CC BY-NC-ND
0Available online at www.sciencedirect.com
ScienceDirect
Procedía Engineering
CrossMark
ELSEVIER
Procedía Engineering 154 (2016) 936 - 942
www.elsevier.com/locate/procedia
12th International Conference on Hydroinformatics, HIC 2016
Mohammed Esmaila*, ALI Masriab , Abdelazim Negm c
a Environmental Engineering Department, School of Energy and Environmental Engineering, Egypt-Japan University of Science and Technology, E-JUST, P.O.Box 179, New Borg Al-Arab City, Postal Code 21934, Alexandria, Egypt. b Lecturer, Irrigation and Hydraulics Department, Faculty of Engineering, Mansoura University, Mansura, Egypt. C Head of Natural Resources Lab., Environmental Engineering Department, Egypt-Japan University of Science and Technology, E-JUST, Alexandria, Egypt (Seconded from Faculty of Engineering, Zagazig University, Zagazig, Egypt)
Land use/land cover (LULC) change is considered one of the most important signals of regional environmental changes. In this study, supervised classification and post-classification change detection techniques were applied to Landsat images acquired in 1987 and 2015 to map LULC changes along the north part of the Nile delta coastal zone specifically, at Damietta promontory. Landsat images were radio-metrically and geometrically corrected, and then, multi-temporal post-classification analysis was performed to detect LULC changes rates of the Nile delta coast around Damietta promontory where the Damietta branch (of the Nile River) meets the Mediterranean Sea.
Four categories including seawater, developed (agriculture and urban), and undeveloped areas were selected to evaluate their temporal changes by comparing the processed images. The objective of this study is to map and assess the rate of changes LULC changes at the study area which can help the decision makers to replan the use of natural resources efficiently.
By applying remote sensing/GIS technique, areas of rapid change are identified and targeted for more detailed monitoring in the field. Changes among different LULC classes were assessed. The results show that the sew water and urban area were increased.
© 2016 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.Org/licenses/by-nc-nd/4.0/).
Peer-review under responsibility of the organizing committee of HIC 2016
* Corresponding author. E-mail address: mohammed.esmail@ejust.edu.eg
1877-7058 © 2016 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.Org/licenses/by-nc-nd/4.0/).
Peer-review under responsibility of the organizing committee of HIC 2016
doi:10.1016/j.proeng.2016.07.515
Abstract
Mohammed Esmail et al. / Procedia Engineering 154 (2016) 936 - 942 Keywords: Change detection; Land use/land cover; Remote sensing; coastal zone; Damietta, Nile River
1. Introduction
Sustainable development at coastal zone is urgent nowadays worldwide.Consequently, monitoring is is required to reach the sustainability, and ecological protection. The LULC changes have become an essential parameter in the recent strategies to achieve the integrateds coastal zone management that have a great impact on economic development and natural resources managment [1]. Similar to other delta as worldwide, the Nile Delta has been experiencing significant coverage due to natural and man-made intervention. It has been changed from an acccreted zone by the action of sediments transported from the Nile river to the coastal zone to an eroded area after building different water control structures along the Nile, from from South at Aswan to North at the lower Delta, i.e, Aswan Low Dam in 1902 and the Aswan High Dam (AHD) in 1964 [2]. The barrages and dams have a marked control on the magnitude of peak discharges from the river and have virtually eliminated sediment delivery to the coast since its construction[3]. Consequently, an observable change in the land cover of the coastal zone has been occurred during last decades.
The Damietta area has been subjected to large landscape changes and reformations at recent decades. The area still under development and is expected to experience some more changes during the next few years. This demonstrates the urgent need for assessment of the present coverage pattern and land-uses of this area. The present study attempts to determine details of LULC information of the Damietta estuary and surrounding area using high resolution remote sensing Images. Land cover types and surface characteristics were described and their areas estimated and mapped using the standard remote sensing methods[4].
Digital change detection is the process of describing changes in land use properties. It is considered one of the recent important tools in acquiring free data for to manage natural resources in various field. Among different field, it has been widely used in coastal area for monitoring and resources management . In addition, Geographical Information System (GIS) has been applied integrally with remote sensing to estimate the evolution and discover the different changes occurred during a certain helping to make decisions [5]. The simple access to various satallite images from various satellite platforms have helped to generate information on varied aspects of the coastal and marine environment. Remote sensing technique has widely been used in environmental change detection studies. In these studies, different tools have been used in an integrated way to formulate a valuable management action plans [6]. For monitoring coastal area, various methods has been used to detect changes via digital images obtained from different satallite platforms. The most familiar change detection techniques include simple band ratio, principal component analysis, vegetation index algorithms, and others. Major data sources include MSS, TM, ETM, System Pour l'Observation de la Terre (SPOT), moderate resolution imaging spectroradiometer (MODIS), advanced very high resolution radiometer (AVHRR), and other sensors [7]. Post classification change detection is a perfect technique applied to address thematic changes in satellite data. The advantage of using this approach is its applicability to data from different sensors with a high degree of confidence [8]. In addition, post-classification statistics proved to be an effective method as it can analyze the coastal processes and develop reliable models to predict shoreline trend represented in eroded and accreted areas, This is because datasets from two or more dates are independently classified, thereby minimizing problems arising from atmospheric correction and/or sensor degradation [9]. Analysis of different remote sensing images, including Landsat MSS, TM, ETM+ and Spot satellite along parts of the Nile Delta has identified areas of erosion followed by beach accretion. Different studies have been carried out to assess the land cover change along Damietta area [3]. For example, a previous study [9] reported that, a severe change has been occured as a result of agricultural and tourist development projects. These changes led to vegetation degradation and water logging that area. Another study [10], found that barren lands at the north of Damietta have poor potentiality for agriculture. These lands could be utilized for other land use, such as residential, industrial or for touristic activities. While those located east of Damietta could be utilized for aquaculture or as bioremediation facilities for biological treatment of agricultural wastewaters. On the other hand, the current study aims to use the RS/GIS to monitor the
changes in land use over the period from t1987 to 2015 to help decision making to set the right policies and take the right decision to protect the area from further unwanted changes.
2. Materials and Methods
2.1. Study area
The study area of the Damietta Promontory area lies on the north-eastern Nile delta coast and extends about 11.0 km east, 11.0 km west, 5.0 km north, and 12.0 km south referenced to the tip of the promontory,(Figure 1). gThe promontory is one of the most vulnerable areas as it suffers from a continous erosionIt is considered a vital area as it used for transportation, agriculture, and fishing activities. It suffers from different coastal problems represented in accretion inside the inlet, and shoreline erosion that began after the construction of the Aswan High Dam [11].
Fig. 1. Study Area, Damietta Promontory area at the terminal of Damietta Branch, Land-sat
image, 2015.
2.2. Datasat
The image data was obtained at unequal intervals between 1987 and 2015, covering a time span of 28 years. The images have been acquired in summer season and in good quality, with no effective clouds. Four sets of satellite images were used as presented in Table 1: two from the Landsat TM (30 m spatial resolution) acquired in 1987 and 1999; two Landsat ETM (30 m spatial resolution) acquired in 2003 and 2015.
Table 1. Landsat images used in this research.
Acquired Date SpaceraftID/Sensor Path/Row Pixel Size (m) Coordinate System/Datum Zone
Apr 22, 1987 Landsat_5/TM 176/38 30 UTM/WG84 36
Sep 06, 1999 Landsat_7/TM 176/38 30 UTM/WG84 36
Feb 05, 2003 Landsat_7/ETM 176/38 30 UTM/WG84 36
Nov 29, 2015 Landsat_8/ETM 176/38 30 UTM/WG84 36
2.3. Image Pre-Processing
A first step in the pre-processing is to check all images any defects such as striping. Then, all images were clipped to focus on our study area. After that, all images were corrected geometrically and radiometrically. At last, all images were stacked and classified. Image processing has been implemented using ENVI software version 4.8.
2.3.1 Geometric Correction
Geometric correction of the different satallite images is of great importance for change detection since the potential exists for registration errors to be interpreted as LULC change, leading to an overestimation of actual change[12]. Based on the metadata documentation, the TM, and ETM+ images used in the present investigation are ortho-rectified products. They are indeed in the World Geodetic System (WGS84) datum and the Universal Transverse Mercator (UTM) projection system[3] .
Using ENVI 4.8 software, the Satellite map from Google earth software (figure.2) was used as a base map to geo-reference the Landsat image acquired in 2015 through image to map referencing. Then, the Landsat TM image in 2015 was considered as the master image that was utilized to register the other images through image-to-image registration. To accurately register each image, about 20 ground control points (GCP) was examined and matched with all images. These points includedthe most distinguished geomorphologic features. the calculated route mean square error (RMSE) was found about 0.43 pixels, that ensures a good geo-referencing of the images.
Fig. 2. Points Used in Geometric Correction, Satellite Image Image (Google earth), 2015.
2.3.2 Radiometric Correction
Firstly, the dark-object subtraction method to correct the atmospheric scattering caused by haze, dust or smoke [3]. Then, the current radiometric correction was implemented in one step using radiometric correction tool in ENVI software, which combines the sun and view angle effects, and the sensor calibration with the atmospheric correction. Image Classification
2.4. Image Classification
The overall objective of the image classification procedure is to automatically categorize all pixels in an image into LULC classes or themes. One of the most important applications of satellite remote sensing is to detect changes in land use to discern those areas on digital images that change features of interest between two or more dates [1][3].
The LULC classes are typically mapped from digital remotely sensed data. The spectral signatures of land and water in the region was used to identify each class that already familiar from the different previous studies.
The classification process has been implemented using all spectral bands in the Landsat images (1987, 1999, 2003 and 2015) except the thermal bands .
Based on many trials , supervised classification using support vector machine technique has been implemented to the four images . Four LULC classes have been selected as follows; seawater, agriculture, urban and undeveloped areas. The classified images were then refined using 5x5 majority/minority analysis to remove odd pixels in the classified matrix and reduce noise in the output maps.
Table 2. Points and RMS error used in geometric correction.
Points Map (x, y) Image (x, y) Predict (x, y) RMS
x y x y X y Error
P1 362290.9500 3478062.1400 1417.5000 5046.2500 1417.7800 5046.1200 0.31
P2 382800.8900 3482750.1200 2101.8700 4889.5300 2101.6300 4889.6300 0.26
P3 367061.0300 3464814.8300 1576.5000 5486.7500 1576.5400 5486.9000 0.15
P4 392070.7900 3485032.2900 2410.7400 4813.4100 2410.7600 4813.3700 0.04
P5 393515.0000 3483967.0000 2458.5000 4849.0000 2458.8700 4848.8300 0.41
P6 393254.0400 3482173.3100 2450.1000 4908.5700 2450.1000 4908.5900 0.02
P7 393756.9300 3480116.6000 2466.7500 4977.0000 2466.7900 4977.1000 0.11
P8 393381.2500 3476172.3300 2454.3100 5108.5800 2454.1200 5108.5200 0.20
P9 398002.6800 3476394.7400 2608.4400 5101.0000 2608.1800 5101.0400 0.26
P10 410761.4500 3469354.7200 3032.9500 5335.4800 3033.1400 5335.5700 0.21
P11 408214.2500 3471106.8500 2948.3100 5277.1700 2948.3300 5277.1700 0.02
P12 399040.0400 3474578.9200 2642.7200 5161.5300 2642.6900 5161.5400 0.04
P13 395598.2000 3466524.2600 2527.7100 5429.9900 2527.6500 5429.9600 0.07
P14 391126.7600 3470625.7900 2378.7800 5293.3500 2378.7700 5293.3300 0.02
P15 380994.3100 3465868.9700 2040.6700 5452.0000 2040.9100 5451.8100 0.31
P16 376812.6600 3466287.6800 1901.7200 5438.0000 1901.5600 5437.8700 0.20
P17 373324.0500 3467204.3100 1785.2300 5407.3100 1785.3200 5407.3600 0.10
P18 366358.4900 3466435.8800 1553.0000 5433.0000 1553.1500 5432.9500 0.16
P19 370731.3500 3474875.2800 1699.1400 5151.8800 1699.0800 5152.0400 0.17
P20 369593.5600 3471396.6500 1661.5000 5267.7500 1661.0800 5267.8400 0.43
3. Results and Discussions
Support vector machine algorithm techniques among the other supervised calssification techniques gives a good overall accuracy ranged from 0.97 to 1.0 . The areas were computed for the land cover types in each of the classified images shown table 2 in square kilometres. Table 2 presents the descriptions for the land cover types used for the analysis of the study area.
It is clear from the figure 3. that, seawater area has increased from 240.289 km2 in 1987 to 248.017 km2 in 2015. This means that the eroded land during this period is about the difference between the two areas. In addition, the urban area increased by 1.466%, which led to a decrease in the agriculture area through advancing towards the sea. On the other hand, the undeveloped areas cover decreased due to the urban extension on both sides of the Damietta promontory, and also due to the construction of the coastal protection measures in front of the promontory.
Damietta harbour, also, have some changed in water area as increase on basin to serve the commercial demands. Land area also was changed whereas urban areas in harbour part were increase.
Table 3. Area and percentage of change of different land cover classes of 1984 and 2014 classified images.
Area % Area (km2) Area Change %
Classes —
1987 2015 1987 2015 (1987-2015)
Sea Water 68.407% 71.437% 240.289 248.017 3.030%
Agriculture 15.926%
Undeveloped 05.936%
Urban 09.731%
13.000% 55.940
04.365% 20.851
11.197% 34.182
45.135 -2.926%
15.154 -1.571%
38.875 1.466%
M41 bA 1_UU,«*J
Damietta Promontory (2003) Damietta Promontory (2015)
ItohaiiIW ' iL,Ps=i.i:tauiHi
Sea Water Urban Agriculture Undeveloped
Fig. 3. Land-use supervised-classification for the different Landsat images 1987, 1999 and 2015.
4. Conclusions
Land cover changes have been investigated at Damietta Promontory using Landsat images spacing of 28 years period, acquired in 1987, 1999, 2003, and 2015. These data was geo-referenced using a satellite image, and then corrected geometrically, and radio-metrically. Changes in land cover were detected through supervised classification.
It was found that within the last 28 years, 3.030% of the coastal area at the promontory has been eroded. A considerable increase in urban settlements has taken place by.1.466% was developed. The agriculture areas have been decreased to 2.926%. Damietta harbour was sophisticated and undeveloped areas were decreased.
Acknowledgements
The first author would like to thank Egyptian Ministry of Higher Education (MoHE) for granting him the Ph.D. scholarship. Also, thanks and appreciation is due to Egypt-Japan University of Science and Technology (E-JUST) and JICA for their support and for offering the tools needed for this research.
References
[1] A. A. Alesheikh, A. Ghorbanali, and N. Nouri, "Coastline change detection using remote sensing 1," vol. 4, no. 1, pp. 61-66, 2007.
[2] N. W. Re, "Evolution of rosetta promontory on nile delta coast during the period from 1500 to 2005 , Egypt Evolution of Rosetta Promontory during the Last 500 Years , Nile Delta Coast , Egypt," no. March, 2016.
[3] A. Masria, K. Nadaoka, A. Negm, M. Iskander, N. B. El-arab, E. Informatics, T. De Noronha, and A. Carleton, "Detection of Shoreline and Land Cover Changes around Rosetta Promontory, Egypt, Based on Remote Sensing Analysis," pp. 216-230, 2015.
[4] W. M. Moufaddal, "Surface characteristics and coverage pattern of the new damietta port area, as mapped by high resolution remote sensing," 2007.
[5] S. Ramachandran, R. Krishnamoorthy, S. Sundramoorthy, Z. F. Parviz, A. Kalyanamuthiah, and K. Dharanirajan, "Management of Coastal Environments in Tamilnadu and Andaman & Nicobar Islands based on Remote Sensing and GIS approach. MAEER MIT," Pune Journal, IV (1516), Spec. issue Coastal, Environ. Manag., pp. 129-140, 1997.
[6] A. El-zeiny, "Space-borne technology for monitoring temporal changes along Damietta shoreline ," vol. 4, no. 1, pp. 459-468, 2016.
[7] M. E. Hereher, "Mapping coastal erosion at the Nile Delta western promontory using Landsat imagery," pp. 1117-1125, 2011.
[8] C. Song, C. E. Woodcock, K. C. Seto, M. P. Lenney, and S. A. Macomber, "Classification and Change Detection Using Landsat TM Data : When and How to Correct Atmospheric Effects ?," vol. 4257, no. 00, 2000.
[9] A. Shalaby, A. S. A, and R. Tateishi, "Shalaby A , Tateishi R . Remote sensing and GIS for mapping and monitoring land cover and land- use changes in the northwestern coastal zone of Egypt . Applied Geography Remote sensing and GIS for mapping and monitoring land cover and land-use changes in ," no. March 2016, 2007.
[10] M. I. El-gammal, R. R. Ali, and R. Eissa, " Land use assessment of barren areas in Damietta Governorate , Egypt using remote sensing," Egypt.ScienceDirect, J. Basic Appl. Sci., vol. 1, no. 3-4, pp. 151-160, 2014.
[11] A. R. Zidan, O. S. Rageh, T. Sarhan, and M. Esmail, "Effect of breakwaters on Wave Energy Dissipation (Case Study : Ras El-Bar Beach, Egypt)," Int. Water Technol. J., vol. 2, no. 4, pp. 268-283, 2012.
[12] D. A. Stow, "Reducing the effects of misregistration on pixel-level change detection," Int. J. Remote Sens., vol. 20, no. 12, pp. 24772483, 1999.
