Mapping-of-Pliocene-Pleistocene-Rock-Units-Using-Enhanced-Thematic-Mapper-Plus-ETM-Case-Study-Wadi-Qasab-Area-South-East-Sohag-Egypt

This study describes several processing methods, applied to a Landsat Enhanced Thematic Mapper Plus image of 2007, to develop a reliable method to discriminate between different exposed

Mapping of Pliocene-Pleistocene Rock Units Using Enhanced Thematic Mapper Plus ETM+: Case Study, Wadi Qasab Area, South East Sohag, Egypt

Conference Paper · November 2013

CITATIONS

3

READS

712

4 authors, including:

Some of the authors of this publication are also working on these related projects:

geomorphology View project

Bosy A El-Haddad

Sohag University

11PUBLICATIONS    187CITATIONS    

SEE PROFILE

Dr Ahmed M Youssef Saudi Geological Survey

105PUBLICATIONS    3,671CITATIONS    

SEE PROFILE

All content following this page was uploaded by ِDr Ahmed M Youssef on 21 March 2014.

MAPPING OF PLIOCENE-PLEISTOCENE ROCK UNITS USING ENHANCED

THEMATIC MAPPER PLUS ETM+: CASE STUDY, WADI QASAB AREA,

SOUTH EAST SOHAG, EGYPT

and Abdel Hammed El-Sharter*

*Geology Department, Faculty of Science, Sohag University, Egypt

**Geological Hazards Department, Applied Geology Sector, Saudi Geological Survey,

P.O Box 54141, Jeddah 21514, KSA

ABSTRACT Field work for discrimination between different rock units is a very tedious one and a time consuming However, by using remote sensing techniques the differentiation become easier and time save The current work is very important in the field of sedimentary rocks, especially because few authors used remote sensing techniques in sedimentary rock discrimination and can provide a good background for more work This study describes several processing methods, applied to a Landsat Enhanced Thematic Mapper Plus image of 2007, to develop a reliable method to discriminate between different exposed Pliocene - Pleistocene sedimentary rock units in Wadi Qasab area The current area in the Nile Valley was formed during the different stages of the Nile evolution and the stratigraphy of these rock units is complicated By using different enhancement techniques with the help

of a supervised classification method, it was found that principal component analysis and minimum noise fraction are the most suitable methods to discriminate between different types of Pliocene – Pleistocene sediments, as well as to map these sediments from the surrounding rock units Field investigation was used to verify the remote sensing findings

Keywords: ETM+, Sedimentary rocks, principal component analysis, minimum noise fraction, Egypt

1.1 Study area

The Study area (Wadi Qasab area) is situated in the low desert zone, South East of Sohag Governorate, Egypt, in the midway between Cairo and Aswan (Figure 1) The study area is located between the Latitudes 26o 12' 00'' and 26o 29' 30'' N and between the Longitudes 31o 68 00'' and 32o 14' 00'' E

P-P VIII-1 – VIII-13 (NOV 2013) ASSIUT-EGYPT

Figure (1): (A) Egypt map, (B) Sohag - Qena area, and (C) Study area (Wadi Qasab area)

1.2 Remote sensing background

Remote sensing instruments measure reflected or emitted radiation in the visible, near Infrared, thermal infrared or microwave portion of the electromagnetic spectrum to obtain information about the earth’s surface from a distance Satellites images have long been used as an effective exploration tool that can be used on detection of associated hydrothermal minerals, structural elements and lithological mapping (Jensen, 2000; Drury, 2001; Gupta, 2003) The spectral resolution of the new remote sensing data such

as ETM+ can help in differentiation of varieties of lithological units Landsat data has been used previously in arid and semi arid environments to locate areas with iron oxides and/or hydrous minerals (Abram et al., 1983; Tangestani and Moore, 2001) The arid and semi arid environment are suitable for application of remote sensing data for lithological mapping due to poor vegetation cover Youssef, et al., 2009; and Youssef 2008a&b used remote sensing technique for discriminating different sedimentary rock units In general,

it is hard to differentiate between the sedimentary rock units due to their similarities in chemical and mineral composition

1.3 Aim of the study

In this research a trial was made to use ETM+ data for discrimination between Madamud Formation which belong to Pliocene (Paleo-nile Phase) and Pleistocene rock units Issawia, Armant, and Qena Formations (which are represent Desert and Pre-nile phases respectively) and to help in understanding the sequence of sedimentation during

the Nile evolution

2 GEOLOGICAL SETTING OF THE STUDY AREA

The study area is characterized by presence of several sedimentary facies that could be recognized as in Figure (2) and Table (1) The current paragraphs will discuss in detail the geological setting of the area

Figure (2): Geological map of Wadi Qasab area (Modified after Mahran, 1993)

Table (1): Stratigraphic sequence of the study area

Pleistocene

low-energy environment (Omer and Issawi, 1998)

quartzite and siliceous sandstones, and it's size varying from 2 to 20cm in diameter all embedded in a matrix of

reddish-brown soil

metamorphic fragments (Said, 1981)

Late Pliocene /

Early

Pleistocene

Early Pliocene

Armant /Issawia

Madamud

Clastic facies at the lake margins and carbonate facies in

the central zones (Said, 1971)

Bedded brown and gray clays intercalated with thin beds and lenses of silt and fine sand, and fluviatile -dominated sediments made up of sand, silt and mud intercalations

(Omer and Issawi, 1998) 2.1 Pliocene sediments

2.1.1 Madamud Formation

Madamud Formation can be described as fine grained silt and clay beds of chocolate brown color followed by a bed of alternating lamination of fluvial fine grained sand and silts and topped by a thin bed of grayish brown calcareous clay (Said 1981) It’s thickness ranges between 6m and 9m mineral composition is mainly Montmorillonite with little Kaolinite and some accessory minerals including quartz, biotite, muscovite, pyrite,

epidote and zircon

2.2 Pleistocene sediments

2.2.1 Armant Formation (Early Pleistocene)

Armant Formation is made up of alternating beds of locally derived gravels and fine grained clastic rocks The gravel beds are cemented by tuffaceous materials and the pebbles are subangular and poorly sorted The fine grained clastic beds are calcareous, sandy argillaceous, or phosphatic In other places, the Armant Formation is made up of alternating pebbles beds, marl, and horizontally bedded travertine with plant reeds The succession includes unidentified plant remains most probably belonging to Pliocene In the study area Armant Formation composed of sandstone, siltstone, mudstone and shale, accumulation in the central core, changes laterally eastward into coarse grained sandstones interfingring with sandstones and conglomerates These lacustrine facies appear to have graded into alluvial fan deposits in which large quantities of coarse clastics come from easterly trending wadis inherited fractures and faults and funneled their loads of coarse clastics as proximal fan deposits filling the upstream parts of wadis (Mahran, 1993) Armant Formation and Issawia Formation represent the desert phase in the Nile evolution

2.2.2 Issawia Formation (Early Pleistocene)

Issawia Formation deposits are represented by 12m conglomerates interfingring with coarse breccias The conglomerate sediments accumulated on irregular slope developed

on the Eocene rock as talus (Mahran, 1993) Issawia Formation belonging to the desert Phase in the evolution of River Nile The breccias sediments are massive bedded travertine with minor conglomerate lenses Pebbles of local derivation, the matrix composed of red mud, which are topped in many places by hard red breccias The hard red breccias are quarried for ornamental use in many places along the Nile Valley

2.2.3 Qena Formation (Middle Pleistocene)

Qena sands form an exceedingly uniform and skirts the eastern and western bank of the River Nile in Upper Egypt Qena Formation is mainly composed of sands and gravels lacking fragments of basement complex parentage to reflect proximal source, where neither (or negligible) igneous nor metamorphic pebbles are recorded The mineralogical characteristics of these sediments indicate that they are mostly derived from the Paleozoic and Mesozoic sedimentary rocks distributed in the Eastern Desert (Omer, 1996) They overlaid by the Abbasia gravel According to (Said, 1981), Qena sand belongs to the Pre-Nile stage of the River Nile evolution They are made up of massive cross bedded quartzose sandstone unit with specks of feldspar grains

2.2.4 2.2.4 Abbassia Formation (Middle Pleistocene)

Said (1981) described Abbassia Formation as pebbles of igneous rocks, quartzite and siliceous sandstones, and its size varying from 2 to 20 cm in diameter all are embedded reddish-brown sandy matrix In the study area the Abbassia Formation overlay Qena sands as coarse conglomerates come from reactivated drainage wadis and fractures in the east exhibiting proximal fan subfacies, which was graded westward into fine conglomerates interfingring with cross bedded sandstones and siltstones Abbassia Formation represents the Pre-Nile stage of the river Nile evolution

2.2.5 Dandara Formation (Late Pleistocene)

In the study area Dandara Formation is represented by flood plain deposits composed

of an alternation fine siliclstic beds (siltstone, claystone, and fine grained sandstones) were accumulated on the bank of the River Nile in the east and west side Toward the east, in local inland lakes, laminated fine siliclastics (siltstone and claystone) and stromatolitic limestone and lime mud were laid down (Mahran 1993) Dandara Formation represents the Neo-Nile stage of the Nile evolution

3 METHODOLOGY AND RESULTS 3.1 Materials used in the current study

The material used in this study includes Landsat 7 ETM+ imagery Path175, Row 42, acquired on September 2007 Landsat ETM+ is multi-spectral remote sensing data that have four bands in the Visible and Near Infrared (VNIR) regions of the electromagnetic spectrum with 30 m spatial resolution (visible blue = band 1; green = band2; red = band 3; and NIR= band 4); two bands in the Shortwave Infrared (SWIR) region with 30 m spatial resolution (bands 5 and 7), one band in the Thermal Infrared (TIR) region with 60

m spatial resolution (band 6), and one panchromatic band 8 with a 15 m spatial resolution The image is cloud free and geometrically corrected to a UTM Zone 36 North and WGS84 datum

For this purpose a fused image, from the panchromatic and the multi-spectral channels, has been generated On the other hand, the data of Enhanced Thematic Mapper (ETM+) with a resolution of 15 m have been used The main technical characteristics and specification of the sensors used with ETM is shown in (Table 2)

Table (2): Technical characteristics of ETM+ data

Satellite

sensor

Spectral Bands

Spatial resolution

Average revisiting time; viewing angle/ swath

width

Enhanced

Thematic

Mapper

Band 1, 2, 3, 4,5,7 Band 6 Band 8

30 m

60 m Panchromatic 15 m

16 days

183 km wide swath

at an altitude of 705 km

The image processing has been done using the “Environment for Visualizing Images” (ENVI 4.6) software The ETM+ image with 15 m spatial resolution was subset to focus

on the study area The results have been verified with the field investigation (many field trips between 2012 and 2013 to get a good understanding about the extent of each unit in the study to help in verifying the remote sensing results) The results are compared with the geological map of the area Figure (2)

3.2 Methods used in the current study

In the current study different image processing have been applied to help in the rock unit discrimination in the study area, including: 1) True and false color composite (band

3, 2, and 1 in RGB) and (band 7, 4, and 2 in RGB) respectively 2) Several data enhancement techniques were employed in this analysis, including Principal component analysis (PCA) and Minimum Noise Fraction (MNF) Also, image classification has been applied for the PCA and MNF images

3.2.1 Principal Component Analysis (PCA)

Principal Component Analysis (PCA), in each one can use either standard or selective analysis, the difference being that in the selective analysis; only certain bands are chosen (Crosta and Moor, 1989) The PCA technique indicates whether the materials are represented by bright or dark pixels based on the sign and magnitude of the eigenvectors

It is a linear procedure to find the direction in input space where most of the energy of the input lies In other words, PCA performs feature extraction The projections of these components correspond to the eigenvalues of the input covariance matrix The principal component analysis is performed first, and then the eignvector loading is trained to generate image that give the information of spectral bands for easy interpretation

The reason for this is that the PCA faster since eigenvalues are stable Natural color and especially the standard false color infrared images don’t show much variation in rock

colors It is often difficult to discern rock contacts, let alone identification the rock type Various computer based techniques have been developed that allow a greater variation in the observed colors in non-standard false-color rendition The Crosta technique is based

on PCA Through the analysis of the eignvector values (Crosta and Moore, 1989), it allows identification of principal components that contain spectral signature about specific minerals as well as contribution of each of the original bands to the components

in relation to spectral response of the material interest

3.2.2 Minimum Noise Fraction (MNF)

Minimum Noise Fraction (MNF), is a technique in which transformation is used to determine the inherent dimensionality of image data especially in hyperspectral data, to segregate noise in the data, and to reduce the computational requirement for subsequent processing (Boardman and Kruse, 1994) The MNF transform is essentially two cascaded principal component transformation (Green et al., 1988) where the first transformation based on an estimated noise covariance matrix decorrelates and scales the noise in the data This first step results in band-to-band correlation The second step is standard principal component transformations of the noise-whitened data For further spectral processing, the inherent dimentionality of the data is determined by examination of the final eigenvalues and the associated images The data can be divided into two parts: one part associated with large eigenvalues and coherent eigen images, and a complimentary part with near unity eigenvalues and noise dominated images By using only the coherent portions, the noise is separated from the data, thus improving the spectral processing results

3.2.3 Image Classification

Image classification is the technique of assigning pixels of the image to classes based

on the spectral reflectance characteristics of each image pixel The parallelepiped method was used to implement the supervised classification Richards (1999) mentioned that parallelepiped classification uses a simple decision rule to classify multispectral data The decision boundaries form an n-dimensional parallelepiped classification in the image data space The dimensions of the parallelepiped classification are defined based upon a standard deviation threshold from the mean of each selected class If a pixel value lies above the low threshold and below the high threshold for all n bands being classified, it is assigned to that class If the pixel value falls in multiple classes, ENVI assigns the pixel

to the last class matched Areas that do not fall within any of the parallelepiped classes are designated as unclassified The supervised classification, parallelepiped, was used on the both PCA and MNF images

3.3 Results of Remote Sensing Application

Generally, the most significant advantage of multispectral imagery is its ability to detect the differences between surface materials by combining their spectral bands in cases where, within a single band, different materials cannot be discriminated and may have the same appearance On the other hand, by using specific band combinations, different materials might be contrasted against their back-ground (Sedlak, 2002) A True Color Composite (TCC); False Color Composite (FCC); Principal Component Analysis

(PCA); and Minimum Noise Fraction (MNF) have been applied on four alluvial fans (Figure 3) In addition, training samples were selected from the PCA and MNF images using the region of interest technique

Figure (3): Four Fans have been used for the current analysis

3.3.1 True and False Image Analysis

A true color composite image (bands 321 in RGB) has been prepared for the study area (Four selected fans) (Figure 4) On the other hand, a false color composite image (bands 742 in RGB) is generated (Figure 4) The results of this data analysis for TCC and FCC do not show any distinct features to discriminate between Pliocene deposits (Madamud Formation) and Pleistocene deposits (Armant, Qena, Abbassia, and Dandara Formations)

Figure (4): True and False color composites of Landsat ETM+ bands 742 in (RGB) on

selected four fans of the study area

3.3.2 Principal Component Analysis (PCA) for Lithological Investigation

The PCA technique was applied on six ETM+ bands 1, 2, 3, 4, 5, and 7 of the study area (Four selected fans) to enhance the spectral variability of the content on the image The result of this enhanced PCA method indicates that the Pliocene - Pleistocene rock units (Madamud Formation, Armant Formation, Qena Formation, Abbassia Formation, and Dandara Formation) can be differentiated from each other Figure (5.a1) shows the PCA image bands 412 on RGB that indicate different rock units as follow; 1= Armant Fm., with yellow color; 2 = Abbassia Fm., with green color; 3 = Qena Sand, with dark violet color; 4 = Madamud Fm, with pale orange; and 5 = Dandara Fm with pink color

To verify these results, the supervised classification (Parallelepiped method) was carried out on the PC image (bands 412 in RGB) Figure (5.a2) The training samples were selected on the different rock units based on the PC image, as well as from field observations Using the supervised classification method on the PCA image, it is easy to distinguish between these rock units (Figure 5.a2) in which, Armant Fm appears in red color, Abbassia Fm in green color, Qena sand in blue color, Madamud Fm in yellow color, and Dandara Fm in cyan color