebir Fault Zone are also detected Quaternary depositsTriassic marine siliciclastics volcaniclastics sandstone andlimestone are manifested as pink to purple tones consistingof lands with low elevation (Fig 4) Lakes and main riversystems are portrayed as blue to dark blue in the image Infact the lines range from blue to dark blue on the image arefaults and fracture zones occupied by streams The river pat-tern is one of the most important factors contributing to thelineament because it reflects the nature of the existing frac-ture system (Suzen and Toprak 1998)

Figure 5 shows a ScanSAR HV polarization image thatis superimposed on a general topography map of the Kelan-tan state (Department of Minerals and Geoscience Malaysia2003) It is evident that the morphology of the study area islargely controlled by rock type and structure High-elevationareas (500ndash100 and lt 1000 m) in the Kelantan state aremountainous areas associated with Main Range granite (inthe west) and Boundary Range granite (in the east) Hilly

plain and coastal areas with an altitude between 500 and 50 mare associated with sedimentary rocks

Figure 6 shows the resultant image map for NndashS NEndashSW and NWndashSE (R 0 G 45 B 135) directional fil-ters A major change in deformation style is obvious fromthe west to the east in Fig 6 Structural analysis reveals fourdistinct parts from the west to the east including (i) west-ern part of the scene by ductile fabrics (ii) western of theBRSZ affected mainly by brittle deformation (iii) ductilendashbrittle deformation between the BRSZ and Lebir Fault Zoneand (iv) brittlendashductile fabrics between Lebir Fault Zone andeastern coastal line Lineament occurrence in Fig 6 is mainlylinked to the NndashS trending of the BRSZ and Lebir FaultZone Generally major faults are strike-slip with both dex-tral and sinistral movements which trend NndashS and NWndashSENWndashSE-trending strike-slip faults moved sinistrally in theLebir Fault Zone The sinistral movement along the LebirFault Zone is responsible for the formation of folding andreverse faulting adjacent to the fault and surrounding areaThese structures characterized transpressive tectonic regimein Peninsular Malaysia (Richter et al 1999 Harun 2002)

The collision zone and compressional structures appearclearly in the west of the BRSZ in Main Range granite(Fig 6) Deformation in this region shows the shorteningzone oriented parallel to the BRSZ Several faults joints andfractures represent brittle deformation events in the regionthat mostly strike NWndashSE Generally most of the short linea-ments are clustered in the collision zone Ductile deformationin the western margin of the image (Fig 6) includes uprightasymmetrical mega-folds with axial surfaces oriented WndashEHence the contractional strain direction affected by this do-main is from NEndashSW followed by dextral shearing The de-formed area zone between the BRSZ and Lebir Fault Zonerepresents fault systems and folding area NWndashSE strikingstrike-slip faults and NEndashSW normal faults are dominatedbrittle structural elements within this domain Ductile defor-

Figure 3 List of extracted factors from Landsat-8 PALSAR and fieldwork (a) slope (degree) (b) aspect (degree) (c) lithology types(d) landuse type (e) precipitation (mm) (f) NDVI (g) distance to main road (h) distance to main river (m) and (i) distance to fault line (km)

mation is demonstrated by several open upright folds whichhave WndashE to NEndashSW axial plans (Fig 6) Brittlendashductilefabrics in the eastern part of the image between Lebir FaultZone and eastern coastal line illustrate curved shear zone thatoccupied several NndashS and NWndashSE striking faults fracturesand joints A mega-concentric fold surrounds the shear zonewith a WE striking axial surface According to the orienta-tion of the lineaments sinistral movement along the LebirFault Zone has generated the tectonic features Some NndashS

and NEndashSW-trending normal faults and small curvatures arealso identifiable near the eastern coastal line (Fig 6)

Figure 7 shows a fine-mode merged image of the northernand southern parts of the Kelantan state In this figure HHpolarization channel was assigned to be red HV polarizationchannel to be green and HH+HV polarization channels tobe blue for generating an RGB image for the study area Thiscolor combination produced an image map contains impor-tant information related to water bodies wetlands and ge-

Figure 4 RGB color combination (HH HV and HH+HV polariza-tion channels) of PALSAR-2 Scan SAR scene covering the northernpart of the Peninsular Malaysia Black rectangle shows the selectedarea for detailed analysis using fine-mode images of the northernand southern parts of the Kelantan state in this study

Figure 5 ScanSAR HV polarization image of the Kelantan statesuperimposed by general topography map

ological structural features River systems and lakes appearblack especially in the northeastern part of the image andwetlands as mauve (Fig 7) Smooth surfaces such as calmwater bodies appear dark in SAR images due to the reflectionof the radar signal Hence no returning radar signal could bedetected in receiving antenna (Thurmond et al 2006 Pour

Figure 6 ScanSAR image map derived from NndashS (0) NEndashSW(45) and NWndashSE (135) directional filters for the northern partof the Peninsular Malaysia The explanation for the figure dashedblack lines are major faults dot-dashed lines are folds and curvilin-ear red lines are faults and fractures

and Hashim 2015a) Besides the best soil moisturewetnessinformation is achievable from L-band microwaves becauseit comes out from deeper soils In fact vegetation is almosttransparent and roughness effects are negligible using L-bandmicrowaves for soil wetness mappingmonitoring (Jacksonand Schmugge 1989 Lacava et al 2005) Soil moisture isone the most important factors in hydrogeological hazardsespecially since the soil response is affected by its status ofsaturation Accordingly the identification of wetlands is veryimportant for flood forecasting and prevention in the state ofKelantan Wetlands are more distributed in the northern partof the study area (Fig 7) A vast wetland (dark mauve) isrepresented clearly in the central north segment of the im-age Several wetlands such as light to dark mauve are ob-servable in central-east south and southwestern part of theimage (Fig 7)

The RGB image map (Fig 7) was compared with the ge-ological map of the Kelantan state It is evident that most ofthe detected wetlands are associated with sedimentary andmetamorphic rocks A few of the wetlands are located in thegranite bedrock areas in the west (the Main Range Granite)and east (the Boundary Range granite) of the state of Kelan-tan The granite bedrock areas are all characterized by dis-sected hilly to mountainous terrain that gives rise to isolatedhighlands and mountain ranges (Fig 7) These granitic areasrepresent the outcrops of a number of granite batholiths thatare generally elongated along a NndashS to NNWndashSSE direction(Bignell and Snelling 1977)

Figure 7 RGB color composite image of HH HV and HH+HVpolarization channels derived from fine-mode merged image of thenorthern and southern parts of the Kelantan state

LSMs for before and after flood episodes were producedusing related factors contributing to landslides event in theKelantan state (Fig 8) Overall the number of landslideswas 452 About 367 occurred in the area of rubber planta-tion area while 343 occurred in forest areas Location oflandslides occurrences at different land usecover type in Ke-lantan is given in Fig 9 Forest is the dominant land usecoverin Kelantan besides rubber and palm oil

The amplitude of the radar signal is highly sensitive to thephysical properties of the ground surface which producesthe brightness of the surface to vary much more than op-tical or infrared images (Robinson et al 1999) Hence acombination of different polarization channels of fine-modePALSAR-2 data contains considerable information for rockdiscrimination In this study an RGB color composite was

produced by assigning HV+HH polarization channels tored HVHH polarization channels to green and HV polar-ization channel to blue for detailed rock discrimination inthe Kelantan state Figure 10 represents the RGB merged im-age for the northern and southern parts of the Kelantan stateGranitic rocks are manifested as light green (dark green inhigh-altitude area probably due to layover effect on SAR im-age Gelautz et al 1998 Franceschetti and Lanari 1999)while sedimentary and metamorphic rocks appear as a va-riety of tones such as brown light brown dark green lightblue and light red in the image (Fig 10) In comparison withthe geological map of the study area it seems that brownand dark green hues are metamorphic rocks and light brownlight blue and light red are sedimentary rocks (clay shaleconglomerate sandstone and limestone) River systems andlakes are observable clearly as black meandering streams andwater bodies especially in the northern segment of the region(Fig 10)

Four directionally filtered images of fine-mode observa-tion which contain enhanced information for a set of linea-ments in NndashS EndashW NEndashSW and NWndashSE directions wereused for lineament mapping in the Kelantan state Figure 11ashows a structural map for the Kelantan state which is de-rived from the resultant image of directional filtering to HVpolarization channel It should be noted that the locations ofwetlands are also portrayed in this figure The most importantstructural features in the image map are fault zones river sys-tems and drainage lines patterns (Fig 11a) Within the studyarea two large fault zones are presented BRSZ in the north-west and Lebir Fault Zone in the southeast The NndashS NEndashSW and NNEndashSSW lineament trends are commonly domi-nant in the image map The dominant lineaments tend to runin the NndashS direction which is mainly linked to the N-S trend-ing of the BRSZ (in the west) and Lebir Fault Zone (in theeast) Additionally few short NWndashSE-trending lineamentsare detected in the western and eastern parts of the study area(Fig 11a) The NndashS and NEndashSW striking system distributedin the southeastern segment of the region is particularly re-lated to Lebir Fault Zone The pattern of the lineament mapin the northwestern part of the image map displays the oc-currence of BRSZ which contains lineaments in NndashS andNNEndashSSW directions Most of the short and smaller faultsfollow the NndashS NEndashSW and NNEndashSSW trends as the majorfault systems in the Kelantan state It seems that the NWndashSE faults are the youngest faults in the study area due to thelow frequency of lineaments in this trend The major NndashS-trending faults are interpreted as oldest structures in Peninsu-lar Malaysia and related to the amalgamation of Gondwana-derived terranes during the PermianndashTriassic and NEndashSWand NWndashSE-trending faults are interpreted to be Late Trias-sic to Jurassic in age (Metcalfe 2013b)

The occurrence and concentration of geological structuralfeatures in any region are related to rock type thickness ofuppermost weathered mantle and brittleness of rocks (Harriset al 1960) In the Kelantan state high concentrations of lin-

Figure 8 LSM before (a) and after (b) 2014 flooding episode in Kelantan state

Figure 9 Land usecover map of the Kelantan state

eaments are associated with granitic rocks in higher terrainsand low concentrations of lineaments are usually associatedwith lower terrains of metamorphic and sedimentary rocksTjia and Harun (1985) analyzed regional structures of Penin-sular Malaysia and reported that many lineaments are verywell displayed in areas underlain by granite whereas in otherareas they are poorly shown Moreover Akhir (2004) iden-tified that high lineament concentrations in the Upper PerakValley Perak state Peninsular Malaysia are closely relatedto higher terrain which is formed by more-resistant rockssuch as granite and sandstone whereas low lineament con-centrations are associated with lower terrain which is mainlyformed by less-resistant rock types such as shale and slateAs well volcanic rocks in the Upper Perak Valley encom-pass a low lineament concentration (Akhir 2004)

The rivers in the study area are structurally controlledThey display zigzag patterns due to the presence of fracturesjoints and faults with changes in orientation (Fig 11a) Thedrainage system in the Kelantan River basin shows dendriticsub-dendritic and rectangular patterns (Fig 11a) It is evi-dent that the drainage pattern is apparently being controlledby structure and lithology in the study area However severalfactors such as topography soil type bedrock type climateand vegetation cover influence input output and transportof sediment and water in a drainage system The geologicalstructures and lithologic variation have given rise to differentdrainage patterns (Summerfield 1991) For instance a den-dritic and sub-dendritic pattern with a large number of trib-utaries is typical of drainage in areas of impermeable crys-talline rock such as gneiss andor sediment of uniform resis-tance (horizontal strata) The pattern is characteristic of es-sentially flat-lying andor relatively homogeneous rocks andimpervious soils with the lack of structural control A rectan-gular pattern is usually caused by jointing or faulting of theunderlying bedrock It is usually associated with massive in-trusive igneous and metamorphic rocks (Summerfield 1991)Therefore it is assumed that the area with the dendritic andsub-dendritic pattern is subjected to hydrogeological hazardssuch as flooding because of low infiltration runoff Rectangu-lar drainage pattern is a susceptible zone that could be easilyaffected by landslide due to the slope of the land lithostruc-tural conditions and speed of runoff In the Kelantan Riverbasin most of the dendritic and sub-dendritic drainage pat-terns are detected in central part of the river basin (Fig 11a)which consists of sedimentary rocks However most of therectangular drainage patterns are associated with igneous andmetamorphic rocks in the western part of the Kelantan Riverbasin (Fig 11a) A structural and topographical feature mapof the Kelantan River basin is shown in Fig 11b It is evidentthat most of the dendritic and sub-dendritic drainage patternsare located in lowlands and the rectangular drainage patterndominates in highlands in the Kelantan River basin

Figure 10 RGB color composite image of HV+HH HVHH andHV polarization channels derived from fine-mode merged image ofthe northern and southern parts of the Kelantan state

Precipitation influences the occurrence of landslides morethan the distance to fault which can be seen as having thelowest impact on landslide susceptibility According to thelandslide susceptibility map 65 of the whole area of Ke-lantan is found to be under the category of a low landslidesusceptibility zone while 35 class located within high sus-ceptibility zone Table 4 shows information on statistical test-ing of landslides due to rainfall amount for the Kelantan stateduring the December 2014 flood episode Two assumptionswere used for statistical testing of the landslide in this anal-ysis (a) the value of rainfall is one of the major factors con-tributing to the landslide tragedy in Kelantan which is as-sumed to be H0 and (b) the value of rainfall is not the ma-jor factor contributing to the landslide tragedy in Kelantanwhich is determined to be HA The ANOVA test shows thatF calculatedltF significance thus the null hypothesis is ac-cepted The test statistic is the Fstatistic value of 00938 Using

α of 005 F005197= 3939 Since the Fstatistic value is muchsmaller than the critical value (p value= 07601) thus thenull hypothesis H0 is accepted at equal rainfall means andconclude that rainfall is one of the major factors contributingto the landslide tragedy in the Kelantan

Before the flood the mean amount of rainfall is muchlower than mean of rainfall during flood episode where fewlandslide points during 2014 fall under the mean of nor-mal rainfall During the flood episode as the mean of rain-fall is very high many landslides points fall higher than themean value It is evident that a landslide is triggered whenthe amount of average monthly rainfall reaches plusmn400 andplusmn800 mm for daily rainfall

Further analysis has been carried out after the detectionof landslides in Kelantan state and landslide susceptibilitymapping through the GIS-based statistical model Since thelandslides occur mainly in the forest information about whatdensity of trees will cause slope failure is crucial any mitigat-ing action plan to be undertaken in the future by the local au-thorities Thus ground surface displacement for all identifiedlandslides occurrences and relation to forest density shouldbe carried out To know the impact of a landslide event thatshifts the variety of different density of forest in an affectedarea the area of the landslide before and after a flood episodemust be analyzed by associating its size with forest densityStock plots indicate that before a rainy season (Fig 12a)many landslides in Kelantan tend to occur in moderate for-est density (close to 050 NDVI value) while landslides aremore likely to occur at a location with a slightly lower NDVIvalue (almost 040) during the flood episode (Fig 12b) Incontrast large landslides are not highly dependent on howdense the forest is but are more likely due to other factorssuch as slope failure geological structures (faults and frac-tures) and heavy rainfall

Field observations were conducted to compare the de-tected high potential risk and susceptible zones with highlydamaged areas in recent flooding events in the KelantanRiver basin The analysis of field investigation data indicatesthat many of flooded areas were associated with high po-tential risk zones for hydrogeological hazards such as wet-lands urban areas floodplain scroll meander bend and den-dritic and sub-dendritic drainage patterns Most of the hy-drogeological hazards zones are located in flat topographicregions Flat topography with impervious soilssurface haslow infiltration runoff which yields more storm runoff dur-ing heavy monsoon rainfall compared to other regions Fig-ure 13a and b show meander bends and floodplain scrollzones in the Kelantan River basin which were flooded areasduring the 2015 flooding event

Numerous landslide-affected points were recorded in thehigh-altitude segment of the south and southwestern part ofthe Kelantan state As mentioned above high drainage den-sity of the rectangular system is governed in this domainThe drainage density affects runoff in that a high drainagedensity drains runoff water rapidly decreases the lag time

Figure 11 (a) Structural lineament map of the Kelantan state derived from directional filtering to HV polarization channel (b) Structuraland topographical feature map of the Kelantan River basin

Figure 12 (a) Stock plot of forest density versus area of landslide before the flood (b) The stock plot of forest density versus area of alandslide during flood episode

and increases the peak of the hydrograph Consequently theslope of the land in the south and southwestern regions in-creases the speed and extent of water and sediment trans-portation to the Kelantan River basin during heavy mon-soon rainfall Most of the landslide-affected points were lo-cated in the topographic slope of metamorphic and qua-ternary rock units However some landslides occurred infracture zones of weathered igneous rock units Some largelandslide-affected zones were recorded in the intersection of

longitude and latitude between 508minus02primeprime N 10158prime53primeprime E508prime14primeprime N 10159prime06primeprime E 508prime24primeprime N 10159prime21primeprime E and509prime03primeprime N 10159prime41primeprime E These landslide points were as-sociated with NndashS NNEndashSSW and NEndashSW-trending faultzones Figure 13c and d show large landslide-affected zonesin the southwestern part of the Kelantan state

Table 4 Information on statistical testing of landslide due to rainfall amount

Regression statistics

Multiple R 00311R2 00010

Adjusted R2minus00093

Standard error 1414588Observations 99

Degree of Sum of Mean of Fstatistic Fsignificancefreedom squares sum square (p value)

Regression 1 187659 187659 00938 07601Residual 97 1 941 02866 20 01060

Sum 98 1 942 90526

Figure 13 (a) Meander bend (b) Floodplain scroll zone in the Kelantan River basin (c d) Large landslide-affected zones in the southwesternpart of the Kelantan state

5 Conclusions

Results of this investigation indicate that Landsat-8 andALOS-2 have proven to provide successful advanced remotesensing satellite data for disaster monitoring in tropical envi-ronments Analysis of the Landsat-8 and ALOS-2 data pro-vided significant information for identifying high potentialrisk and susceptible zones for natural hazards of geologi-cal origin in the Kelantan River basin Malaysia Wetlandsfloodplain scroll meander bend dendritic and sub-dendriticdrainage patterns and urban areas were identified as high po-

tential risk zones for hydrogeological hazards Landslide re-currence regions were detected in a high-altitude segment ofthe south and southwestern part of the Kelantan state whichis dominated with a high density of rectangular drainage pat-tern and topographic slope of metamorphic and quaternaryrock units Some of the large landslide zones were associatedwith NndashS NNEndashSSW and NEndashSW-trending fault systemsStructural and topographical geology maps were producedfor the Kelantan River basin that could be used to facilitatethe planning of geohazards mitigation In conclusion the re-sults of this investigation have great potential assistance in

terms of a total solution to flood disaster management in theKelantan River basin by providing an important source of in-formation to assess the potential for many natural hazards ofgeological origin

Data availability This investigation used original satellite datawhich are discussed in Sect 3

Competing interests The authors declare that they have no conflictof interest

Acknowledgements This study was conducted as a part ofthe Transdisciplinary Research Grant Scheme (TRGS) re-search project ldquoDetecting mapping and characterization oflandslides and landslide assessment in Kelantan River basinduring December 2014 flooding using Satellite SAR datardquo votenos RJ13000078094L831 and RJ13000078094L837 Ministryof Higher Education (MOHE) Malaysia We are thankful tothe Universiti Teknologi Malaysia for providing the facilitiesfor this investigation We also would like to express our greatappreciation of Rosa Lasaponara and the anonymous reviewers fortheir very useful and constructive comments and suggestions forimprovement of this paper

Edited by Rosa LasaponaraReviewed by three anonymous referees

References

Abdullah A Nassr S Ghaleeb A Remote sensing and ge-ographic information system for fault segments mapping astudy from Taiz area Yemen J Geol Res 2013 201757httpsdoiorg1011552013201757 2013

Akhir J M Lineament in enhanced remote sensing images an ex-ample from the Upper Perak Valley Perak Darul Ridwan GeolSoc Malaysia Bull 48 115ndash119 2004

Amri K Mahdjoub Y and Guergour L Use of Landsat 7 ETM+for lithological and structural mapping of Wadi Afara Heouinearea (TahifetndashCentral Hoggar Algeria) Arab J Geosci 41273ndash1287 2011

Arikawa Y Osawa Y Hatooka Y Suzuki S and KankakuY Development Status of Japanese Advanced Land Ob-serving Satellite-2 in Vol 7826 Proceedings of the SPIERemote Sensing Conference Sensors Systems and Next-Generation Satellites XIV edited by Meynart R Neeck SP and Shimoda H 20ndash23 September 2010 Toulouse Francehttpsdoiorg10111712866675 2010

Bannert D The application of remote sensing to natural hazard ofgeologic orgion-experience learned from GRAS-program of Un-esco and IUGS in Vol XXXIII Part B7 International Archivesof Photogrammetry and Remote Sensing Amsterdam 2000a

Bannert D Geological application of remote sensing (GARS) anIUGSUNESCO joint program in the new millennium Episodes23 37ndash39 2000b

Bignell J D and Snelling N J Geochronology of Malayan gran-ites Overseas Geology and Mineral Resources IGS Londonp 72 1977

Blau P Japanrsquos H-IIA Rocket successfully Launches ALOS-2Radar Satellite Spaceflight 101 httpwwwspaceflight101comh-iia-f24-launch-updates-alos-2html last access 24 May 2014

Carr J R Numerical analysis for the geological science edited byCarr J R Prentice-Hall Inc Englewood Cliffs New Jerseyp 592 1995

Chavez P C and Bauer B An automatic optimum kernel-size se-lection technique for edge enhancement Remote Sens Environ12 23ndash38 1982

Choi J Oh H J Lee H J Lee C and Lee S Combining land-slide susceptibility maps obtained from frequency ratio logisticregression and artificial neural network models using ASTERimages and GIS Eng Geol 124 12ndash23 2012

Clark C D and Wilson C Spatial analysis of lineaments Com-put Geosci 20 1237ndash1258 1994

Dai F C and Lee C F Landslide characteristics and slope insta-bility modeling using GIS Lantau Island Hong Kong Geomor-phology 42 213ndash228 2002

Department of Minerals and Geoscience Malaysia Quarry Re-source Planning for the State of Kelantan Osborne and ChappelSdn Bhd Kuala Lumpur Malaysia 2003

Eagleman J R and Lin W C Remote Sensing of Soil Moisture bya 21-cm Passive Radiometer J Geophys Res 81 3660ndash36661976

Eckardt R Berger C Thiel C and Schmullius C Removal ofOptically Thick Clouds from Multi-Spectral Satellite Images Us-ing Multi-Frequency SAR Data Remote Sensing 5 2973ndash3006httpsdoiorg103390rs5062973 2013

Eliason E M and McEwen A S Adaptive Box Filters for Re-moval of Random Noise from Digital Images Photogramm EngRemote Sens 56 453ndash458 1990

ERSDAC ndash Earth Remote Sensing Data Analysis Center PALSARuserrsquos guide 1st Edn Japan Space SystemsEarth Remote Sens-ing Division Japan March 2006

Franceschetti G and Lanari R Synthetic aperture radar process-ing CRC Press Boca Raton 1999

Gelautz M Frick H Raggam J Burgstaller J and Leberl FSAR image simulation and analysis of alpine terrain ISPRS JPhotogramm Remote Sens 53 17ndash38 1998

Ghani A A Plutonism in Geology of Peninsular Malaysia editedby Hutchison C S and Tan D N K Geological Society ofMalaysia Kuala Lumpur 211ndash232 2009

Guzzetti F Carrara A Cardinali M and Reichenbach P Land-slide hazard evaluation a review of current techniques and theirapplication in multi-scale study Central Italy Geomorphology31 181ndash216 1999

Hamimi Z El-Sawy E S K El-Fakharani A Matsah M Shu-joon A and El-Shafei M K Neoproterozoic structural evo-lution of NE-trending Ad-Damm Shear Zone Arabian ShieldSaudi Arabia J Asian Earth Sci 99 51ndash63 2014

Haralick R M Sternberg S R and Zhuang X Image AnalysisUsing Mathematical Morphology IEEE T Pattern Anal MachIntel PAMI-9 532ndash550 1987

Harris J F Tailor G L and Walper J L Relation of deforma-tional fractures in sedimentary rocks to regional and local struc-ture Am Assoc Petrol Geol Bull 44 1853ndash1873 1960

Harun Z Late Mesozoic-Early Tertiary faults of PeninsularMalaysia in Geological Society of Malaysia Annual GeologicalConference 26ndash27 May 2002 Kota Bharu Kelantan Malaysia2002

Hashim M Ahmad S Johari M A M and Pour A B Au-tomatic lineament extraction in a heavily vegetated region us-ing Landsat Enhanced Thematic Mapper (ETM+) imagery AdvSpace Res 51 874ndash890 2013

Henderson F M and Lewis A J Principles and applications ofimaging radar manual of remote sensing in Vol 2 3rd EdnJohn Wiley and Sons New York 886 pp 1998

Heng G S Hoe T G and Hassan W F W Gold mineralizationand zonation in the state of Kelantan Geol Soc Malaysia Bull52 129ndash135 2006

Igarashi T ALOS Mission requirement and sensor specificationsAdv Space Res 28 127ndash131 2001

Irons J R Dwyer J L and Barsi J A The next Landsat satelliteThe Landsat Data Continuity Mission Remote Sens Environ145 154ndash172 2012

Jackson T J and Schmugge T J Passive Microwave RemoteSensing System for Soil Moisture Some Supporting ResearchIEEE T Geosci Remote 27 225ndash235 1989

Japan Aerospace Exploration Agency ALOS-2PALSAR-2Level 11152131 GeoTIFF Product Format DescriptionhttpwwweorcjaxajpALOS-2endocformathtm (lastaccess 22 December 2015) 2014

Jensen J R Introductory Digital Image Processing A remotesensing perspective 3rd Edn Pearson Prentice Hall Upper Sad-dle River NJ 276ndash287 2005

Ju J and Roy D P The availability of cloud-free Landsat ETM+data over the conterminous United States and globally RemoteSens Environ 112 1196ndash1211 2008

Kankaku Y Osawa Y Hatooka Y and Suzuki S Overviewof Advanced Land Observing Satellite-2 (ALOS-2) in Pro-ceedings of ISPRS Technical Commission VIII Symposium 9ndash12 August 2010 Kyoto Japan 2010

Komac M A landslide susceptibility model using the Analyti-cal Hierarchy Process method and multivariate statistics in pe-rialpine Slovenia Geomorphology 74 17ndash26 2006

Lacava T Greco M and Leo E Monitoring soil wetness varia-tions by means of satellite passive microwave observations thehydroptimet study case Nat Hazards Earth Syst Sci 5 583ndash592 2005

Lee J S and Jurkevich I Speckle filtering of synthetic apertureradar images a review Remote Sens Rev 8 313ndash340 1994

Malczewski J GIS and Multi-criteria Decision Analysis John Wi-ley and Sons New York 1999

Mardiana S Ahmad A and Osman K Capability ofRADARSAT data in monsoon flood monitoring GIS Develop-ment 1ndash6 httpa-a-r-sorgaarsproceedingACRS2000PapersENV00-10htm (last access July 2017) 2000

Melgani F Contextual reconstruction of cloud contaminated mul-titemporal multispectral images IEEE T Geosci Remote 44442ndash455 2006

Metcalfe I The Bentong-Raub Suture Zone J Asian Earth Sci18 691ndash712 2000

Metcalfe I Tectonic evolution of the Malay Peninsula J AsianEarth Sci 76 195ndash213 2013a

Metcalfe I Gondwana dispersion and Asian accretion Tectonicand palaeogeographic evolution of eastern Tethys J Asian EarthSci 66 1ndash33 2013b

Morelli M and Piana F Comparison between remote sensed lin-eaments and geological structure in intensively cultivated hills(Monferrato and Langhe domains NW Italy) Int J RemoteSens 27 4471ndash4493 2006

Nazaruddin D A Fadilah N S M Zulkharnain Z Omar SA S and Ibrahim M K M Geological Studies to Support theTourism Site A Case Study in the Rafflesia Trail Near KampungJedip Lojing Highlands Kelantan Malaysia Int J Geosci 5835ndash851 2014

Plank S Twele A and Martinis S Landslide mappingin vegetated areas using change detection based on opti-cal and polarimetric SAR data Remote Sensing 8 307httpsdoiorg103390rs8040307 2016

Pour B A and Hashim M Structural geology mapping us-ing PALSAR data in the Bau gold mining district SarawakMalaysia Adv Space Res 54 644ndash654 2014

Pour B A and Hashim M Structural mapping using PALSARdata in the Central Gold Belt Peninsular Malaysia Ore GeolRev 64 13ndash22 2015a

Pour B A and Hashim M Integrating PALSAR and ASTER datafor mineral deposits exploration in tropical environments a casestudy from Central Belt Peninsular Malaysia Int J Image DataFusion 6 170ndash188 2015b

Pour B A Hashim M and van Genderen J Detection of hy-drothermal alteration zones in a tropical region using satellite re-mote sensing data Bau gold field Sarawak Malaysia Ore GeolRev 54 181ndash196 2013

Pour B A Hashim M and Marghany M Exploration ofgold mineralization in a tropical region using Earth Observing-1 (EO1) and JERS-1 SAR data a case study from Bau gold fieldSarawak Malaysia Arab J Geosci 7 2393ndash2406 2014

Pradhan B Flood susceptible mapping and risk area delineationusing logistic regression GIS and remote sensing J Spat Hy-drol 9 2009

Pradhan B and Youssef A M A 100-year maximum floodsusceptibility mapping using integrated hydrological and hy-drodynamic models Kelantan River Corridor Malaysia JFlood Risk Manage 4 189ndash202 httpsdoiorg101111j1753-318X201101103x 2011

Pradhan B Shafiee M and Pirasteh S Maximum flood pronearea mapping using RADARSAT images and GIS Kelantan riverbasin Int J Geoinform 5 11ndash23 2009

Raharimahefa T and Kusky T M Stuctural and remote sensingstudies of the southern Betsimisaraka Suture Madagascar Gond-wana Res 10 186ndash197 2007

Raharimahefa T and Kusky T M Structural and remote sensinganalysis of the Betsimisaraka Suture in northeastern MadagascarGondwana Res 15 14ndash27 2009

Rahman C A and Mohamed K R Pemetaan Awalan SumberWarisan Geologi Negeri Kelantan in Geological Heritage ofMalaysia (Geoheritage Mapping and Geosite Characterization)edited by Komoo I Tjia H D and Leman M S LESTARIUMK Bangi 2001

Raj J K Geomorphology in Geology of Peninsular Malaysiaedited by Hutchison C S and Tan D N K Geological Societyof Malaysia Kuala Lumpur 5ndash29 2009

Ramli M F Tripathi N K Yusof N Shafri H Z M and Rah-man Z A Lineament mapping in a tropical environment usingLandsat imagery Int J Remote Sens 30 6277ndash6300 2009

Razak K A Santangelo M Van Westen C J Straatsma M Wand de Jong S M Generating an optimal DTM from airbornelaser scanning data for landslide mapping in a tropical forest en-vironment Geomorphology 190 112ndash125 2013

Research Systems Inc ENVI Tutorials Research Systems IncBoulder CO 2009

Richard J A and Jia X Remote Sensing Digital Image AnalysisSpringer-Verlag New York p 363 1999

Richter B Schmidtke E Fuller M Harbury N and SamsudinA R Palaeomagnetism of Peninsular Malaysia J Asian EarthSci 17 477ndash519 1999

Robinson C Kusky T El-Baz F and El-Etr H Using RadarData to Assess Structural Controls from Variable Channel Mor-phology Examples from the Eastern Sahara in Proceeding ofthe Thirteen International Conference on Applied Geologic Re-mote Sensing 1ndash3 March 1999 Vancouver Canada 1999

Rosenqvist A Shimada M and Chapman B An overview of theJERS-1 SAR Global Boreal Forest Mapping (GBFM) projectin Vol 2 Proceedings of Geoscience and Remote Sensing Sym-posium IGARRS 04 20ndash24 September 2004 Anchorage AKUSA 1033ndash1036 2004

Roy D P Wulder M A Loveland T A et al Landsat-8 Sci-ence and product vision for terrestrial global change researchRemote Sens Environ 145 154ndash172 2014

Russ J C The Image Processing Handbook CRC Press Boca Ra-ton FL p 744 2002

Saaty T L The Analytical Hierarchy Process McGraw Hill NewYork 1980

Saaty T L and Vargas G L Prediction Projection and Forecast-ing Kluwer Academic Publishers Dordrecht 1991

Saaty T L and Vargas G L Models Methods Concepts and Ap-plications of the Analytic Hierarchy Process Kluwer AcademicPublisher Boston 2001

Sabins F F Remote sensing Principal and Interpretation3rd Edn W H Freeman and Co Long Grove Illinois USA1996

Schowengerdt R A Remote sensing models and methods for im-age processing 3rd Edn Academic Press Elsevier BurlingtonMA 229ndash243 2007

Schwartz M O Rajah S S Askury A K Putthapiban P andDjaswadi S The Southeast Asian Tin Belt Earth Sci Rev 3895ndash293 1995

Shahabi H and Hashim M Landslide susceptibility map-ping using GIS-based statistical models and remote sens-ing data in tropical environment Scient Rep 5 9899httpsdoiorg101038srep09899 2015

Sheng Y and Xia Z G A comprehensive evaluation of filters forradar speckle suppression in Vol 3 Remote sensing for sus-tainable future Geoscience and Remote Sensing SymposiumIGARSS96 Lincoln NE USA 1559ndash1561 1996

Shimada M ALOS-2 science program in Proceedings ofIGARSS (IEEE international Geoscience and Remote Sensingsymposium) 21ndash26 July 2013 Melbourne Australia 2013

Shimada M and Isoguchi O JERS-1 SAR mosaics of South-EastAsia using calibrated path images Int J Remote Sens 231507ndash1526 2002

Shimada M Watanabe M Motooka T Kankaku Y and SuzukiS Calibration and validation of the PALSAR-2 in Proceedingsof the IGARSS (International Geoscience and Remote SensingSymposium) 2015 26ndash31 July 2015 Milan Italy 2015

Summerfield M A Global Geomorphology An Introduction to theStudy of Landforms in Part IV chap 16 John Wiley Inc Long-man New York 405ndash430 1991

Suzen M L and Doyuran V Data driven bivariate landslide sus-ceptibility assessment using geographical information systemsa method and application to Asarsuyu catchment Turkey EngGeol 71 303ndash321 2004

Suzen M L and Toprak V Filtering of satellite images in geolog-ical lineament anlyses an application to a fault zone in CentralTurkey Int J Remote Sens 19 1101ndash1114 1998

Suzuki S Advanced Land Observing Satellite-2 ldquoDAICHI-2rdquo (ALOS-2) ndash Mission talk by team leaders JAXAhttpglobaljaxajpprojectssatalos2leadershtml last accessMay 2014

Suzuki S Kankaku Y Imai H and Osawa Y Overview ofALOS-2 and ALOS-3 in Vol 8528 Proceedings of SPIE EarthObserving Missions and Sensors Development Implementa-tion and Characterization II 29 October 2012 Kyoto Japanhttpsdoiorg10111712979184 2012

Sveinsson J R and Benediktsson J A Speckle reduction andenhancement of SAR image in the wavelet domain in Vol 1Remote sensing for sustainable future Geoscience and RemoteSensing Symposium IGARSS96 Lincoln NE USA 63ndash661996

Tan B K ldquoSuture Zonerdquo in peninsular Malaysia and ThailandImplications for paleotectonic reconstruction of Southeast AsiaJ S Asian Earth Sci 13 243ndash249 1996

Tehrany M S Pradhan B and Jebur M N Spatial prediction offlood susceptible areas using rule based decision tree (DT) anda novel ensemble bivariate and multivariate statistical models inGIS J Hydrol 504 69ndash79 2013

Thome K Palluconi F Takashima T and Masuda K Atmo-spheric Correction of ASTER IEEE T Geosci Remote 361119ndash1211 1998

Thurmond A K Abdelsalam M G and Thurmond J BOptical-radar-DEM remote sensing data integration for geologi-cal mapping in the Afar Depression Ethiopia J Afr Earth Sci44 119ndash134 2006

Tjia H D Major faults of Peninsular Malaysia on remotely sensedimages Sains Malaysiana 18 101ndash114 1989

Tjia H D and Harun Z Regional structures of PeninsularMalaysia Sains Malaysiana 14 95ndash107 1985

Tripathi N K and Gokhale V G K Directional morphologicalimage transforms for lineament extraction from remotely sensedimages Int J Remote Sens 21 3281ndash3292 2000

Unjah T Komoo I and Mohamad H Pengenalpastian SumberWarisan Geologi di Negeri Kelantan in Geological Heritage ofMalaysia (Geoheritage Mapping and Geosite Characterization)edited by Komoo I Tjia H D and Leman M S LESTARIUMK Bangi 2001

van der Pluijm B A and Marshak S Earth Structure ndash An Intro-duction to Structural Geology and Tectonics WCBndashMcGrawndashHill New York 1997

Vincent RK 1997 Fundamentals of Geological and Environmen-tal Remote Sensing Prentice Hall New Jersey pp 366

Woodhouse I H Introduction to microwave remote sensingCRC Press Taylor amp Francis Group Boca Raton 2006

Yalcin A GIS-based landslide susceptibility mapping using an-alytical hierarchy process and bivariate statistics in Ardesen(Turkey) Comparisons of results and confirmations Catena 721ndash12 2008

Yalcin A Reis S Aydinoglu A C and Yomralioglu T A GIS-based comparative study of frequency ratio analytical hierarchyprocess bivariate statistics and logistics regression methods forlandslide susceptibility mapping in Trabzon NE Turkey Catena85 274ndash287 2011

Yamamoto T Kawano I Iwata T Arikawa Y Itoh H Ya-mamoto M and Nakajima K Autonomous Precision OrbitControl of ALOS-2 for Repeat-Pass SAR Interferometry in Pro-ceedings of IGARSS (IEEE International Geoscience and Re-mote Sensing Symposium) 21ndash26 July 2013 Melbourne Aus-tralia 2013

Abstract

Introduction

Geology of the study area

Materials and methods

Satellite remote sensing data and characteristics

Methods

Results and discussion

Conclusions

Data availability

Competing interests

Acknowledgements

References

Figure 1 Location of the Kelantan state in Peninsular Malaysia

2014 and 2015) which have led to significant damage to live-stock agricultural production homes and businesses in theKelantan River basin (Pradhan 2009 Pradhan et al 2009Pradhan and Youssef 2011 Tehrany et al 2013 Nazarud-din et al 2014) The problem stems from the inappropriateuse of lands that are vulnerable to erosion quick water runoffand slope failure Two distinct wet seasons from Septem-ber to December and February to May are reported by theMalaysian Meteorological Department (MMD) in PeninsularMalaysia (Mardiana et al 2000) A flood episode in Decem-ber 2014 caused several landslides one of which occurred foreach kilometer of the area of Kelantan state Mapping land-slides in tropical environments is difficult because landscapesare generally covered by thick tropical rainforest (Plank etal 2016) In addition the cloudy weather and rain affectedthe quality of the optical satellite remote sensing data (Mel-gani 2006 Ju and Roy 2008 Ramli et al 2009 Hashim etal 2013 Razak et al 2013 Eckardt et al 2013 Shahabiand Hashim 2015) The main target of this investigation isto identify high potential risk and susceptible zones in theKelantan River basin for mapping landslide occurrence zoneusing satellite remote sensing technology and geographic in-formation system (GIS) techniques

Advances in remote sensing technology allow the applica-tion of synthetic aperture radar (SAR) data in the geological

structural analysis for tropical environments (Shimada andIsoguchi 2002 Morelli and Piana 2006 Ramli et al 2009Pour and Hashim 2014 2015a b Pour et al 2013 2014)Structural field mapping is often difficult in heavily vegetatedterrain This is the case with the study area where dense veg-etation cover deep weathering and scarcity of bedrock ex-posure hamper geological structure mapping over a long dis-tance In addition the use of optical remote sensing data islimited due to the persistent cloud coverage of the study areafor the most part of the year (Ramli et al 2009 Hashim etal 2013) SAR data contain the potential to penetrate cloudand vegetation and unlike the optical data are dependentupon the surface roughness of the materials ideally suitedto mapping lineaments (faults and fractures) in tropical en-vironments Lineaments are related to large structural frac-tures which represent zones of weakness in the brittle partof the lithosphere The presence of linear tectonic structuresis one of the most important factors in geological hazard oc-currences (Bannert 2000a b) Lineaments are representedby faults (linear features) lithological contacts between rockunits and drainage patterns in any area (van der Pluijm andMarshak 1997) Observation from satellite images (LandsatThematic Mapper) in the Himalayan Mountains of Nepal andChina revealed a clear connection between active faults as-sociated with earthquakes and the occurrence of large land-slides (Bannert 2000a b) Therefore delineation of faultsand fractures using advanced remote sensing technology inany region is a necessity to assess the potential for many nat-ural geological hazards Numerous investigations have usedfaults drainage patterns and lithology factors as importantfactors to measure and map the susceptibility of geologicalhazard (Guzzetti et al 1999 Dai and Lee 2002 Suzen andDoyuran 2004 Abdullah et al 2013)

Landsat-8 was launched on 4 February 2013 carrying theOperational Land Imager (OLI) and the Thermal InfraredSensor (TIRS) sensors These two instruments collect im-age data for nine visible near-infrared shortwave infraredbands and two longwave thermal bands (Irons et al 2012)They have a high signal-to-noise radiometer performanceenabling 12 bit quantization of data allowing for more bitsfor better land cover characterization by sharper color con-trast Landsat-8 provides moderate-resolution imagery from15 to 100 m of Earthrsquos surface and polar regions (Roy et al2014)

The Advanced Land Observing Satellite-2 (ALOS-2) waslaunched on 24 May 2014 as the successor of ALOS-1 (launched on 24 January 2006 and decommissioned inMay 2011) (httpglobaljaxajppress20140520140524_daichi2html Blau 2014) The ALOS-2 is exclusively in-stalled with the Phased Array type L-band Synthetic Aper-ture Radar-2 (PALSAR-2) using microwaves to maximizedits ability compare to the ALOS-1 on which three sensors(one microwave and two optical devices) were on board(Igarashi 2001 Suzuki et al 2012)

32 Methods

Nxminus1summ=0

Nyminus1sumn=0

g = f middotw (2)

i+Wy2summ=iminusWy2

j+Wx2sumn=jminusWx2

Sum 1000 1000 1000 1000 1000 1000 1000 1000 1000 1000

Factors Weight

LSM=nsumi=1

(RiWi) (5)

episode ()(mm)

Sum 98 1 942 90526

5 Conclusions

References

Abstract

Introduction


Methods


Conclusions

Data availability

Competing interests

Acknowledgements

References

32 Methods

Nxminus1summ=0

Nyminus1sumn=0

g = f middotw (2)

i+Wy2summ=iminusWy2

j+Wx2sumn=jminusWx2

Sum 1000 1000 1000 1000 1000 1000 1000 1000 1000 1000

Factors Weight

LSM=nsumi=1

(RiWi) (5)

episode ()(mm)

Sum 98 1 942 90526

5 Conclusions

References

Abstract

Introduction


Methods


Conclusions

Data availability

Competing interests

Acknowledgements

References

32 Methods

Nxminus1summ=0

Nyminus1sumn=0

g = f middotw (2)

i+Wy2summ=iminusWy2

j+Wx2sumn=jminusWx2

Sum 1000 1000 1000 1000 1000 1000 1000 1000 1000 1000

Factors Weight

LSM=nsumi=1

(RiWi) (5)

episode ()(mm)

Sum 98 1 942 90526

5 Conclusions

References

Abstract

Introduction


Methods


Conclusions

Data availability

Competing interests

Acknowledgements

References

32 Methods

Nxminus1summ=0

Nyminus1sumn=0

g = f middotw (2)

i+Wy2summ=iminusWy2

j+Wx2sumn=jminusWx2

Sum 1000 1000 1000 1000 1000 1000 1000 1000 1000 1000

Factors Weight

LSM=nsumi=1

(RiWi) (5)

episode ()(mm)

Sum 98 1 942 90526

5 Conclusions

References

Abstract

Introduction


Methods


Conclusions

Data availability

Competing interests

Acknowledgements

References

Nxminus1summ=0

Nyminus1sumn=0

g = f middotw (2)

i+Wy2summ=iminusWy2

j+Wx2sumn=jminusWx2

Sum 1000 1000 1000 1000 1000 1000 1000 1000 1000 1000

Factors Weight

LSM=nsumi=1

(RiWi) (5)

episode ()(mm)

Sum 98 1 942 90526

5 Conclusions

References

Abstract

Introduction


Methods


Conclusions

Data availability

Competing interests

Acknowledgements

References

Sum 1000 1000 1000 1000 1000 1000 1000 1000 1000 1000

Factors Weight

LSM=nsumi=1

(RiWi) (5)

episode ()(mm)

Sum 98 1 942 90526

5 Conclusions

References

Abstract

Introduction


Methods


Conclusions

Data availability

Competing interests

Acknowledgements

References

episode ()(mm)

Sum 98 1 942 90526

5 Conclusions

References

Abstract

Introduction


Methods


Conclusions

Data availability

Competing interests

Acknowledgements

References

Sum 98 1 942 90526

5 Conclusions

References

Abstract

Introduction


Methods


Conclusions

Data availability

Competing interests

Acknowledgements

References

Sum 98 1 942 90526

5 Conclusions

References

Abstract

Introduction


Methods


Conclusions

Data availability

Competing interests

Acknowledgements

References

Sum 98 1 942 90526

5 Conclusions

References

Abstract

Introduction


Methods


Conclusions

Data availability

Competing interests

Acknowledgements

References

Sum 98 1 942 90526

5 Conclusions

References

Abstract

Introduction


Methods


Conclusions

Data availability

Competing interests

Acknowledgements

References

Sum 98 1 942 90526

5 Conclusions

References

Abstract

Introduction


Methods


Conclusions

Data availability

Competing interests

Acknowledgements

References

Sum 98 1 942 90526

5 Conclusions

References

Abstract

Introduction


Methods


Conclusions

Data availability

Competing interests

Acknowledgements

References

Sum 98 1 942 90526

5 Conclusions

References

Abstract

Introduction


Methods


Conclusions

Data availability

Competing interests

Acknowledgements

References

References

Abstract

Introduction


Methods


Conclusions

Data availability

Competing interests

Acknowledgements

References

Abstract

Introduction


Methods


Conclusions

Data availability

Competing interests

Acknowledgements

References

Abstract

Introduction


Methods


Conclusions

Data availability

Competing interests

Acknowledgements

References

Abstract

Introduction


Methods


Conclusions

Data availability

Competing interests

Acknowledgements

References

Documents

Application of Landsat-8 and ALOS-2 data for structural ...eprints.utm.my/id/eprint/74865/1/MazlanHashim2017... · 1286 A. B. Pour and M. Hashim: Application of Landsat-8 and ALOS-2