Coupling of DEM and remote-sensing-based approaches for semi ...

10 downloads 66548 Views 1MB Size Report
Coupling of DEM and remote-sensing-based approaches for semi-automated detection of regional geostructural features in Zagros mountain, Iran. Authors ...
Arab J Geosci (2013) 6:91–99 DOI 10.1007/s12517-011-0361-0

ORIGINAL PAPER

Coupling of DEM and remote-sensing-based approaches for semi-automated detection of regional geostructural features in Zagros mountain, Iran Saied Pirasteh & Biswajeet Pradhan & Hojjat O. Safari & Mohammad Firuz Ramli

Received: 20 January 2011 / Accepted: 18 May 2011 / Published online: 31 May 2011 # Saudi Society for Geosciences 2011

Abstract In recent years, remote-sensing data have increasingly been used for the interpretation of objects and mapping in various applications of engineering geology. Digital elevation model (DEM) is very useful for detection, delineation, and interpretation of geological and structural features. The use of image elements for interpretation is a common method to extract structural features. In this paper, linear features were extracted from the Landsat ETM satellite image and then DEM was used to enhance those objects using digital-image-processing filtering techniques. The extraction procedures of the linear objects are performed in a semi-automated way. Photographic elements and geotechnical elements are used as main keys to extract the information from the satellite image data. This paper emphasizes on the application of DEM and usage of various filtering techniques with different convolution kernel size applied on the DEM. Additionally, this paper discusses about the usefulness of DEM and satellite digital data for extraction of structural features in SW of Zagros mountain, Iran. S. Pirasteh : B. Pradhan (*) Institute of Advanced Technology, Spatial Numerical Modeling Laboratory, University Putra Malaysia, 43400 UPM, Serdang, Malaysia e-mail: [email protected] e-mail: [email protected] H. O. Safari Faculty of Earth Sciences, Golestan University, Sari, Iran M. F. Ramli Faculty of Environmental Studies, University Putra Malaysia, 43400 UPM, Serdang, Malaysia

Keywords Remote sensing . Image processing . DEM . Linear feature mapping . GIS . Zagros mountain

Introduction Faults, folds, and lineaments interpreted from remotely sensed data are often used as indicator of major fractures in near surface (Ali Syed and Pirasteh 2003; Juhari and Ibrahim 1997; Jana 2002). Many researchers have mapped and interpreted the structure of a region based on the information extracted from aerial photographs and satellite images (Berberian 1995; Sattarzadeh et al. 2000; Pradhan et al. 2006, 2010a, b, c; Pradhan 2010d; Ayazi et al. 2010; Farrokhnia et al. 2010; Pradhan and Youssef 2010; Safari et al. 2009; Youssef et al. 2009, 2011). The structural map can further be used for geoengineering applications such as tunneling, dams, etc. Ali and Pirasteh (2004) and Pirasteh and Syed Ahmad (2005) showed that using Laplacian convolution filter on band-4 of the Enhanced Thematic Mapper (ETM+) Landsat data increase the ability to extract structural features. However, the potential of the DEM for the extraction of these feature and comparison with the satellite images have not been performed earlier. Bolstad and Stowe (1994) produced DEMs from 1:40,000 scale aerial photographs and SPOT panchromatic pairs with automatic correlation technique and tested the accuracy of these DEMs and their derivative surfaces (slope and aspect). Accuracies of DEMs derived from different data models were assessed by Li (1994). One of the DEMs was derived from photogrammetrically measured contour data, and the other was the gridded data from aerial photographs. As a result, he suggested that contouring could be a better sampling strategy for better accuracy.

92

Generally, DEM uses for building 3D view of an area. DEM can be created from either stereo pairs derived from satellite data or aerial photographs or generated from digital terrain elevation data. DEM can be readily combined with image data (both optical and SAR) for a number of different purposes (Lillesand and Kiefer 1987). For example, DEM was used to calibrate and geocode SAR

Fig. 1 Study area showing the Zagros mountains in southwest Iran

Arab J Geosci (2013) 6:91–99

data classification is very limited due to higher cost involved in preparing the DEM. Mostly, files in DEM format are taken from contour maps. Various techniques of image preprocessing, integration of ancillary data have shown positive results to improve the filtering results. This study carried out to explore the use of DEM in enhancing structural and

Arab J Geosci (2013) 6:91–99

geomorphology features with different filtering kernel size and classification techniques and further comparison with the information extraction from the satellite image to understand the potential of the DEM for structural features extraction. This study demonstrates the comparison of filtering on DEM and ETM+ Landsat data.

Study area The study area is a part of the Zagros mountains which lies in the southwest (Fig. 1) of Iran is one of the best exposed regions for geostructural features. This paper tries to express and answer the following question: (a) potential of DEM for evaluation and extraction of geostructural features, (b) the usefulness of DEM with or without satellite image to extract the geostructural features with certain degree of accuracy, (c) identifying of a suitable filtering and convolution kernel to optimize the enhancement for extraction of the geostructural features from DEM, and (d) to generate a 3D image of the area in order to increase the accuracy of the semi-automated interpretation in conjunction with several field data.

Data and methodology used Digital image processing and DEM Generally speaking, different methods may be applied on the image to extract the geostructural and geomorphology features information (Lee and Pradhan 2006, 2007; Pirasteh et al. 2008, 2009, 2010a, 2010b, 2011; Lee 2005; Giacinto et al. 2000; Pradhan et al. 2006; Pradhan and Lee 2009, 2010a, b, c; Pradhan and Buchroithner 2010; Pradhan and Pirasteh 2010). One of these methods is filtering and convolution high-pass technique on the satellite image like ETM+ Landsat. However, this paper discusses about various convolution kernel size and matrix using different filtering and convolution techniques (i.e., high pass, low pass, directional, and Laplacian) on DEM to compare them with each other and further to ETM+. This will indicate how DEM and which technique is more affected to extract structural features. Firstly, the contrast of six bands of the ETM+ data are digitally enhanced. The images of all the bands are compared with respect to contrast and definition of structural features. As a result of visual evaluation, Landsat ETM+ band-4 data which record the information at the wavelength between 0.75 and 0.90 μm were selected for this study, since it shows good contrast and display structural features like folds, faults, and lineaments as compared with the other bands. However, the infrared

93

band-4 of ETM+ Landsat image has been considered for filtering convolution kernel size too. Both directional and Laplacian convolution filter has been attempted within ENVI version 4.2 software environment. Figure 2 illustrates both filters kernel size and marix applied on the ETM+ satellite image. The information extracted from both DEM and ETM+ data are finally merged and compared in order to understand the potential of the DEM. Digital-imageprocessing filtering convolution techniques have been applied on the DEM, and different images are produced. The high-pass technique has been applied on DEM with 3× 3, 5×5, 7×7, and 11×11 martix convolution kernel size (Fig. 3). The lineaments features are further enhanced. Low-pass filtering techniques have also been applied on DEM to enhance the objects (Fig. 4). Kernel size matrix applied on DEM is 3×3, 5×5, and 7×7. Laplacian filtering convolution kernel size of 3×3 and 5×5 have also been attempted to extract the objects through a semi-automated interpretation on the DEM (Fig. 5). Finally, the directional technique with zero angle with kernel size 3×3 and 5×5 have been carried out to produce the images with better enhancement and ability to interpret the objects (Fig. 6). The source DEM has been produced using Rivertools software with text format extracted from aerial photographs and later generated digital topography map by Iranian Survey Organization (ISO). The data have been converted to text format using Microstation J software. Preprocessing, which is designed to remove any undesirable raster characteristics is produced by the initial processing of the raw data and remove noise and correct geometric distortions.

Fig. 2 Values produced by filtering techniques

94 Fig. 3 Showing high-pass techniques applied on DEM

Arab J Geosci (2013) 6:91–99

a 3*3

b 5*5

c 11*11

Fig. 4 Showing low-pass filtering techniques on DEM

a 3*3

b 5*5

Arab J Geosci (2013) 6:91–99 Fig. 5 Showing Laplacian filtering techniques applied on DEM

95

a 3*3

The geometric correction of the ETM+ satellite image was performed by registering the image to 1:25,000 scale digital topography map sheets by selecting 100 ground control points. This has been carried out using ENVI software. The resulted root mean square error was less than 1 pixel (20 m). This was achieved by using the first-order nearest neighborhood transformation method (Pradhan 2010a, b, c, d, e, Pradhan 2011). This is because the difference in contours lines from the digital topography map provided by ISO is 20 m. A geometrically corrected raster image was then overlaid on the DEM to produce a 3D image for increasing the accuracy and representing the lineaments and geomorphology features. The geostructural features were detected from the ETM+ data using image techniques elements after applying the digital-imageprocessing techniques. These techniques enhanced the Fig. 6 Showing directional filtering applied on DEM

3*3

b 11*11

objects and further facilitate the analyst to make on-screen digitization. Furthermore, the objects have been mapped by interpreting and using on-screen digitations along with the field observations. This interpretation is defined as semiautomated interpretation. The keys for interpretation are photography elements such as tone, shape, association, and geotechnical elements such as erosion, topography, and vegetation. The same features are extracted from DEM separately after applying all aforementioned filtering techniques. The linearity and sharp lines, slope, light gray-tone pixels, and elevation topography are main keys for interpretation on DEM. Subsequently, several field observations were carried out in order to validate the extracted information from ETM+ and DEM data (Fig. 7). Finally, the extracted information were converted into the ARCGIS version 9.2 environments to produce the final structural map. 7*7

96

Arab J Geosci (2013) 6:91–99

Fig. 7 Showing lineaments and folds in Zagros mountains, southwest Iran

Results and discussion In this paper, the most common way of presenting the geostructural features and the potential of DEM using the digital-image-processing techniques is disucssed. We applied different filtering techniques and convolution kernel size on DEM. The geostructural features extracted from infrared band-4 of ETM+ Landsat enhanced the feature for structural features detection. This emphasizes and supports the information extracted from DEM. The information extracted from DEM has shown a reasonable understanding and matches with the information extracted from ETM+. To further enhance the geological structural information, filtering techniques such as Laplacian and directional convolution filters are applied to Band-4 Landsat ETM+ data. The result shows that the number of filters of the values produced (Fig. 2) enhanced the general characteristics of folds, faults, lineaments, and thrusts of the area. The folds, faults, and lineaments in the study area are seen through different enhancements and filtering on images and some are only seen in a directionally filtered images. The reorganization of folds, faults, and lineaments on the images are based on vegetation linearity, tonal changes in the images,

drainage pattern, topographic breaks, landscapes, and discontinuity in the lithology, tectonic landforms such as klippe, nappe, fenster and scarps. Different filtering kernel size on DEM reveals that the directional filter with 7×7 kernel is more powerful to interpret the geostractural features visually and on-screen digital interpretation. The comparison of the different generated images using various filtering techniques are depicted in Table 1 to show the potential of the DEM for semi-automated interpretation from DEM. Moreover, the combinition of the extracted information from DEM and ETM+ satellite data is very similar This was further confirmed when the field observations information have been applied. In some pixels that do not allow the interpretation of the structural features on ETM+, using DEM can be a good combinition to extract more information and validate the performances of ETM+. The DEM and ETM+ were used to generate the 3D image of Zagros mountain (Fig. 8) and overlaid the structural features on 3D image for increasing the accuracy and validating with the ground observation. When we applied high-pass filtering with convolution kernel sizes 3×3 to 13×13 on DEM, it has been found that the structural features are more enhanced and can easily detect the lineament features. But, when the same method was applied on ETM+ Landsat band 4, it did not suggest a better enhancement. Although satellite data have its own potential to detect the objects and extract the information, it has been shown that working on DEM using a high-pass kernel size to extract structural features are more preferable than the ETM+ data itself. It is because the ability of the analyst to interpret the object increases mainly when there is vegetation cover and makes it difficult to interpret the geostructural and geomorphological features on the image. The low-pass filtering process on DEM shows that when different convolution kernel size was applid, no changes can be seen on the original DEM. Furthermore, the Laplacian filtering with different convolution kernel size on DEM shows that if the kernel size is 3×3 and 5×5, the source DEM remain dark, but increasing the kernel size; the DEM starts to make the features enhance. This enhancement is not as much as what we have seen in the high-pass convolution filter. The directional filtering with zero angle for kernel size 3×3, 5×5,

Table 1 Comparison of the various filtering techniques applied on DEM Convolution kernel size

High pass

Low pass

3×3 5×5 7×7 9×9

Very dark Low enhanced Medium enhanced High enhanced

Normal Normal Normal Normal

as source source as source as source as

DEM DEM DEM DEM

Laplacian

Directional

Very dark Very dark Low enhanced Medium enhanced

Very dark Very low Low enhanced Medium-high enhanced

Arab J Geosci (2013) 6:91–99

97

Fig. 8 3D view and structural interpretation of Landsat ETM+ FCC 4-3-2 in Zagris Mountain, SW Iran. Approximate scale: 1:450,000. Field photograph at right side of the figure illustrates slickensided surface and showing related movement along the fault plane

7×7, and 9×9 shows that the features would be enhanced when the kernel size increases and most of the features can clearly be seen on filtered DEM. But not as much as we could see in the high-pass filtered DEM.

Fig. 9 Structural map of southwest Iran

Concluding remarks Comparing the filtering techniques on DEM suggests that the high-pass filtering with 9×9 convolution kernel size is

98

reliable to be used for the detection of geostructural features. Furthermore, the comparison between high-pass filtered DEM and ETM+ Landsat with band-4 infrared indicates that the combination of both raster data makes it powerful and strengthens the ability to have a semiautomated interpretation. The information extracted from the ETM+, DEM, and field observations have been merged together in GIS environment and produced a structural map (Fig. 9) for future geoengineering application purposes. As a final conclusion, this study shows the advantages of the remotely sensed data and GIS techniques for detecting the linear features by using semiautomatic filtering techniques. It also reveals that the DEM can be a very useful tool for the study of geomorphology within the GIS environment. Similar studies can be adopted for different study areas with mountainous topography.

References Ali Syed A, Pirasteh S (2003) Remote sensing and gis study of tectonics and net erosion rates in the Zagros Structural Belt, southwestern Iran. Mapp Sci Remote Sens 40(4):253–262 Ali Syed A, Pirasteh S (2004) Geological application of Landsat ETM for mapping structural geology and interpretation: aided by remote sensing and gis. Int J Remote Sens 25(21):4715–4727 Ayazi MH, Pirasteh S, Arvin AKP, Pradhan B, Nikouravan B, Mansor S (2010) Disasters and risk reduction in groundwater: Zagros mountain southwest Iran using geoinformatics techniques. Disast Adv 3(1):51–57 Berberian M (1995) Master “blind” thrust faults hidden under the Zagros Folds: active basement and surface morphotectonics. Tectonophysics 241:193–224 Bolstad PV, Stowe T (1994) An evaluation of DEM accuracy: elevation, slope and aspect. Photogramm Eng Remote Sensing 60(11):1327–1332 Farrokhnia A, Pirasteh S, Pradhan B, Pourkermani M, Arian M (2010) A recent scenario of mass wasting and its impact on the transportation in Alborz Mountains. Iran using geo-information technology. Arabian J Geosci. doi:10.1007/s12517-010-0238-7 Giacinto G, Roli F, Bruzzone L (2000) Combination of neural and statistical algorithms for supervised classification of remotesensing images. Pattern Recogn Lett 21:385–397 Jana MM (2002) Application of remote sensing in the study of geomorphic processes and landform in piedmont zone of Darjeeling Sub-Himalaya. Photonirvachak-J Ind 30(1–2):61–72 Juhari MA, Ibrahim A (1997) Geological application of Landsat Themetic Mapper imagery: mapping and analysing of lineaments in NW Penisula Malaysia. Available at: www.gisdevelopment. net/aars/1997/. pp 1–7 Lee S (2005) Application and cross-validation of spatial logistic multiple regression for landslide susceptibility analysis. Geosci J 9(1):63–71 Lee S, Pradhan B (2006) Probabilistic landslide hazards and risk mapping on Penang Island, Malaysia. J Earth Syst Sci 115 (6):661–672 Lee S, Pradhan B (2007) Landslide hazard mapping at Selangor, Malaysia using frequency ratio and logistic regression models. Landslides 4(1):33–41

Arab J Geosci (2013) 6:91–99 Li Z (1994) A comparative study of the accuracy of digital terrain models (DTMs) based on various data models. ISPRS J Photogramm 49(1):2–11 Lillesand TM, Kiefer RW (1987) Remote sensing, and image interpretation, 2nd edn. Wiley, Singapore Pirasteh S, Syed Ahmad A (2005) Lithostratigraphic study from Dezful to Brojerd-Dorood areas SW Iran using digital topography, remote sensing and GIS. Indian Petr Geol J 13(1):1–13 Pirasteh S, Woodbridge K, Rizvi SMA (2008) Geo-information technology (GiT) and tectonic signatures: the River Karun & Dez, Zagros Orogen in south-west Iran. Int J Remote Sens 30:389–395 Pirasteh S, Pradhan B, Mahmoodzadeh A (2009) Stability mapping and landslide recognition in Zagros mountain South West Iran: a case study. Disast Adv 2(1):47–53 Pirasteh S, Rizvi SMA, Ayazi MH, Mahmoodzadeh A (2010a) Using microwave remote sensing for flood study in Bhuj Taluk, Kuchch District Gujarat, India. International Geoinformatics Research and Development Journal 1(1):13–24 Pirasteh S, Safari HO, Pradhan B, Attarzadeh I (2010b) Lithomorphotectonics analysis using Landsat ETM data and GIS techniques: Zagros Fold Belt (ZFB), SW Iran. Intl Geoinf Res Dev J 1(2):28–36 Pirasteh S, Pradhan B, Rizvi SM (2011) Tectonic process analysis in Zagros mountain with the aid of drainage networks and topography maps dated 1950–2001 in GIS. Arabian J Geosci 4 (1–2):171–180. doi:10.1007/s12517-009-0100-y Pradhan B (2010a) Remote sensing and GIS-based landslide hazard analysis and cross-validation using multivariate logistic regression model on three test areas in Malaysia. Adv Space Res 45 (10):1244–1256 Pradhan B (2010b) Application of an advanced fuzzy logic model for landslide susceptibility analysis. Intl J Comput Intel Syst 3 (3):370–381. doi:10.2991/ijcis.2010.3.3.12 Pradhan B (2010c) Manifestation of an advanced fuzzy logic model coupled with geoinformation techniques for landslide susceptibility analysis. Environ Ecol Stat. doi:10.1007/s10651-0100147-7 Pradhan B (2010d) Role of GIS in natural hazard detection, modeling and mitigation. Disast Adv 3(1):3–4 Pradhan B (2010e) Landslide susceptibility mapping of a catchment area using frequency ratio, fuzzy logic and multivariate logistic regression approaches. J Indian Soc Remote Sens 38(2):301–320 Pradhan B (2011) Use of GIS based fuzzy relations and its cross application to produce landslide susceptibility maps in three test areas in Malaysia. Environ Earth Sci 63(2):329–349 Pradhan B, Buchroithner MF (2010) Comparison and validation of landslide susceptibility maps using an artificial neural network model for three test areas in Malaysia. Environ Eng Geosci 16 (2):107–126 Pradhan B, Lee S (2009) Landslide risk analysis using artificial neural network model focusing on different training sites. Intl J Phys Sci 3(11):1–15 Pradhan B, Lee S (2010a) Delineation of landslide hazard areas on Penang Island, Malaysia, by using frequency ratio, logistic regression, and artificial neural network models. Environ Earth Sci 60(5):1037–1054 Pradhan B, Lee S (2010b) Landslide susceptibility assessment and factor effect analysis: back-propagation artificial neural networks and their comparison with frequency ratio and bivariate logistic regression modelling. Environ Modell Softw 25(6):747–759 Pradhan B, Lee S (2010c) Regional landslide susceptibility analysis using back-propagation neural network model at Cameron Highland, Malaysia. Landslides 7(1):13–30

Arab J Geosci (2013) 6:91–99 Pradhan B, Pirasteh P (2010) Comparison between prediction capabilities of neural network and fuzzy logic techniques for landslide susceptibility mapping. Disast Adv 3(2):26–34 Pradhan B, Youssef AM (2010) Manifestation of remote sensing data and GIS on landslide hazard analysis using spatial-based statistical models. Arabian J Geosci 3(3):319–326. doi:10.1007/ s12517-009-0089-2 Pradhan B, Singh RP, Buchroithner MF (2006) Estimation of stress and its use in evaluation of landslide prone regions using remote sensing data. Adv Space Res 37:698–709 Pradhan B, Lee S, Buchroithner MF (2010a) A GIS-based backpropagation neural network model and its cross application and validation for landslide susceptibility analyses. Comput Environ Urban Syst 34:216–235 Pradhan B, Lee S, Buchroithner M (2010b) Remote sensing and GISbased landslide susceptibility analysis and its cross-validation in three test areas using a frequency ratio model. Photogramme Fernerkund Geoinf 1:17–32. doi:10.1127/1432-8364/2010/0037

99 Pradhan B, Pirasteh S, Varatharajoo R (2010c) Enhancement of automated lineament extraction from IRS- 1B satellite imagery for part of Himalayan region. Intl J Geo Inf 6(2):41–50 Safari HO, Pirasteh S, Pradhan B (2009) Upliftment estimation of the Zagros Transverse Fault in Iran using geoinformatics technology. Remote Sens 1(4):1240–1256. doi:10.3390/rs1041240 Sattarzadeh Y, CosGrove JW, Vita-Finzi C (2000) The interplay of faulting and folding during the evoluation of the Zagros. In: Cosgrove, J.W.& Ameen. M.S. (eds) Forced Folds and Fractures. Special publication, 169, Geological Society, London. pp 187–196 Youssef AM, Pradhan B, Gaber AFD, Buchroithner MF (2009) Geomorphological hazard analysis along the Egyptian Red Sea coast between Safaga and Quseir. Nat Hazards Earth Syst Sci 9:751–766 Youssef AM, Pradhan B, Sabtan AA et al (2011) Coupling of remote sensing data aided with field investigations for geological hazards assessment in Jazan area, Kingdom of Saudi Arabia. Environmental Earth Sciences, doi:10.1007/s12665-011-1071-3