EFFECTIVE AEROSOL OPTICAL THICKNESS RETRIEVAL ALGORITHM USING MODIS 500 METRE DATA AHMAD MUBIN BIN WAHAB UNIVERSITI TEKNOLOGI MALAYSIA EFFECTIVE AEROSOL OPTICAL THICKNESS RETRIEVAL ALGORITHM USING MODIS 500 METRE DATA AHMAD MUBIN BIN WAHAB A thesis submitted in fulfilment of the requirements for the award of the degree of Master of Science (Remote Sensing) Faculty of Geoinformation and Real Estate Universiti Teknologi Malaysia MARCH 2015 iii DEDICATION Specially dedicated to my beloved parents, wife, siblings and all my fellows friends iv ACKNOWLEDGEMENT First and for most I would like to thank to Allah S.W.T for his presence, protection and guidance me during the whole study period. I wish to express my sincerity to my supervisor Dr. Md. Latifur Rahman Sarker for his encouragement, guidance, supports, advices, and motivation during my study. Without his continued support, this thesis would not have been the same as presented here. I would like to thanks Department of Geoinformation, Faculty of Geoinformation and Real Estate, and Universiti Teknologi Malaysia (UTM) for assisting during my study. Special thank also goes to School of Graduate Studies (SPS) UTM and Ministry of Higher Education (MOHE) for funding my study. I am also indebted to Janet Nichole for providing Aerosol Robotic Network (AERONET) data for Hong Kong Polytechnic University station and Hok Tsui station. I am grateful to all my family members especially my mother and my wife who always support me to fulfil this study. Your presence has a great value for me. Last but not least, my sincere gratitude to all my friends and others who have provided in the completion of this thesis. v ABSTRACT Aerosol estimation from satellite data is crucial for the air quality assessment, visibility estimation, and climate modelling. Numerous retrieval algorithms have been developed for aerosol optical thickness estimation but there are still uncertainties in estimation due to several factors that need to be addressed for the development of an effective retrieval algorithm. Therefore, the main goal of this study is to develop an effective aerosol retrieval algorithm using Moderate Resolution Imaging Spectroradiometer 500 metre data considering the effects of different Radiative Transfer codes, surface reflectance estimation techniques, local aerosol models, and atmospheric transmission contributions. The aerosol estimation algorithm has been developed using several processing steps include i) estimation of aerosol reflectance from satellite data, ii) local aerosol models characterization using aerosol inversion product, iii) estimation of aerosol reflectance as function of aerosol optical thickness using different Radiative Transfer codes and direct method, iv) retrieval of aerosol optical thickness by comparing the residual of aerosol reflectance between satellite data and Radiative Transfer codes using Look-up Tables based on optimal spectral shape fitting function, and v) validation of retrieved aerosol optical thickness with in-situ ground measurement. Results indicate that aerosol optical thickness can be successfully retrieved from the satellite data using Second Simulation of a Satellite Signal in the Solar Spectrum vector code, 2-channel of Moderate Resolution Imaging Spectroradiometer data, and surface reflectance derived from the Radiative Transfer code based atmospheric correction using continental and desert aerosol model together. The proposed algorithm is very effective and retrieved aerosol optical thickness from Moderate Resolution Imaging Spectroradiometer 500 metre data with the accuracy of 96% and low uncertainty for the both study sites. This finding highlights the potential of this algorithm to retrieve aerosol optical thickness from satellite data with high accuracy and good spatial information compared to the 10 kilometres satellite aerosol product. vi ABSTRAK Anggaran aerosol daripada data satelit adalah penting untuk penilaian kualiti udara, anggaran kebolehlihatan, dan pemodelan iklim. Pelbagai algoritma penerbitan telah dibangunkan untuk anggaran ketebalan optik aerosol tetapi masih terdapat keraguan dalam anggaran yang disebabkan oleh beberapa faktor yang perlu diberi perhatian untuk pembangunan algoritma penerbitan aerosol yang efektif. Oleh itu, tujuan utama bagi kajian ini adalah untuk membangunkan algoritma penerbitan aerosol menggunakan data Pengimejan Spektroradiometer Resolusi Sederhana 500 meter dengan mempertimbangkan kesan-kesan bagi kod Pemindahan Sinaran yang berbeza, teknik anggaran kepantulan permukaan, model aerosol tempatan, dan penyumbangan penghantaran atmosfera. Algoritma anggaran aerosol telah dibangunkan menggunakan beberapa langkah pemprosesan termasuk i) anggaran bagi kepantulan aerosol daripada data satelit, ii) pencirian model aerosol tempatan menggunakan produk penterbalikkan aerosol, iii) anggaran bagi kepantulan aerosol sebagai fungsi ketebalan optik aerosol menggunakan kod Pemindahan Sinaran yang berbeza dan kaedah secara langsung, iv) penerbitan ketebalan optik aerosol dengan membandingkan nilai perbezaan kepantulan aerosol di antara data satelit dan kod Pemindahan Sinaran yang menggunakan Jadual Carian berdasarkan fungsi penyesuaian bentuk spektral yang optimal, dan v) pengesahan ketebalan optik aerosol yang diterbitkan dengan pengukuran di tanah lapangan. Hasil kajian menunjukkan bahawa ketebalan optik aerosol dapat diterbitkan dengan jayanya daripada data satelit menggunakan kod vektor Simulasi Kedua bagi Isyarat Satelit di dalam Spektrum Suria, 2-saluran data Pengimejan Spektroradiometer Resolusi Sederhana, dan kepantulan permukan diterbitkan daripada kod Pemindahan Sinaran berdasarkan pembetulan atmosfera menggunakan model aerosol benua dan padang pasir bersama-sama. Algoritma yang dicadangkan adalah sangat efektif dan ketebalan optik aerosol yang diterbitkan daripada data Pengimejan Spektroradiometer Resolusi Sederhana 500 meter dengan ketepatan sebanyak 96% dan ketidakpastian yang rendah bagi kedua-dua kawasan kajian. Penemuan ini menekankan keupayaan algoritma ini untuk menerbitkan ketebalan optik aerosol daripada data satelit dengan ketepatan yang tinggi dan maklumat ruang yang bagus berbanding dengan produk satelit aerosol 10 kilometer. vii TABLE OF CONTENTS CHAPTER TITLE PAGE DECLARATION ii DEDICATION iii ACKNOWLEDGEMENT iv ABSTRACT v ABSTRAK vi TABLE OF CONTENTS vii LIST OF TABLES xi LIST OF FIGURES xiii LIST OF ABBREVIATIONS xvii LIST OF SYMBOLS xxi LIST OF APPENDICES xxiii 1 INTRODUCTION 1 1.1 Background of the Study 1 1.2 Problem Statement 6 1.3 Objectives of the Study 10 1.4 Scope of the Study 10 1.5 Study Area 12 1.6 Significance of the Study 15 1.7 Outline of the Thesis 16 2 LITERATURE REVIEW 18 2.1 Atmospheric Aerosol 18 2.1.1 Aerosol Parameters and Properties 22 viii 2.1.2 Aerosol Effects 25 2.1.3 Aerosol Measurements 27 2.1.4 Satellite Aerosol Measurements 29 2.1.5 Satellite Aerosol Remote Sensing Techniques 32 2.2 Radiative Transfer Theory 36 2.2.1 Scalar Radiative Transfer Model 37 2.2.2 Vector Radiative Transfer Model 39 2.2.3 Rayleigh Scattering 40 2.2.4 Mie Scattering 46 2.2.5 Gas Absorption 51 2.2.6 Successive Orders of Scattering (SOS) Method 55 2.2.7 Discrete Ordinates Radiative Transfer 56 (DISORT) method 2.3 Radiative Transfer Codes 58 2.3.1 MODerate Resolution Atmospheric 59 TRANsmission (MODTRAN) code 2.3.2 Santa Barbara DISORT Atmospheric 60 Radiative Transfer Program (SBDART) code 2.3.2 Second Simulation of a Satellite Signal in the 61 Solar Spectrum (6S) code 2.4 Concluding Remarks 63 3 RESEARCH METHODOLOGY 66 3.1 Data Preparation 68 3.1.1 Data Selection 68 3.1.2 Data Pre-processing 71 3.2 Aerosol Reflectance Estimation 75 3.2.1 Top of Atmosphere (TOA) Reflectance 77 3.2.2 Rayleigh Reflectance 79 3.2.3 Total Atmospheric Transmission 81 3.2.4 Land Surface Reflectance 83 3.2.5 Hemispheric Reflectance 86 3.2.6 Total Gaseous Transmission 87 ix 3.3 Local Aerosol Model Characterization 89 3.3.1 Identify Number of Clusters (k) 92 3.3.2 Clustering Analysis 93 3.4 Aerosol Optical Thickness (AOT) Retrieval 94 3.4.1 Aerosol Optical Thickness Retrieval based on 95 Look-up Tables using Santa Barbara DISORT Radiative Transfer (SBDART) code 3.4.2 Aerosol Optical Thickness Retrieval based on 99 Look-up Tables using Second Simulation of a Satellite Signal in the Solar Spectrum Vector (6SV) code 3.4.3 Aerosol Optical Thickness Retrieval without 102 Look-up Tables 3.5 Validation and Comparison of Aerosol Optical 105 Thickness at 0.55 µm 4 RESULT AND ANALYSIS 107 4.1 Local Aerosol Model Characterization 107 4.1.1 Selection of the Number of Clusters (k) 108 4.1.2 Development of Local Aerosol Models 112 4.1.3 Sensitivity Test of the Local Aerosol Models 114 4.2 Surface Reflectance Estimation 117 4.3 Validation of MODIS 500 m Aerosol Optical 120 Thickness using Aerosol Optical Thickness from Aerosol Robotic Network (AERONET) Station 4.3.1 Validation of Aerosol Optical Thickness 121 Retrieved from MODIS 500 m using Look-up Tables of Santa Barbara DISORT Radiative Transfer (SBDART) code 4.3.2 Validation of Aerosol Optical Thickness 124 Retrieved from MODIS 500 m using Look-up Tables of Second Simulation of a Satellite Signal in the Solar Spectrum Vector (6SV) x code 4.3.3 Validation of Aerosol Optical Thickness 128 Retrieved From MODIS 500 m Using Direct Analytical Equation (DAE) 4.4 Comparison of Accuracy of Different Aerosol Optical 134 Thickness Retrieval Models 4.5 Comparison between Aerosol Optical Thickness from 137 MODIS 500 m and Aerosol Optical Thickness from MODIS Aerosol Product 4.6 Spatial Distribution of Aerosol Optical Thickness 140 Generated from the Best Aerosol Optical Thickness Retrieval Model 4.6.1 Comparison of Spatial Distribution of MODIS 140 Aerosol Optical Thickness at 500 m and MODIS Aerosol Product over Hong Kong and Pearl River Delta Region 4.6.2 Spatial Distribution of AOT over Hong Kong 142 Region 4.7 Concluding Remarks 145 5 DISCUSSION 146 6 CONCLUSION AND RECOMMENDATIONS 151 6.1 Conclusion 151 6.2 Recommendations 153 REFERENCES 155 171-174 Appendices A - B