Land surface albedo retrieval via Himawari-8/AHI and quality assessment
- Abstract
- Land surface albedo (LSA) acts an important role in climate change because it determines absorbed solar radiation on Earth’s surface, thus, it is chosen as one of the Essential Climate Variable (ECV). For the past decades, studies have been conducted to retrieve more accurate LSA from satellite images by addressing some issues in LSA retrieval algorithms such as removing atmospheric effect and describing surface anisotropy characteristics. In this dissertation, the LSA retrieval algorithm for Himawari-8/AHI was proposed for the first retrieval of LSA from multi-spectral geostationary satellite data over the Asia-Oceania region. Furthermore, the results of validation and uncertainty assessment were presented to provide objective quality and strengthen the application of our results. The proposed algorithm is composed of three major steps: 1) atmospheric correction, 2) Bidirectional reflectance distribution function (BRDF) modeling, and 3) Narrow-to-broadband (N2B) conversion. The atmospheric correction was performed using the pre-constructed Second Simulation of a Satellite Signal in the Solar Spectrum (6S) look-up table (LUT). To improve the accuracy of TOC reflectance, the minimum curvature surface method was applied to refine the interval for solar zenith angle and viewing zenith angle of 6S LUT. Then, the BRDF modeling was performed using kernel-based model. In this step, the optimization method using normalized reflectance was presented. This method improved the quality of BRDF modeling results, particularly when the number of observations was less than 15. Finally, through multiple regression analysis with good quality Moderate resolution imaging spectroradiometer (MODIS) LSA data, the N2B conversion coefficients optimized for AHI channels were derived and the broadband LSAs for black-sky (BS) and white-sky (WS) were estimated. The estimated broadband LSA for both BS and WS showed similar spatial distribution with those of MODIS. A quality assessment was performed on three aspects: traceability, validation, and uncertainty). First, the traceability chain was presented to visualize the traceability of the proposed algorithm. In addition, it showed the direction of uncertainty propagation and includes a description of the data and methods used in each process. Next, as the results of validation, the estimated broadband LSAs for both BS and WS showed a low RMSE (snow-free: 0.0176 for BS and 0.0236 for WS, snow-covered: 0.0523 for BS and 0.0529 for WS) compared to the MODIS LSA product. The validation results compared to the ground measurements from Aerosol Robotic Network (AERONET), Australian and New Zealand flux tower network (OzFlux) showed the RMSE of 0.023. It was comparable to that of MODIS LSA and satisfied to requirements of climate model (0.02-0.05). In addition, our results reasonably captured the seasonal variation for Cheorwon of Korea Flux Network (KoFlux) and can reveal more LSA data even when clouds are frequently present, such as during the summer monsoon season. Lastly, in uncertainty assessment, the uncertainties in each processing step were presented. The results of this dissertation can provide high-quality LSA with high spatial and temporal completeness from geostationary satellite observations in the Asia–Oceania region. We believe that estimated broadband LSA and results of validation and uncertainty analysis can be utilized in various field, such as validation of model results, simulation of weather and climate, and radiation budget studies.
Key words: Himawai-8/AHI; Atmospheric correction; Bidirectional reflectance distribution function modeling; land surface albedo; quality assessment
- Issued Date
- 2021
- Awarded Date
- 2021. 2
- Type
- Dissertation
- Publisher
- 부경대학교
- Affiliation
- 부경대학교 대학원
- Department
- 대학원 지구환경시스템과학부공간정보시스템공학전공
- Advisor
- 한경수
- Table Of Contents
- CHAPTER 1: GENERAL INTRODUCTION 1
1.1. Background 2
1.2. Problems 5
1.3. Objectives of this dissertation 10
1.4. Structure of this dissertation 13
CHAPTER 2: USED DATA AND STUDY AREA 15
2.1. Used data 16
2.1.1. Himawari-8/AHI Level 1B product 17
2.1.2. JAXA cloud and aerosol products 20
2.1.3. ECMWF atmospheric data 22
2.1.4. MODIS land surface albedo product 23
2.1.5. Auxiliary data 25
2.1.6. Ground measurement 26
2.2. Study area 30
CHAPTER 3: IMPROVEMENTS OF 6S LUT BASED TOC REFLECTANCE 33
3.1. Introduction 35
3.2. Methods 37
3.2.1. 6S RTM 38
3.2.2. Pre-construction of the 6S LUT 40
3.2.3. Accuracy assessment 41
3.3. Results 42
3.3.1. Evaluation of interpolated 6S LUT 42
3.3.2. Evaluation of accuracy for surface reflectance 47
3.4. Examples of outputs produced in atmospheric correction 57
3.5. Summary and conclusion 59
CHAPTER 4: OPTIMIZATION OF BRDF MODELING USING NORMALIZED RELFECTANCE 61
4.1. Introduction 63
4.2. Methodology 67
4.2.1. Roujean model and composite strategy 67
4.2.2. BRDF normalization 71
4.2.3. Optimization of BRDF 72
4.3. Results and discussion 74
4.3.1. Analysis to select the number of optimization iterations 74
4.3.2. Analysis of RMSE in BRDF modeling 77
4.3.3. Comparison of spatial distribution 81
4.3.4. Evaluation of surface anisotropy simulation through time-series analysis 83
4.4. Summary and Conclusion 89
CHAPTER 5: LAND SURFACE ALBEDO RETRIEVAL THROUGH KERNEL INTEGRATION AND NARROW-TO-BROADBAND CONVERSION 91
5.1. Introduction 93
5.2. Methodologies 94
5.2.1. Kernel integration and narrowband LSA retrieval 94
5.2.2. N2B conversion and retrieval of broadband LSA 97
5.3. Results and analysis 100
5.3.1. Results of narrowband LSA retrieval 100
5.3.2. Evaluation of N2B conversion equation 102
5.3.3. Results of broadband LSA retrieval 106
5.4. Summary 108
CHAPTER 6: VALIDATION AND UNCERTAINTY ASSESSMENT 111
6.1. Introduction 113
6.2. Algorithm traceability 116
6.2.1. Description of traceability and methods 116
6.2.2. Traceability chain of LSA retrieval algorithm 117
6.3. Validation 134
6.3.1. Description of validation and methods 134
6.3.2. Results of validation and inter-comparison 137
6.4. Uncertainty analysis 150
6.4.1. Description of uncertainty analysis and methods 150
6.4.2. Uncertainty analysis for each processing steps 150
6.5. Summary and conclusion 162
CHAPTER 7: GENERAL CONCLUSION 165
7.1. Conclusion 166
7.1.1. Atmospheric correction employing MCS technique 166
7.1.2. BRDF modeling with optimization process 168
7.1.3. LSA retrieval through angular integration and N2B conversion 169
7.1.4. Validation and uncertainty analysis 170
7.2. Recommendations for the further research 171
References 173
- Degree
- Doctor
- Appears in Collections:
- 대학원 > 지구환경시스템과학부-공간정보시스템공학전공
- Files in This Item:
-
-
There are no files associated with this item.
-
Items in Repository are protected by copyright, with all rights reserved, unless otherwise indicated.