ASSESSMENT OF ATMOSPHERIC CORRECTION METHODS FOR OPTIMIZING HAZY SATELLITE IMAGERIES
UMARA FIRMAN RIZI DIANSYAH
GRADUATE SCHOOL BOGOR AGRICULTURAL UNIVERSITY BOGOR 2014
STATEMENT I, Umara Firman Rizi Diansyah hereby declare that this thesis entitled : Assessment Of Atmospheric Correction Methods For Optimizing Hazy Satellite Imageries Is a result of my own work under the supervision of advisory board during the period of December 2013 until December 2014 and that it has not been published ever. The content of this thesis has been examined by the advisory board and external examiner
Bogor,
December 2014
Umara Firman Rizi Diansyah G051120041
RINGKASAN UMARA FIRMAN RIZI DIANSYAH. “Penilaian Metode Atmospheric Correction Untuk Optimasi Citra Satelit Berkabut. Dibawah bimbingan oleh MUHAMMAD BUCE SALEH and ANTONIUS BAMBANG WIJANARTO. Efek atmosfer yang mencemari citra satelit dapat dihilangkan dengan mendapatkan nilai reflektansi permukaan pada citra penginderaan jauh dengan menggunakan koreksi atmosfer. Proses ini berguna untuk meningkatkan akurasi klasifikasi citra sehingga data yang diperoleh dapat dibandingkan dan diatur dalam sejumlah solusi pendekatan untuk pemantauan lingkungan, manajemen sumber daya, evaluasi dampak aplikasi deteksi bencana dan perubahan. Koreksi atmosfer memiliki berbagai kategori yang berasal dari pantulan fisik tanah yang meninggalkan cahaya termasuk permukaan dan tanah reflektansi dan suhu, dan koreksi relatif atmosfer. Salah satu kategori adalah penghapusan kabut atmosfer dan tutupan awan. Hasil proses koreksi atmosfer sangat penting untuk mengoptimalkan citra satelit berkabut berkenaan dengan permukaan objek untuk mendeteksi perubahan tutupan lahan dan penggunaan lahan. Dengan mempertimbangkan perbedaan dalam proses pembentukan kabut penelitian ini difokuskan pada masing-masing jenis citra satelit dari dua jenis zona yang dipilih, pertama kecamatan Balaraja yang dapat mewakili daerah pedesaan dan kecamatan Penjaringan yang mewakili daerah perkotaan. Maksud dari penelitian ini adalah untuk menguji kesesuaian tiga jenis metode koreksi haze terhadap kejelasan obyek permukaan di wilayah tutupan vegetasi dan non vegetasi, berkenaan dengan mengurangi atau menghilangkan kabut di suatu wilayah dalam sebuah citra satelit optik yang memiliki karakteristik tertentu, untuk memperoleh hasil dari penghapusan kabut dalam penelitian ini menggunakan tiga metode yaitu Dark Object Substraction (DOS), Histogram Match (HM) dan Virtual Cloud Point (VCP). Tujuan dari penelitian ini adalah untuk menguji kesesuaian tiga jenis metode koreksi kabut terhadap objek permukaan pada tutupan lahan di wilayah vegetasi dan wilayah non vegetasi. Berkenaan dengan penghapusan kabut di wilayah yang memiliki karakteristik tertentu pada citra satelit optik. Dengan mempertimbangkan proses pembentukan kabut di daerah pertanian, perkebunan, pemukiman dan industri sehingga daerah penelitian dibagi menjadi dua wilayah yaitu pedesaan yang didominasi vegetasi dan perkotaan yang didominasi non vegetasi. Wilayah pedesaan dipilih Kecamatan Balaraja di Tangerang dan wilayah perkotaan dipilih Kecamatan Penjaringan di Jakarta Utara. Dengan menerapkan persamaan untuk menghitung Haze Optimal Transformasi yaitu HOT = DNbluesin (∂) -DNredcos (∂), hasil utama dari penelitian ini adalah sebagai berikut: untuk Avnir gambar di daerah pedesaan, berdasarkan kriteria visual, korelasi dan uji Z dapat disimpulkan bahwa untuk Avnir2 citra di daerah pedesaan dapat dikoreksi dengan menggunakan metode DOS dan VCP. Untuk citra Avnir2 di daerah perkotaan, metode DOS memiliki hasil yang baik pada band 1, 2 dan 3, Dapat disimpulkan bahwa untuk citra Avnir2 di daerah
perkotaan dapat dikoreksi dengan menggunakan metode DOS. Untuk citra Landsat7 di daerah pedesaan, metode DOS memiliki hasil yang baik pada band 1 dan 2, Dapat disimpulkan bahwa untuk Landsat7 di daerah pedesaan dapat dikoreksi dengan menggunakan metode DOS dan VCP. Untuk citra Landsat7 di daerah perkotaan, metode DOS memiliki hasil yang baik pada band 1 dan 2, sedangkan metode VCP memiliki hasil yang baik pada band 3. Dapat disimpulkan bahwa untuk citra Landsat7 di daerah perkotaan dapat dikoreksi dengan menggunakan metode DOS. Semakin baik hasil citra terkoreksi, semakin kecil nilai optimasi kabut. Proses koreksi atmosfer pada penelitian ini dibagi dalam dua jenis citra Avnir2 dan Landsat7. Alasan digunakannya citra optik tersebut berdasarkan pada hipotesis bahwa penggunaan tiga metode memberikan hasil yang sama bila diterapkan pada citra yang memiliki rentang band yang berbeda. Berdasarkan interpretasi visual dari perbandingan tiga metode DOS dan VCP didapatkan hasil lebih baik di daerah pedesaan dan perkotaan ditunjukkan dengan melihat kejelasan kerapatan vegetasi di daerah pedesaan, kontur, warna dan objek-objek lain di daerah perkotaan dan hubungan signifikan antara citra terkoreksi dengan citra non-kabut pada Avnir2, begitu pula dengan citra terkoreksi dengan citra non-kabut pada Landsat7. Berdasarkan interpretasi visual, korelasi dan uji Z disemua kategori dihasilkan perbedaan yang signifikan. Berdasarkan Tabel 12 untuk citra Avnir2 di wilayah rural dapat dikoreksi menggunakan metode DOS danVCP, tetapi di wilayah urban lebih baik diterapkan metode DOS. Untuk citra Landsat7 di wilayah rural dapat dikoreksi menggunakan metode DOS dan VCP, tetapi di wilayah urban lebih baik diterapkan metode DOS. Kata kunci : Avnir dan Landsat7, daerah non vegetasi urban, daerah vegetasi rural, efek atmosfer, eliminasi haze.
SUMMARY UMARA FIRMAN RIZI DIANSYAH. “Assessment Of Atmospheric Correction Methods For Optimizing Hazy Satellite Imageries. Under the supervision of MUHAMMAD BUCE SALEH and ANTONIUS BAMBANG WIJANARTO. Atmospheric effects that contaminate satellite imagery can be eliminated by obtaining the value of surface reflectance on remote sensing imagery using atmospheric correction. This process is useful to improve the accuracy of image classification so that the data obtained can be compared and arranged in a number of solution approaches for environmental monitoring, resource management, evaluation of the effects of disaster and change detection applications. Atmospheric correction has a variety of categories derived from physical reflectance of the true ground leaving radiance including surface and ground reflectance and temperature, and relative atmospheric correction. One of the categories is the elimination of atmospheric haze and cloud masking. The results of atmospheric correction process is very important to optimize the hazy satellite imagery with respect to the object surface to detect changes in land cover and land use. Taking into account the differences in the process of formation of haze this study focused on each type of satellite images of the two types of zones are selected, the first sub-district Balaraja which can represent rural areas and sub-district Penjaringan representing urban areas. The intent of this study was to test the suitability of three types of haze correction method to clear the surface of the object in the area of vegetation and nonvegetation cover, with regard to reducing or eliminating the haze in the region in an optical satellite imagery that has certain characteristics, to gain from the elimination of haze in this study using three methods: Dark Object Substraction (DOS), Histogram Match (HM) and Virtual Cloud Point (VCP). The purpose of this study was to test the suitability of three types of haze correction method of the object on the surface of the land cover in the area of vegetation and non-vegetation area. With regard to the elimination of haze in the area that have certain characteristics in optical satellite imagery. Taking into account the formation of haze in the area of agriculture, plantations, settlements and industry so that research area is divided into two regions, namely rural and urban vegetation dominated predominantly non vegetation. Rural areas selected in the Tangerang District of Balaraja and urban areas have been Penjaringan in North Jakarta. By applying the equation to calculate the Haze Optimize Transformation is HOT = DNbluesin (∂) -DNredcos (∂), the main results of this study are as follows: for AVNIR image in rural areas, based on visual criteria, correlation and Z test it can be concluded that for Avnir2 image in rural areas can be corrected by using DOS and VCP. For Avnir2 image in urban areas, the DOS method has good results in bands 1, 2 and 3, can be concluded that for the image Avnir2 in urban areas can be corrected by using DOS. For Landsat7 image in rural areas, DOS method has good results in bands 1 and 2, can be concluded that for Landsat7 in rural areas can be corrected by
using DOS and VCP. For Landsat7 image in urban areas, the DOS method has good results in bands 1 and 2, while the VCP method has good results in band 3. It can be concluded that for the image Landsat7 in urban areas can be corrected by using DOS. The better results corrected image, the smaller the value of optimization haze Atmospheric correction process is divided into two kinds of images Landsat7 and Avnir2 . Reason uses the optical image is based on the hypothesis that the use of three methods gave similar results when applied to images that have a range of different bands. Based on visual interpretation of the comparison of three methods of DOS and VCP obtained better results in rural and urban areas indicated by viewing clarity vegetation density in rural areas, contours, colors and other objects in urban areas and significant relationship between the corrected image with the image of a non haze on Avnir2, as well as the corrected image with the image of a non-haze on Landsat7. Based on visual interpretation, correlation and Z test in all categories produced a significant difference. According to Table 12 for the image Avnir2 in rural areas can be corrected using the DOS danVCP, but in the urban area is better applied methods DOS. For Landsat7 image in rural areas can be corrected using DOS and VCP, but in the urban area is better applied methods DOS. Keywords : Avnir and Landsat7, urban non vegetation area, rural vegetation area, atmospheric effects, haze elimination
Copyright @ 2014, Bogor Agricultural University Copyright are protected by law It is prohibited to cite all or part of this thesis without referring to and mentioning the source; Citation only permitted for the sake of education, research, scientific writing, report writing, critical writing, or reviewing scientific problems. Citation does not inflict the name and honor of Bogor Agricultural University. It is prohibited to republish and reproduce all part of this thesis without the written permittion from Bogor Agricultural University.
ASSESSMENT OF ATMOSPHERIC CORRECTION METHODS FOR OPTIMIZING HAZY SATELLITE IMAGERIES
UMARA FIRMAN RIZI DIANSYAH
A Thesis submitted for the degree of Master of Science in Information Technology for Natural Resources Management Program Study
GRADUATE SCHOOL BOGOR AGRICULTURAL UNIVERSITY BOGOR 2014
External Examiner:
Prof Dr Lilik Budi Prasetyo
Research Title Name Student ID
: Assessment Of Atmospheric Correction Methods For Optimizing Hazy Satellite Imageries : Umara Firman Rizi Diansyah : G051120041
Approved by, Advisory Board
Dr Antonius Bambang Wijanarto Co-Supervisor
Dr Muhammad Buce Saleh Supervisor
Endorsed by,
Program Coordinator
Dean of the Graduate School
Dr Ir Hartrisari Hardjomidjojo, DEA
Dr Ir Dahrul Syah, MScAgr
Date of examination:
Date of Graduation:
December 30, 2014
ACKNOWLEDGMENT In the name of Allah Azza Wajalla, the Most Gracious and the Most Merciful. Alhamdulillah, all praises to Allah for the strengths and His blessing in completing this thesis, I would like to express my highly appreciation to the following: 1. Dr Muhammad Buce Saleh, as my supervisor for his supervision support. His invaluable help in the forms of constructive comments and suggestions throughout the experimental and thesis works have contributed to the success of this research and Dr Antonius Bambang Wijanarto, as the co-supervisor for his ideas and knowledge regarding this topic during my research. 2. Prof Dr Lilik Budi Prasetyo as the external examiner for his positive inputs and ideas. 3. Dr Ir Hartrisari Hardjomidjojo, DEA, as the program coordinator and all my lecturers for giving knowledge and experience. 4. All my friends at MIT, IPB for their help and support in finishing my assignment and study. 5. MIT secretariate and all of its staff members for helping me to arrange the administration, technical procedures, and facilities. 6. All of my family. Thank you for always supporting me.
Bogor, December 2014
Umara Firman Rizi Diansyah
LIST OF CONTENTS LIST OF TABLES LIST OF FIGURES 1.
INTRODUCTION Background Problem Statement Hypothesis Objective Benefits of Research Limitation
2.
3
RESEARCH METHODOLOGY Time and Location Required Tool Data Sources Methodology Methods of Haze Removal Dark Object Substraction Histogram Match Virtual Cloud Point Quality Assessment
vii vii 1 1 4 4 4 5 5 5 5 6 6 7 7 8 8 9 9
Visual interpretation
9
Correlation Analysis
10
Z-test
10
RESULT AND DISCUSSION Avnir image in the rural area Visual Interpretation Z Analysis Avnir image in the urban area Visual Interpretation Z Analysis Landsat image in the rural area Visual Interpretation Correlation Analysis
11 11 11 12 13 13 14 14 14 15
Z Analysis Landsat image in the urban area Visual Interpretation Correlation Analysis Z Analysis Assessment using Haze Optimize Transformation (HOT) 4
15 17 17 17 18 19
CONCLUSIONS
20
REFERENCES APPENDICES
21
CURRICULUM VITAE
25
23
LIST OF TABLES 1
Data Requirement
Spectral Bands / Wavelength 3 Z values for Avnir in the rural area 4 Z values for Avnir in the urban area 5 The correlation coefficient between the image of the three correction method with non-haze image for Landsat7 in rural area 6 Z values for Landsat image in the rural area 7 Z values for Landsat of non-Haze (NH) in rural area 8 The correlation coefficient between the image of the three correction method with non-haze (NH) image for Landsat in urban areas 9 Z values for Landsat image in the urban area 10 Z values for Landsat image in the urban area vs non haze (NH) image 11 Results of Haze Optimize Transformation Calculation 12 Selection haze correction method based on visual criteria, correlation and Z test for all categories 2
6 6
12 14 15 16 16 17 18 18
19 19
LIST OF FIGURES
1 2 3 4 5 6 7 8 9 10 11 12 13 14
Formation of Haze Workflow of Atmospheric Correction Research Area (Balaraja and Penjaringan subdistrict) Procedure of Haze removal Structure of Dark Object Substraction (DOS) Structure of Virtual Cloud Point (VCP) Avnir image in Haze and Non-Haze rural area Corrected Image using DOS, HM and VCP in Avnir image and rural area Avnir image in Haze and Non-Haze urban area Corrected Image using DOS, HM and VCP in Avnir image and urban area Landsat image in Haze and Non-Haze rural area Corrected Image using DOS, HM and VCP in Landsat image and rural area Landsat image in Haze and Non-Haze urban area Corrected Image using DOS, HM and VCP in Landsat image and urban area
2 3 5 7 8 9 12 13 13 14 15 16 17 19
