Sunteți pe pagina 1din 15

ISPRS Journal of Photogrammetry and Remote Sensing 66 (2011) 818832

Contents lists available at SciVerse ScienceDirect

ISPRS Journal of Photogrammetry and Remote Sensing


journal homepage: www.elsevier.com/locate/isprsjprs

Auto-detection and integration of tectonically signicant lineaments from SRTM DEM and remotely-sensed geophysical data
Alaa A. Masoud a,, Katsuaki Koike b,1
a b

Geology Department, Faculty of Science, Tanta University, 31527 Tanta, Egypt Graduate School of Science and Technology, Kumamoto University, 2-39-1 Kurokami, Kumamoto 860-8555, Japan

a r t i c l e

i n f o

a b s t r a c t
A set of techniques was developed for automatically detecting tectonic lineaments from multi-source remotely-sensed data at various scales. The techniques include adaptive shading of grid data to enhance linear features, a segment-tracing algorithm to extract line segments from the shaded grid data, grouping of the segments by concatenating short segments, and connecting them by proximity and co-linearity criteria to form a lineament that represents signicant tectonic structure. B-spline smoothing was adopted for lineament representation. Finally, a technique for assessing the orientations and styles of faulting (normal, reverse, and strike-slip types) was developed for use in characterizing the extrapolated fracture planes. The applicability of the developed techniques was examined using 30 arc-second topography/ bathymetry grids, 1-min gravity anomaly grids, and 2-min total eld magnetic intensity grids covering Egypt and its surroundings. Lineaments derived from data types so diverse in composition and from various depths corresponded well with the referenced tectonic features over much of the region. Prominent trends and faulting styles of lineaments provided important clues as to the timing of their development as well as strong support for a structural inheritance model. Results demonstrated the effectiveness of the developed techniques combined with integration of remotely-sensed data in detecting regional fracture systems accurately and in characterizing geodynamics over a long timeframe. 2011 International Society for Photogrammetry and Remote Sensing, Inc. (ISPRS). Published by Elsevier B.V. All rights reserved.

Article history: Received 2 January 2010 Received in revised form 15 August 2011 Accepted 16 August 2011

Keywords: Adaptive shading Segment tracing Segment grouping B-spline Tectonic model Egypt

1. Introduction Lineaments are one of the essential topographic features used in exploring for resources such as groundwater, hydrocarbons, minerals, and geothermal energy, as well as in mapping hazard susceptibility from earthquakes and landslides (e.g., Guild, 1974; Dix and Jackson, 1981; Sibson, 1986a,b; Boucher, 1995; Rowan and Bowers, 1995; Rowland and Sibson, 2004; Masoud et al., 2007). In this regard, characteristics of lineaments such as spatial extent, density, intersection, and orientation have proved signicant because they indicate zones and trends of high permeability and/or low pressure that may serve as pathways for migration, and thus targets for increased reserve. Lineaments may represent faults that control basin development and distribution of reservoirs (Warner, 1997). Regional lineaments are commonly interpreted as surface expressions of geologic weak zones at tectonic boundaries of basins and plates, as well as faults and rock fractures (e.g., Oakey, 1994; Fichler et al., 1999; Kudo et al., 2004; Salem et al., 2005; Milbury et al., 2007; Austin and Blenkinsop, 2008).
Corresponding author.
1

E-mail address: alaamasoud09@live.com (A.A. Masoud). Present address: Graduate School of Engineering, Kyoto University, Kyoto, Japan.

With recent advances in computer hardware and spatial-analysis techniques, computer-assisted lineament analysis on a large scale has become practical for characterizing geologic structures and tectonics. Many methods of automatically extracting lineaments from digital data have been proposed, such as from satellite images and the Digital Elevation Model (DEM). Such methods are mostly based on edge-detection techniques using spatial and morphological lters (e.g., Morris, 1991; Szen and Toprak, 1998; Tripathi et al., 2000). In cases where the results are extrapolated using lters, the frequency and connectivity of the line segments is strongly affected by the scale of the source data and the detection parameters (Argialas and Mavrantza, 2004). This may prevent plausible representation of tectonically signicant lineaments relevant to long rock fractures and faults. Therefore, it is indispensible to develop an auto-detection technique that can increase the frequency and connectivity of detected lineaments so that resultant lineament maps resemble fault maps by geologists. Another important point to be considered for auto-detection of tectonically signicant lineaments is the selection of suitable data sources. Generally, satellite images representing reectance and backscattering characteristics of the earths surface in response to electromagnetic waves at various wavelengths are used for lineament extraction. However, linear articial features unrelated to

0924-2716/$ - see front matter 2011 International Society for Photogrammetry and Remote Sensing, Inc. (ISPRS). Published by Elsevier B.V. All rights reserved. doi:10.1016/j.isprsjprs.2011.08.003

A.A. Masoud, K. Koike / ISPRS Journal of Photogrammetry and Remote Sensing 66 (2011) 818832

819

fractures, such as land use boundaries and land cover, also tend to be detected in satellite images. Instead of such data, it may be more promising to use grids of data from multiple sources that encompass the widely varied composition and depths represented by the topography and the subsurface geophysical attributes, especially when integrated at various scales. This can lead to a better understanding of the relationship between tectonic trends and anomalies from which they develop. Based on the above background, this study has three purposes. The rst is to develop a method that can enhance line features in grid data. For this, we use grids of composite topographybathymetry data, gravity data, and aeromagnetic data. The second is to develop methods for detection and connection of lineaments. The third is an effective integration of lineaments from such multisource data that describe different geologic attributes. The resultant lineaments are used in geological characterization to estimate orientations of fracture planes and fault types. We demonstrate, through a case study of a wide area around Egypt situated at the intersection of the African, Arabian, and Eurasian plate boundaries, that the techniques developed are useful for better understanding regional tectonics and geodynamics.

2.1. DEM shading With the increase in regions for which elevation data has been extrapolated using high spatial resolution DEM, geomorphologic and lineament analyses using DEM have also increased signicantly. Because shading of DEM can enhance the linear features, shading has played an important role in such analyses. A variety of shading methods have been proposed such as the Lambertian reection (Yoli, 1967; Foley et al., 1990; Horn, 1982), the Phong illumination (Bui-Tuong, 1975), the Blinn reection (Blinn, 1977), and ray-tracing (Foley et al., 1990), which are generally based on a physical model for illumination at specied azimuth and tilt angles over a DEM. Assuming a constant albedo equal to unity and Lambertian reectance properties for the whole DEM surface, the gray shade intensity value, G (x, y), for a pixel located at (x, y) in the DEM space becomes proportional only to the squared inner product of the directional vector along the illumination s, and the surface normal vector t as follows:

tx sx t y sy sz Gx; y cs t c p tx2 t y2 1
2

!2 c : constant 1

2. Methodology With these purposes in mind, the methods developed for enhancement of line features in grid data, identication and grouping of lineaments, calculating orientations of fracture planes, and fault type modeling are described below. The developed techniques are applicable not only to the DEM, but also to gravity and magnetic grid data, as was the case in the research presented herein.

where:

s sx ; sy ; sz T
and

t t x ; t y ; 1T

The horizontal components sx and sy control the illumination azimuth, while the vertical sz denes the tilt angle. This shade intensity at a given position varies with illumination azimuth

Fig. 1. Shaded DEM at constant tilt angle of 45 from the horizon with four illumination azimuths from west, north, east, and south, for the test DEM data (middle gure) in SW Sinai Peninsula in Egypt. Location of the test DEM is shown in Fig. 7. White arrows demarcate areas of suppression and shift in position of the linear features.

820

A.A. Masoud, K. Koike / ISPRS Journal of Photogrammetry and Remote Sensing 66 (2011) 818832

and tilt. Therefore, linear features striking perpendicularly to the illumination azimuth are optimally emphasized, whereas they are suppressed if oriented parallel to the illumination (Cooper, 2003). This bias can be conrmed by the DEM for a part of our study area located in southern Sinai, Egypt, where a at plain (Qaa Plain) lies adjacent to high mountain ranges (Fig. 1). In this gure, the position of the intense linear shades resulting from the same topographic features varies greatly even for opposite illumination azimuths. To reduce the bias from the illumination effect on the appearance of linear features, Zhou and Dorrer (1995) presented a method consisting of a wavelet transform of the DEM and then adjustment of the main illumination direction. Prechtel (2000) developed an alternative technique, identifying clusters of similarly-oriented cells from which a triangulation is derived for the deection of direction. Masoud and Koike (2006) applied a multidirectional technique that assigns the average of shade intensities for several illumination directions to the kernel center for accentuating tectonic lineaments. However, the problem in the above shading methods is that the tilt of the illumination source is kept constant. The tilt change should affect the shade intensities, so the characteristics of the line segments extrapolated from such shading may vary accordingly. In order to address this problem, we propose a technique for adaptive-tilt multi-directional shading (ATMDS) to obtain the maximum shade intensity at each DEM grid node. ATMDS sets the six northward illumination azimuths at 30 intervals clockwise from west and automatically sets the tilt angle dened by the vector component sz in Eq. (1) to range from 0 to 45 upward from the horizon as shown in Fig. 2. The reason for selecting these limited directions is because opposite southward illumination azimuths and tilt angles greater than 45 suppress shade intensities rather than enhancing them. At each grid node, shade intensities for the six illumination azimuths and the tilt angles in the dened range are calculated. Then, the maximum value is selected (Eq. (4)). This calculation is repeated for all the grid nodes.

od with tilt angles of 15, 30, and 45 on the multi-directional shading (as compared to Fig. 1 where the ATMDS method was not applied). Next, linear features (line segments) were extracted from the DEM using the segment-tracing algorithm (STA: Koike et al., 1995, 1998; Koike and Ichikawa, 2006) described below. The resultant extractions and rose diagrams showing the directions of linear features are compared in Fig. 3. Table 1 summarizes descriptive statistics of the length distributions of the line features. As clearly seen, high tilt angle (45) is inferior to identify linear features in the at and lowland areas such as those detected on the Qaa Plain (Fig. 3c). The superiority of ATMDS can be demonstrated by the numerous linear features extrapolated thoroughly from all parts of the test area regardless of the slope of the topography (Fig. 3d). Table 1 also demonstrates the superiority of the ATMDS method in that the number of detected segments and the total length are both greater compared with those detected applying 15, 30, and 45 tilt angles. 2.2. Detection and extraction of linear features Computer-assisted lineament detection and extraction basically rely on edge detectors that model edge points as extrema of data gradients over a threshold and their orientations in images. The edge points that are straightly aligned or smoothly curved are connected to form a longer segment as an element of a lineament. Based on the review by Argialas and Mavrantza (2004) of the widely used edge detectors, edge detectors can be classied into three categories based on the principles they use: line segment detection based on the region-growing algorithm, spatial ltering techniques, and the technique of directional detection of a segment in a local area. We tested the performance of these three categories of detectors that have been proved successful in many case studies: LSD (Grompone von Gioi et al., 2010), EDISON (RIUL, 2003), and STA. LSD is a parameterless linear-time line segment detector based on the region-growing algorithm (Burns et al., 1986). In LSD, a line segment is dened as a straight rectangular region whose edge points share roughly the same gradient angle. EDISON is a feature extraction tool that integrates condence-based edge detection (Meer and Georgescu, 2001) and mean shift-based image segmentation (Comanicu and Meer, 2002). In EDISON, Canny Edge is applied, and the line segment is adjusted to the candidate edge points based on thresholds for gradient, segment length, and hysteresis criteria for type, rank, and condence levels. STA denes a

( Gx; y max c

t x sxi t y syi szj p t x2 t y2 1

!)2

where i 0 ; 30 ; . . . ; 150

and

j 0 $ 45

Figs. 1 and 3 are based on the same DEM, but Fig. 3 illustrates the differences resulting from adopting the ATMDS method. Fig. 3 demonstrates the strong effect of applying the ATMDS meth-

Fig. 2. Schematic idea of the adaptive-tilt multi-directional shading (ATMDS) technique with six illumination azimuths at 30 interval from the northward and tilt angle ranging from 0 to 45. Base of this gure is a perspective view of the test DEM shown in Fig. 1.

A.A. Masoud, K. Koike / ISPRS Journal of Photogrammetry and Remote Sensing 66 (2011) 818832

821

Fig. 3. Shaded DEM of the test area (DEM and the location are shown in Figs. 1 and 7, respectively) at varying tilt angles (left); (a) 15, (b) 30, (c) 45, and (d) the adaptively set tilts, overlain with the detected segments (middle) derived by STA, and rose diagrams and counts of the segments (right).

line composed of adjacent pixels as a vector element by examining local variance of the gray level using 11 11 kernel in the digital image, and connects retained line elements along their expected directions. Threshold values for the extraction and the linkage of line elements are direction-dependent in order to avoid the illumination bias. Some advantages of this method over other ltering methods are its capabilities to (1) trace only continuous valleys and (2) extract more lineaments that are parallel to the illumination azimuth as well as those located in shadow areas. The three detectors were applied to the DEM for the test area (shown in Figs. 3(d) and 4(a)) that was shaded using the ATMDS. Fig. 4(bd) are the line segments extracted by LSD, EDISON, and STA, respectively. The performance of LSD depends on shade intensity; line segments were not extracted from low intensity (dark) zones. Connectivity of the line segments from LSD is the worst

Table 1 Descriptive statistics of the length (km) of the segments by STA for the test area (shown in Fig. 3) to demonstrate sensitivity of tilt angle to detection accuracy of segments. Tilt angle Parameter Mean Standard error Median Standard deviation Sample variance Range Minimum Maximum Sum Count 15 2.61 0.03 2.16 1.59 2.53 19.03 0.18 19.21 5638.23 2159 30 2.57 0.03 2.17 1.43 2.06 16.56 0.723 17.28 5388.73 2092 45 2.51 0.03 2.12 1.42 2.02 15.97 0.72 16.69 4937.75 1965 ATMDS 2.59 0.03 2.18 1.44 2.08 16.55 0.36 16.91 6026.13 2329

822

A.A. Masoud, K. Koike / ISPRS Journal of Photogrammetry and Remote Sensing 66 (2011) 818832

Fig. 4. Line segments for the test area detected from (a) the shaded DEM by ATMDS using (b) LSD, (c) EDISON, and (d) STA, respectively. Yellow arrows point to remarkable ridge features. (For interpretation of the references to colour in this gure legend, the reader is referred to the web version of this article.)

among the three detectors: most extracted segments do not follow the edges of those shade intensities that correspond with the long linear valleys and the boundaries between the mountain and plain. Line segments from EDISON are higher in frequency compared with those derived from LSD. However, there are two drawbacks: low detection accuracy from low shade intensity zones and appearance of many curved segments with short length in areas with rolling topography. Fracture information cannot be derived from such curved segments. Another serious problem is that LSD and EDISON cannot distinguish between line segments corresponding to ridge and valley features. Fractures tend to form linear valley features by selective erosion. More than 50% of the line segments detected are located on ridge features, as marked in Fig. 4(b and c). Conversely, STA can overcome the above problems as line segments are extracted thoroughly even from low shade intensity zones, and their connectivity into straight or slightly curved features is higher. These advantages enable STA to produce a more precise lineament map. Consequently, we selected STA for the lineament analysis. 2.3. Lineaments produced from line segments Line segments detected using STA (Figs. 3d and 4d) provide the potential candidates for lineaments that correspond to fractures and have tectonic signicance. To identify such geologic lineaments, the next step is to concatenate line segments to form one long lineament based on co-linearity and proximity. We have improved the algorithm for concatenation in STA to produce long lineaments from the short line segments by using B-spline as described below. As the rst step, segments are sorted according to their start X and then Y coordinates from the top-left margin of the study area. Each line segment is nominated as a start segment in a group. The

algorithm searches for adjacent segments that cross-cut or are displaced with a dynamic threshold distance related to the start segment geometry (Koike et al., 1995, 1998) and are co-linear at less than or equal to 10 angle. The longest searched segment that satises these criteria and give the largest cumulative length will be listed in the group as the second candidate segment. The cumulative length is calculated from the start point of the start segment to the end point of the candidate segment. The process is repeated for the candidate segment searching for the next segment in the group and so on until there would be no segments that satisfy the conditions. After completing the rst grouping, regrouping that merges cases of more than two groups is sequentially adopted using the same criteria and the average azimuths of the groups. The co-linearity angle (Fig. 5) proved crucial in adjusting for length, frequency, and straightness of the resulting lineaments. Performance analysis of the grouping algorithm showed that the smaller the co-linearity angle, the fewer segments there are that satisfy the condition. Hence, short straight lineaments are more abundant, and vice versa. Selection of the 10 co-linearity criteria is therefore based on the objective of retaining the maximum achievable length, frequency, and straightness, as well as minimizing the directional difference between the original segments and the resultant extrapolated lineament. The second step is to approximate the grouped line segments with one smooth curve using B-spline (Fig. 5). This curve is regarded as a lineament. We selected the cubic B-spline curve for representing lineaments because it well preserves the geometric continuity of the line segments within the group and is independent of the number of control points (Masoud and Koike, 2009). Since B-spline is a well-known piecewise polynomial function widely used for curve generation (e.g., Prautzsch et al., 2002), it is only briey explained herein.

A.A. Masoud, K. Koike / ISPRS Journal of Photogrammetry and Remote Sensing 66 (2011) 818832

823

N=6 P0 u=0 P1 B-spline Curve P2 P3

P4 u=1

Co-linearity angle

Start and end points of segments Pn B-spline curve control points


Fig. 5. Connection of line segments and approximation of them into a smooth curve using B-spline.

We set the inter-group control points (n + 1) as P0, P1, P2, . . ., Pn for n segments. P0 and P1 are the starting and midpoints of the rst segment, and Pn1 and Pn are the mid and end points of the last segment. The rest P2, . . ., Pn2 are the midpoints of the other segments. The B-spline curve of order u (degree u1) is then dened by the control points and the knot vector u:

of the desired plane and the surface slope i (Fig. 6), li should be equal to the vector product of the two normal vectors:

li t i n i 1; 2; . . . ; ms

The only unknown in this equation is n. It is prudent, therefore, to obtain n by applying the least-squares method and solving the normal equation that is deduced from:

Cu

n1 X i1

Ni;p uPi ;

06u61

( ) ms @ X xi li t i nT li t i n 0 @n i1

where C(u) is a piecewise continuous function dening the curve with N = n + 1 discrete control points Pi. Ni,p(u) is a basis function that blends the control points to form a smooth curve. The boundary condition of u is u = 0 at the rst control point (i = 1) and u = 1 at the last control point (i = n + 1).

2.4. Estimating fracture planes and fault modeling from extrapolated lineaments Under the assumption that a fracture along a fault plane has a nite dimension, a lineament, which is interpreted as a trace of a fracture or fault plane on the terrain composed of concatenated line segments, can be used to calculate the fault plane orientation (strike and dip). The trace curvature on the surface depends on the dip angle of the interpreted plane, which can be expressed by a geometrical relationship between the terrain and the plane. A normal vector of a plane is denoted by n. The geometrical relationship can be described through vector analysis (Koike et al., 1998). This technique has been applied successfully to model fault types in Kyushu island in Japan and validated by structural and lithologic data (Koike et al., 2001). In this technique, a slope made by the DEM and its normal vector is represented by ti (i = 1, 2, . . ., ms), where ms is the number of slopes through which the lineaments classied into the same group pass. Let the strike vector of the lineament in a group be li. Because li is equivalent to the intersection

where xi is a weighting coefcient. Furthermore, directional relationship between n and the average vector of ti, t, is used to judge whether the obtained plane represents a normal or reverse fault feature. Dipping criteria of the plane, its direction and angle of dip, are then estimated. We dene three features, normal, reverse, and strike-slip types depending on whether the directions of n and t are the same (normal), opposite (reverse), and undetermined (strike-slip) as shown in Fig. 6. Because strike-slip movements generally do not accompany downthrows on the fault plane, the fault type cannot be specied by the simple geometrical relationship in Fig. 6. Therefore, the undetermined type is assumed to have strike-slip components. The technique is thought to be applicable to the gravity and magnetic lineaments which depict very narrow low anomaly zones that may be caused by planes associated with faulting of different styles and types of movement. 3. Study area and data The techniques developed in our study were applied to detecting tectonically signicant lineaments and characterizing the geologic structures for Egypt and adjacent areas (Fig. 7). The region is geologically and structurally complex, composed of lithologies of different types and ages and showing varying tectonic styles and trends. Better understanding of the kinematics and the spatial

(a)
ti li (t1i , t2i , t3i ) (l1i , l 2i , l3i ) T
T

(b)
Fault plane normal vector Average surface normal vectors t n
Normal fault

Surface Lineament

n
Reverse fault

Fig. 6. (a) Denition of vectors used in estimating azimuth of fracture plane from lineaments, and (b) schematic idea of judging fault type by the directional relationship between the average of surface normal vectors and the normal vector of the interpreted fracture plane.

824

A.A. Masoud, K. Koike / ISPRS Journal of Photogrammetry and Remote Sensing 66 (2011) 818832

Fig. 7. Geological and kinematic structural map of the study area modied from CGMW (2006). Structural features described in Fig. 2 of Guiraud and Bosworth (1999) are shown in pink and those described in Fig. 10 of (Keely, 1994) are shown in black.

extent of the dominant tectonic trends may have implications for identifying the potential for natural resources in the area. The region has been shaped through extension-dominated tectonism alternated with intermittent less-inuential but intense compressive phases (Guiraud and Bosworth, 1999). Extensional tectonics lead to the Neotethys opening along the Mediterranean Basins in the Permian (Guiraud and Bosworth, 1999). Polyphased extension, inversion, and folding resulted in the development of the Syrian Arc belts, commenced in the latest Cretaceous, and continued as a series of pulses into the Late Oligocene (Keeley and Massoud, 1998). The convergence and suturing of AfricaArabia with Eurasia came to effect since the late Cretaceous and proceeded to the present (Ziegler and Roure, 1999). In the Late Oligocene, the Gulf of SuezRed Sea rifting initiated (Omar and Steckler, 1995). During

the Middle Miocene, the Red Sea divergent plate boundary abandoned the Gulf of Suez and broke through the Arabian plate via the Gulf of AqabaDead Sea transform fault, establishing the present-day plate kinematic framework (Guiraud and Bosworth, 1999). Many studies have addressed the repeated reactivation of the older structures by plate-scale stress elds along broad lineaments in the region (e.g., Guiraud and Bellion, 1995; Guiraud and Bosworth, 1999; Bumby and Guiraud, 2005; Abdeen and Greiling, 2005). In these studies, lineaments played an important role in understanding the regions geodynamic evolution. However, these were commonly generalized sketches reviewed from interpretations of geologic observations and scarce geophysical surveys which lack spatial inter-relationships. Integrating lineaments extrapolated from various spatial data describing surface and

A.A. Masoud, K. Koike / ISPRS Journal of Photogrammetry and Remote Sensing 66 (2011) 818832

825

subsurface features seems a more promising way to draw a detailed, reliable structural map of the region. We used three grid data sets: the 30-arc second topography/ bathymetry grid (SRTM30_PLUS) by Becker et al. (2009); the 1arc minute gravity anomalies grid (mGal) by Sandwell and Smith (1997); and the 2-arc minute earth magnetic anomaly (nT) grid by Maus et al. (2007) and Maus et al. (2009). For short, topography/bathymetry is hereafter termed topography. Use of the gravity data allowed an integrated response to density contrasts, and gravity data was closely related to topography by a linear transfer func-

tion. Integrating gravity and topographic data is signicant in studying crustal structures (McNutt, 1979; Smith and Sandwell, 1994). Gravity and magnetic data are related to different rock attributes in the subsurface which provides a basis for joint interpretation of both kinds of data. Coinciding gravity and magnetic anomalies may suggest a common origin. Contrasting gravity and magnetic anomalies may result from erasing the crusts magnetism by heating (Milbury et al., 2007) or from rocks with low magnetic contents (Fichler et al., 1999). Magnetic intensities of rocks occurring below the Curie isotherms can provide insight into the

Fig. 8. STA-derived line segments (left) overlaid on shaded grid data of (a) topography, (b) gravity, and (c) magnetic, and their corresponding lineaments by connection (right) overlaid on the shaded DEM (top), and the shaded gravity (middle) and magnetic data (bottom) as shown in color.

826

A.A. Masoud, K. Koike / ISPRS Journal of Photogrammetry and Remote Sensing 66 (2011) 818832

Table 2 Descriptive length statistics (unit: km) of (a) segments and (b) lineaments derived from the topographic, gravity, and magnetic data (shown in Fig. 8). Data (a) Parameter Mean Standard error Median Standard deviation Sample variance Range Minimum Maximum Sum (km) Count (Number) (b) Mean Standard error Median Standard deviation Sample variance Range Minimum Maximum Sum (km) Count (number) Topography Gravity Magnetic

integrated extrapolation based on all three kinds of data reduces the degree of ambiguity in surface and subsurface crustal structural modeling in the study area. 4. Results and discussion

13 0.03 11.4 7.71 59.48 107.49 4 108.50 823,578 60,131 81 0.93 69.22 36.03 1298.5 262.16 50.01 312.17 120,013 1460

58 0.32 45.6 40.84 1668.66 481.13 4 485.13 913,679 15,689 163 3.28 138.93 73.42 5390.5 493.7 100.11 593.81 81,689 483

60 0.31 49.5 38.66 1495.17 592.05 4 596.05 930,339 15,349 305 5.35 267.5 112.10 12,567.9 691.15 200 891.15 133,677 422

4.1. Line segments and lineaments STA-derived line segments and grouped lineaments from the three grid data sources are shown in Fig. 8, with their corresponding statistics summarized in Table 2. Magnetic-derived segments and lineaments are the longest, but least frequently occurring, among the three. Segments and lineaments derived from the topography are the most frequent, but are shorter in length compared with those derived from the gravity data. Variations in the length and frequency of occurrence of the line segments and lineaments are attributable to the difference in spatial resolutions of the source data. The segments had average lengths of 13, 58, and 60 km and maximum lengths of 108, 485, and 596 km while the lineaments had average lengths of 81, 163, and 305 km and maximum lengths of 312, 593, and 891 km, for the topography, gravity, and magnetic data, respectively. The enhancements in the average and maximum lengths of the lineaments over the segments prove the validity of the grouping technique applied to the various data in enhancing the connectivity of the linear features (Table 2). Superimposition of lineaments extrapolated from all 3 data sources and the reference structural features of Keely (1994) and

subsurface structure and the composition of rocks in the crust. As topography, gravity, and magnetic data are mutually independent,

Fig. 9. Distributions of three kinds of lineaments originated from the topography, gravity, and magnetic data and their superimposition on the geological map of Fig. 7. Color legend of the geological map is shown in Fig. 7. Structural features described in Fig. 2 of Guiraud and Bosworth (1999) are shown in pink and those described in Fig. 10 of keely (1994) are shown in black.

A.A. Masoud, K. Koike / ISPRS Journal of Photogrammetry and Remote Sensing 66 (2011) 818832

827

Guiraud and Bosworth (1999) is shown on Fig. 9. Lineaments from two to three data sources are congruent in many places and show broad concentrated continuous parallel zones, in spite of differences in spatial resolution and geologic attributes. Topographic lineaments showed more abundance in the inland, followed by the magnetic and the gravity lineaments. Gravity and magnetic lineaments dominate in the Mediterranean Sea. Structural zones along the lineaments combined from the 3 data sources agree in orientation and position with the common tectonic features in the region, as well as with the structures mapped in Keely (1994) and Guiraud and Bosworth (1999). They highlight important elements constituting the plate boundaries, intra-plate structural zones, boundaries of rift basins, and intra-basin structural features (Fig. 9), as explained in later sections. Directional trends of the three types of lineaments and their composite rose diagram are shown in Fig. 10, with statistics for

their frequency of occurrence and length summarized in Table 3. In the composite rose diagram, ratios of the counts of individual lineament types to total counts of all types along each direction are calculated. The major trends of lineaments from the three data sources are congruent, which suggests a possible common source for their origin. However, their order of abundance and relative frequencies vary greatly in the individual rose diagrams and the composite. The composite rose diagram shows ve prominent trends: NESW, NNE-SSW, NNW-SSE, NW-SE, and WNW-ESE, arranged counter-clockwise according to decreasing order of abundance (Fig. 10d). The magnetic lineaments show the largest contributions to the NE-SW, NNE-SSW, NW-SE, and WNW-ESE directions, arranged in decreasing order of contribution. Magnetic lineaments probably originate from compositional discontinuity in the underlying mantle and crust. The topographic lineaments contribute the

Fig. 10. Rose diagrams showing strike frequencies of the lineaments from (a) topographic, (b) gravity, and (c) magnetic data, and (d) total of the three kinds of lineaments. In (d), the colors in each direction sector denote the ratios of the three kinds of lineaments: the magnetic lineaments are the most prominent in most sectors.

Table 3 Statistics for the number (n) and length (l) in km of lineaments derived from topography (T), gravity (G), and magnetic (M) data in each directional range. Dominant trends are marked by bold letters. Trend NW Parameter () 8090 7080 6070 5060 4050 3040 2030 1020 010 010 1020 2030 3040 4050 5060 6070 7080 8090 T (n) 25 16 14 43 37 72 87 109 78 55 121 118 161 167 159 115 54 29 1460 T (l) 2200 1675 1246 3132 2836 5665 7368 8858 5895 4638 10,159 10,131 13,843 13,966 12,551 9146 4261 2443 120,013 G (n) 6 8 10 23 8 19 31 34 21 22 47 40 52 65 38 37 16 6 483 G (l) 1068 1594 1871 3565 1283 3495 6008 5621 3200 2954 8521 5975 8263 11,715 6422 6446 2721 967 81,689 M (n) 11 17 12 17 5 9 4 11 20 17 30 31 30 60 61 49 17 21 422 M (l) 3118 5083 4100 5643 1722 2695 1442 3610 5149 4911 9759 9255 10,374 18,489 20,900 16,027 4659 6741 133,677 Total (n) 42 41 36 83 50 100 122 154 119 94 198 189 243 292 258 201 87 56 2365 Total (l) 6386 8352 7217 12,340 5841 11,855 14,818 18,089 14,244 12,503 28,439 25,361 32,480 44,170 39,873 31,619 11.641 10,151 335,379

NE

Total

828

A.A. Masoud, K. Koike / ISPRS Journal of Photogrammetry and Remote Sensing 66 (2011) 818832

Fig. 11. Lineament distributions classied into the three fault types (normal, reverse, and strike-slip) and overlaid on the geological map of Fig. 7. Color legend of the geological map is shown in Fig. 7. Structural features described in Fig. 2 of Guiraud and Bosworth (1999) are shown in pink and those described in Fig. 10 of (Keely, 1994) are shown in black.

most to the NNW-SSE direction, which may indicate shallower crustal fault characteristics. 4.2. Characterizing the style of faulting along lineament zones Fault types were identied for all the derived lineaments as shown in Fig. 11. In the Mediterranean Sea, marked with the subduction of the African and European plates, lineaments with strikeslip and reverse faulting types are dominant. Reverse faults are of marked appearance south of Crete close to the Hellenic arc. In the inland areas, lineaments of strike-slip type of faulting showed more abundance, followed by the normal and then the reverse types. Lineaments of various fault types can mix or alternate along the regional tectonic features in the area. The azimuth and angle of dip of various interpreted fault types along lineaments and their composite are plotted on the rose diagrams and the Schmidt nets (lower hemisphere projection) and shown in Fig. 12. The results show that the most abundant type is the strike-slip type, followed by the normal and then the reverse types. The lineaments of strikeslip and normal fault types are commonly vertical or nearly vertical, while the reverse types show wide dip variations in the range of 2090. The strike-slip type is abundant relative to the normal and reverse types along the NE, NNE-SSW, and NNW-SSE trends. Strike-slip and normal faulting are predominate and contribute nearly equally to the type in the NNE-SSW and NNW-SSE directions. Fault types along the E-W direction vary with location. As for the dip direction, normal fault types show a distinctive south-

easterly dipping, while reverse fault types are concentrated in the opposite (northwesterly) direction. These results agree well with the tectonic framework in the region as discussed in the next section. 4.3. Geologic characterization of the tectonic trends The exceptional continuity (Table 2b), co-linearity, and orientation/positional congruence among the topographic, gravity, and magnetic lineament zones (Fig. 9) and their prominent trends across the region indicate a common source of origin. The origin is likely lithosphere dynamics that acted during the Neoproterozoic era and have been repeatedly reactivated since that time. A noteworthy point is the coincidence with recent (19002006) strong (>5 Mb) earthquake epicenters, which are quoted from the databases of the National Earthquake Information Center (NEIC, 2009) and the Egyptian National Seismograph Network (ENSN), along the extent of the lineament zones and at their intersections, in particular in the Mediterranean Sea area (Fig. 13). These results, underpinned with the persistence of the trends over several depositional phases (Keely, 1994; Guiraud and Bosworth, 1999) and the geodynamic evolution of the region, strongly support the repeated reactivation of tectonism along such deep lineaments to accommodate the subsequent strain, to control the upward propagation of fractures into the younger rocks, and to have acted as the loci for the development of the younger rift-related geological structures (Klitzsch, 1986; Daly et al., 1989; Meshref, 1990; Keely, 1994;

A.A. Masoud, K. Koike / ISPRS Journal of Photogrammetry and Remote Sensing 66 (2011) 818832

829

Fig. 12. Rose diagrams (left) and Schmidt nets (right) for the lineaments of three fault types: (a) normal, (b) reverse, (c) strike-slip, and (d) all the lineaments summing (ac). In rose diagram of (d), the colors in each direction sector denote the ratios of the three kinds of lineaments: strike-slip and normal types are signicant in most direction sectors.

Guiraud and Bellion, 1995; Moustafa, 1997; Unrug, 1997; Guiraud and Bosworth, 1999; Bojar et al., 2002; Abdeen and Greiling, 2005; Bumby and Guiraud, 2005). In order to highlight the correspondence between the lineaments (and associated prominent trends) with signicant tectonic elements in the region, co-linear, continuous lineaments were selected from the concentrated lineament zones in Fig. 9. Then, more regional zones passing over the study area were delineated.

Namely, we manually transformed overlapping long lineaments into one lineament at rst, and connected several lineaments having similar directions by a gentle curve. Resultant Fig. 13 identies ve zones and associated trends that characterize crucial plate boundary and intra-plate features including basin boundaries and intra-basin fractures. The ve major fault trends (tectonic lineaments) extrapolated from the concentrated co-linear long lineaments and the fault types are compared to referenced structural

830

A.A. Masoud, K. Koike / ISPRS Journal of Photogrammetry and Remote Sensing 66 (2011) 818832

Fig. 13. Five major fault trends (tectonic lineaments) interpreted from the concentrated co-linear long lineaments. Recent epicenters (19002006) with relatively large magnitude (M > 5), which were derived from the NEIC and ENSN databases, are overlaid with trends on the geological map of Fig. 7. Color legend of the geological map is shown in Fig. 7. Arrows denote the sense of lateral movement reviewed from literatures.

features. This comparison is summarized below in order of the reactivation period, starting from the oldest: (1) Najd Fault System (NFS) trend (WNW-ESE to NW-SE), positionally corresponds with the magnetic and gravity lineaments, represents principally Late Precambrian sinistral strike-slip transpressive shear zones (Stern, 1985). Another distinctive representative of the NFS trending system is its extraordinary continuity from the southern Sinai to South Cairo, Wadi Natrun, and the Northwestern Mediterranean Sea (Fig. 13). Longacre et al. (2007) demarcated this fault zone in the offshore as a sinistral ocean-continent transform boundary that separates the ocean crust of the southern Tethys from the mildly-extended continental crust of northern Egypt. (2) The Trans-African Lineament (TAL) trend (NE-SW) correlates with the Tethyan shear zones (Keely, 1994) and the Pelusium Line (Neev, 1977; Neev et al., 1982). The variable senses of strike-slip movement induced transtention and transpression stress elds, by which the TAL-parallel lineaments became active during the Tethyan rifting (Keely, 1994). This TAL trend demarcates also the Red Sea extensional faults that have continued to be active from Tertiary to Recent times (e.g. Cochran and Martinez, 1988).

(3) Red SeaGulf of Suez trend (NNW-SSE) correlates with the proto-Clysmic or Erithrean fractures of Keely (1994). The NNW-SSW trending lineaments chiey extracted from the topographic and gravity data are notably situated in the Early Cretaceous deposits (Fig. 12). This corresponds with the fact that the Red SeaGulf of Suez rifting likely commenced in the Cretaceous period (Makris and Rihm, 1991) and reached its climax in the Oligocene period, predominantly controlling the linkage of rift-parallel faults in the Gulf of Suez (Younes and McClay, 1998; Guiraud and Bosworth, 1999). (4) The Gulf of AqabaDead Sea trend (NNE-SSW) demarcates the sinistral transtensive Gulf of AqabaDead Sea fault zone and associated transfer or linking faults which have been active from the Middle Miocene to Recent periods (Guiraud and Bosworth, 1999). A mixture of normal and sinistral strike-slip movements was conrmed along the western margin of the Gulf of Aqaba by Ben-Avraham (1985), which agrees well with the estimated fault types in the present study. (5) Eastern Mediterranean Basin (EMB) related-trends are represented generally by a perpendicular conjugate set of zones oriented nearly E-W resulting from compressional stress elds and N-S resulting from extensional ones associated

A.A. Masoud, K. Koike / ISPRS Journal of Photogrammetry and Remote Sensing 66 (2011) 818832

831

with the northward movement of Africa and still active till today. Lineaments of these trends were derived jointly from the gravity and magnetic anomalies in the Mediterranean offshore area dominated by reverse and strike-slip types (Fig. 11). E-W striking lineaments are commonly concentrated near the southern boundary of the EMB, while the less common N-S oriented lineaments represent intra-basin features. The EMB trends agreement with the well-known Hellenic and the Cyprean Arc (See Fig. 7) is striking. 5. Conclusions Sophisticated techniques for digital processing and integration of remotely-sensed data at various scales over wide compositional and depth variations enabled, for the rst time, successful automatic mapping of signicant tectonic lineaments that agree well with known tectonic fabrics in the region. Three methods were developed: ATMDS for enhancing line features in the grid data; revised STA for extraction and concatenation of line segments; and a technique for modeling fault types along lineaments. ATMDS proved to be effective because the line segments were extracted thoroughly regardless of the gradient of slopes in the grid data. Further, it is an automated one-step process that saves time and efforts and is not biased by the subjectivity that may arise from the interactive setting of the azimuth and the altitude of one-directional source of illumination and the step-wise superimposition of the resulting shaded grids from several directions. By comparing the three edge detectors founded on typical principles, the high capability for detecting line segments that satised the criterion of straight linearity along valley features irrespective of the shade intensities with STA was demonstrated using the shaded grid data. Long lineaments were represented by concatenating the short line segments derived from the grids and smoothing the connection into a gentle curve using B-spline. Effectiveness of estimating orientations and fault types (normal, reverse, and strike-slip) using the lineaments was also demonstrated by comparison to regional fault structures. A case study of the adequacy of applying the developed automated techniques to the area covering Egypt and adjacent areas demonstrated their effectiveness in identifying large-scale tectonic structures based on integration of lineaments extracted from topography, gravity, and aeromagnetic grid data. The continuous lineaments corresponded well, both in position and orientation, with the ve referenced signicant structural features and can be used to update them: the Najd Fault System, Trans-African Lineament, the proto-Clysmic or Erthrean fractures, the Gulf of AqabaDead Sea fault zone, and Eastern Mediterranean Basin structures, which are plate boundaries and intra-plate features of shear zones, boundaries of basins, and intra-basin fractures. The continuities of the ve structures and the faulting types were brought out in more detail by this study. The tectonic signicance of lineaments was conrmed by the concentration of the epicenters of recent (19002006) large seismic activity (M > 5) along the broad lineament zones and at their intersections. In spite of the differences in the main geologic attribute as well as the varied depth range of the three grid data sources, the correspondence of the lineaments from all three sources over much of the region implies the persistence of the prominent trends over wide horizontal and vertical ranges and provides strong support for a structural inheritance model. In this model, basement structural zones have been repeatedly reactivated accompanying the development and upward propagation of the fractures into the Phanerozoic, which was underpinned with the discontinuities in the crusts sediment thickness. The abundance of ve prominent trends (NE-SW, NNE-SSW, NNW-SSE, NW-SE, and WNW-ESE) seems to be related to the frequency of their reactivation periods.

Future research will focus on applying the developed techniques on ne scale various potential eld data where detailed structural information would be available for local limited areas. This can maximize the reliability of the results, in particular for the fault type modeling, and make the validation process much easier. Acknowledgements We are grateful to the anonymous reviewers for their constructive review comments that greatly improved the manuscript. References
Abdeen, M.M., Greiling, R.O., 2005. A quantitative structural study of late PanAfrican compressional deformation in the central Eastern Desert (Egypt) during Gondwana assembly. Gondwana Research 8 (4), 457471. Argialas, D.P., Mavrantza, O.D., 2004. Comparison of edge detection and Hough transform techniques for the extraction of geologic features. International Archives of Photogrammetry, Remote Sensing and Spatial Information Sciences 34 (Part B3), 790795. Austin, J.R., Blenkinsop, T.G., 2008. The Cloncurry Lineament: Geophysical and geological evidence for a deep crustal structure in the Eastern Succession of the Mount Isa Inlier. Precambrian Research 163 (1-2), 5068. Becker, J.J., Sandwell, D.T., Smith, W.H.F., Braud, J., Binder, B., Depner, J., Fabre, D., Factor, J., Ingalls, S., Kim, S-H., Ladner, R., Marks, K., Nelson, S., Pharaoh, A., Sharman, G., Trimmer, R., von Rosenburg, J., Wallace, G., Weatherall, P., 2009. Global bathymetry and elevation data at 30 arc seconds resolution: SRTM30_PLUS. Marine Geodesy 32 (4), 355371. Ben-Avraham, Z., 1985. Structural framework of the Gulf of Elat (Aqaba). Journal of Geophysical Research 90 (B1), 703726. Blinn, J.F., 1977. Models of Light Reection for Computer Synthesized Pictures. Proc. 4th Annual Conference on Computer Graphics and Interactive Techniques, SIGGRAPH 77, 192198. Bojar, A.V., Fritz, H., Kargl, S., Unzog, W., 2002. Phanerozoic tectonothermal history of the ArabianNubian Shield in the Eastern Desert of Egypt: evidence from ssion track and paleostress data. Journal of African Earth Sciences 34 (3-4), 191202. Boucher, R.K., 1995. The relevance of lineament tectonics to hydrocarbon occurrences in the Cooper and Eromanga Basins, South Australia. Journal of Petroleum Exploration Society of Australia 21, 6975. Bui-Tuong, P., 1975. Illumination for computer generated pictures. Communications of the Association of Computing Machinery 18 (6), 311317. Bumby, A.J., Guiraud, R., 2005. The Geodynamic setting of the Phanerozoic Basins of Africa. Journal of African Earth Sciences 43 (1-3), 112. Burns, J.B., Hanson, A.R., Riseman, E.M., 1986. Extracting straightlines. IEEE Transactions on Pattern Analysis and Machine Intelligence 8 (4), 425455. CGMW: Commission for the Geological Map of the World (2006). Structural and kinematic map of the World, 1:50 M scale, A. Haghipour, CGMW/UNESCO Cochran, J.R., Martinez, F., 1988. Structure and tectonics of the northern Red Sea: Catching a continental margin between rifting and drifting. Tectonophysics 150 (1-2), 132. Comanicu, D., Meer, P., 2002. Mean shift: A robust approach toward feature space analysis. IEEE Transactions on Pattern Analysis and Machine Intelligence 24 (5), 603619. Cooper, G.R.J., 2003. Feature detection using sun shading. Computers & Geosciences 29 (8), 941948. Daly, M.C., Chorowicz, J., Fairhead, J.D., 1989. Rift basin evolution in Africa: the inuence of reactivated steep basement shear zones. Geological Society, London, Special Publications 44 (1), 309334. Dix, O.R., Jackson, M.P.A., 1981. Statistical analysis of lineaments and their relation to fracturing, faulting, and halokinesis in the East Texas Basin. Bureau of Economic Geology, The University of Texas at Austin, Austin TX, Report 110, 30. Fichler, C., Rundhovde, E., Olesen, O., Saether, B.M., Ruelatten, H., Lundin, E., Dore, A.G., 1999. Regional tectonic interpretation of image enhanced gravity and magnetic data covering the mid-Norweigian shelf and adjacent mainland. Tectonophysics 306 (2), 183197. Foley, J.D., van Dam, A., Feiner, K., Hughes, J.F., 1990. Computer Graphics: Principles and Practice, 2nd ed. Addison-Wesley, Reading, MA. Grompone von Gioi, R., Jakubowicz, J., Morel, J., Randall, G., 2010. LSD: a fast line segment detector with a false detection control. IEEE Transactions on Pattern Analysis and Machine Intelligence 32 (4), 722732. Guild, P.W., 1974. Distribution of metallogenic provinces in relation to major earth features: Schriftenreihe der erdwissenschaftlichen Kommission der sterreichischen Akademie der Wissenschaften Band 1, pp. 1024. Guiraud, R., Bellion, Y., 1995. Late Carboniferous to Recent Geodynamic Evolution of the West Gondwanian Cratonic Tethyan Margins. In: Nairn, A., Dercourt, J., Vrielynck, B. (Eds.), The Tethys Ocean. The Ocean Basins and Margins. Plenum Press, New York, pp. 101124. Guiraud, R., Bosworth, W., 1999. Phanerozoic geodynamic evolution of northeastern Africa and the northwestern Arabia platform. Tectonophysics 315 (1-4), 73 108.

832

A.A. Masoud, K. Koike / ISPRS Journal of Photogrammetry and Remote Sensing 66 (2011) 818832 Neev, D., 1977. The Pelusium Linea major transcontinental shear. Tectonophysics 38 (3-4), T1T8. Neev, D., Hall, J.K., Saul, J.M., 1982. The Pelusium Megashear System across Africa and associated lineament swarms. Journal of Geophysical Research 87 (B2), 10151030. National Earthquake Information Center (NEIC), 2009. US Geological Survey, http:// neic.usgs.gov. (Accessed 17 June 2009). Oakey, G., 1994. A structural fabric dened by topographic lineaments: Correlation with Tertiary deformation of Ellesmere and Axel Heiberg Islands, Canadian Arctic. Journal of Geophysical Research 99 (B10), 01480227. Omar, G.I., Steckler, M.S., 1995. Fission-track evidence on the initial opening of the Red Sea: two pulses, no propagation. Science 270 (5240), 13411344. Prautzsch, H., Boehm, W., Paluszny, M., 2002. Bzier and B-Spline Techniques. Springer-Verlag, New York. Prechtel, N., 2000. Operational Analytical Hill Shading within an Advanced Image Processing System. In: Proceedings of the 2nd Workshop on High Mountain Cartography, Kartographische Bausteine 18. Technische Universitt Dresden, Dresden, pp. 8598. Rowan, L.C., Bowers, T.L., 1995. Analysis of linear features mapped in Landsat thematic mapper and side-looking airborne radar images of the Reno 1 degree by 2 degree Quadrangles, Nevada and California; implications for mineral resource studies. Photogrammetric Engineering & Remote Sensing 61 (6), 749 759. Rowland, J.V., Sibson, R.H., 2004. Structural controls on hydrothermal ow in a segmented rift system, Taupo Volcanic Zone, New Zealand. Geouids 4 (4), 259 283. Salem, A., Furuya, S., Aboud, E., Elawadi, E., Jotaki, H, Ushijima, K., 2005. Subsurface structural mapping using gravity data of Hohi Geothermal Area, Central Kyushu, Japan. Proceedings World Geothermal Congress, Antalya, Turkey, on CD-ROM. Sandwell, D.T., Smith, W.H.F., 1997. Marine gravity anomaly from Geosat and ERS-1 satellite altimetry. Journal of Geophysical Research 102, 10039-10050. Data is available at http://topex.ucsd.edu/cgi-bin/get_data.cgi. (Accessed 12 June 2009). Sibson, R.H., 1986a. Earthquakes and lineament infrastructure. Philosophical Transactions of the Royal Society, London A 317 (1539), 6379. Sibson, R.H., 1986b. Brecciation processes in fault zones; inferences from earthquake rupturing. Pure Applied Geophysics 124 (1-2), 159175. Smith, W.H.F., Sandwell, D.T., 1994. Bathymetric prediction from dense satellite altimetry and sparse shipboard bathymetry. Journal of Geophysical Research 99 (B11), 2180321824. Stern, R.J., 1985. The Najd Fault System, Saudi Arabia and Egypt: a Late Precambrian rift system? Tectonics 4 (5), 497511. Szen, M.L., Toprak, V., 1998. Filtering of satellite images in geological lineament analyses: an application to a fault zone in Central Turkey. International Journal of Remote Sensing 19 (6), 11011114. Tripathi, N., Gokhale, K., Siddiqu, M., 2000. Directional morphological image transforms for lineament extraction from remotely sensed images. International Journal of Remote Sensing 21 (17), 32813292. Unrug, R., 1997. Rodinia to Gondwana; the geodynamic map of Gondwana supercontinent assembly. GSA Today 7 (1), 16. Warner, T.A., 1997. Integration of remotely sensed geobotanical and structural methods for hydrocarbon exploration in West-central West Virginia. Final report for DOE-sponsored Research Contract DE-FG21-95MC32159. Morgantown, WV, USA, 10 p. Yoli, P., 1967. The mechanisation of analytical hill shading. The Cartographic Journal 4 (2), 8288. Younes, A., McClay, K., 1998. Role of basement fabric on Miocene rifting in the Gulf of SuezRed Sea. In: Proc. 14th Petroleum Conference, October Exploration Vol. 1. Egyptian General Petroleum Corporation, Cairo, pp. 3550. Zhou, X., Dorrer, E., 1995. An Adaptive Algorithm of Shaded-Relief Images from DEMs Based on Wavelet Transforms. Digital Photogrammetry and Remote Sensing 95, SPIE Proceedings Series 2646, 21224. Ziegler, P.A., Roure, F., 1999. Petroleum systems of Alpine-Mediterranean foldbelts and basins, in: Durand, B., Jolivet, L., Horvth, F., Sranne, M. (ed.), The Mediterranean Basins: Tertiary extension within the Alpine orogen. Geological Society, London, Special Publications 156, pp. 517540.

Horn, B.K.P., 1982. Hill shading and the reectance map. Geo-Processing 2, 65146. Keely, M.L., 1994. Phanerozoic evolution of the basins of Northern Egypt and adjacent areas. Geologische Rundschau 83 (4), 728742. Keeley, M.L., Massoud, M.S., 1998. Tectonic controls on the petroleum geology of NE Africa. Geological Society, London, Special Publications 132, 265281. Klitzsch, E., 1986. Plate tectonics and cratonal geology in Northeast Africa (Egypt, Sudan). Geologische Rundschau 75 (3), 755768. Koike, K., Nagano, S., Ohmi, M., 1995. Lineament analysis of satellite images using a Segment Tracing Algorithm (STA). Computers & Geosciences 21 (9), 10911104. Koike, K., Ichikawa, Y., 2006. Spatial correlation structures of fracture systems for identifying a scaling law and modeling fracture distributions. Computers & Geosciences 32 (8), 10791095. Koike, K., Nagano, S., Kawaba, K., 1998. Construction and analysis of interpreted fracture planes through combination of satellite-derived lineaments and digital elevation model data. Computers & Geosciences 24 (6), 573583. Koike, K., Kouda, R., Ueki, T., 2001. Characterizing fracture systems of Kyushu, southwest Japan through satellite-image derived lineaments superimposed on topographic and lithologic data. Bulletin of the Geological Survey of Japan 52 (9), 405423. Kudo, T., Yamamoto, A., Nohara, T., Kinoshita, H., Shichi, R., 2004. Variations of gravity anomaly roughness in Chugoku district, Japan: relationship with distributions of topographic lineaments. Earth Planets Space 56 (5), e5e8. Longacre, M., Bentham, P., Hanbal, I., Cotton, J., Edwards, R., 2007. New Crustal Structure of the Eastern Mediterranean Basin: Detailed Integration and Modeling of Gravity, Magnetic, Seismic Refraction, and Seismic Reection Data. EGM Innovation in EM, Gravity and Magnetic Methods: a new Perspective for Exploration, Capri, Italy, April 1518. Makris, J., Rihm, R., 1991. Shear-controlled evolution of the Red sea: pull apart model. Tectonophysics 198 (2-4), 441466. Masoud, A.A., Koike, K., 2006. Tectonic architecture through LANDSAT-7 ETM+/ SRTM DEM-derived lineaments and relationship to the hydrogeologic setting in Siwa Region, NW Egypt. Journal of African Earth Sciences 45 (45), 467477. Masoud, A.A, Koike, K., 2009. Use of Bzier and B-Spline Curves for Smart Representation of Lineaments Auto-detected from DEM. International Association of Mathematical Geology (IAMG09): Computational Methods for the Earth, Energy and Environmental Sciences, Stanford University, California, USA, 2328 August, (on CD-ROM). Masoud, A., Koike, K., Teng, Y., 2007. Geothermal Reservoir Characterization Integrating Spatial GIS Models of Temperature, Geology, and Fractures. Proc. 12th Conference of International Association for Mathematical Geology, Beijing, China, August 26-31, (on CD-ROM). Maus, S., Barckhausen, U., Berkenbosch, H., Bournas, N., Brozena, J., Childers, V., Dostaler, F., Fairhead, J.D., Finn, C., von Frese, R.R.B., Gaina, C., Golynsky, S., Kucks, R., Lhr, H., Milligan, P., Mogren, S., Mller, D., Olesen, O., Pilkington, M., Saltus, R., Schreckenberger, B., Thbault, E., Caratori Tontini, F., 2009. EMAG2: A 2-arc-minute resolution Earth Magnetic Anomaly Grid compiled from satellite, airborne and marine magnetic measurements. Geochemistry, Geophysics, Geosystems 10 (Q08005). doi:10.1029/2009GC002471. Maus, S., Sazonova, T., Hemant, K., Fairhead, J.D., Ravat, D., 2007. National geophysical data center candidate for the world digital magnetic anomaly map. Geochemistry, Geophysics, Geosystems 8 (Q06017). doi:10.1029/ 2007GC001643. McNutt, M., 1979. Compensation of oceanic topography: an application of the response function technique to the Surveyor area. Journal of Geophysical Research 84 (B13), 75897598. Meer, P., Georgescu, B., 2001. Edge detection with embedded condence. IEEE Transactions on Pattern Analysis and Machine Intelligence 23 (12), 13511365. Meshref, W.M., 1990. Tectonic Framework. In: Said, R. (Ed.), The Geology of Egypt. A.A. Balkema Publishers, Rotterdam, Netherlands, pp. 113156. Milbury, A.E.C., Smrekar, S.E., Raymond, C.A., Schubert, G., 2007. Lithospheric structure in the east region of Marsdichotomy boundary. Planetary and Space Science 55 (3), 280288. Morris, K., 1991. Using knowledge-base rules to map the three-dimensional nature of geologic features. Photogrammetric Engineering & Remote Sensing 57 (9), 12091216. Moustafa, A.R., 1997. Controls on the development and evolution of transfer zones: the inuence of basement structure and sedimentary thickness in the Suez rift and Red Sea. Journal of Structural Geology 19 (6), 755768.

S-ar putea să vă placă și