Integration of Ground Inventory Data with Landsat Imagery to Estimate Aboveground Biomass of Tropical Deciduous Forest in Bago Yoma, Myanmar

학위논문 (석사)-- 서울대학교 대학원 : 산림과학부(산림환경학전공), 2015. 8. 김현석.Even with recently increased awareness of the environmental conservation, the degradation of tropical forests are still becoming the major source for carbon emission to the atmosphere. The aboveground biomass (AGB) of these forests are, therefore, a vital role in global carbon sequestration. As the initial step of the forest conservation in Myanmar, the aboveground biomass of South Zarmani Reserved Forest in Bago Yoma region were estimated using Landsat 8 OLI after the evaluation with 100 sample field inventory plots. Multiple linear regression (MLR) model of band values and their principal component analysis (PCA) model were developed to estimate the AGB using the spectral reflectance from Landsat images and elevation as the input variables. TheMLR model had r2 = 0.43, RMSE = 60.2 tons/ha, relative RMSE = 70.1%, Bias = -9.1 tons/ha, Bias (%) = -10.6%, and p < 0.0001, while the PCA model showed r2 = 0.45, RMSE = 55.1 tons/ha, relative RMSE = 64.1%, Bias = -8.3 tons/ha, Bias (%) = -9.7%, and p < 0.0001. The AGB maps of the study area were generated based on both MLR and PCAmodels. The estimated mean AGB values were 74.74±22.3 tons/ha and 73.04±17.6 tons/ha and the total AGB of the study area are about 5.7 and 5.6 million tons from MLR and PCA, respectively. In conclusion, we were able to generate solid regression models from Landsat 8 OLI image after ground truth and two regression models gave us very similar AGB estimation (less than 2%) of South Zarmani Reserved Forest, Bago Yoma, Myanmar.Table of Contents Abstract i Table of Contents iii List of Tables v List of Figures vi Abbreviations viii 1. Introduction 1 2. Literature Review 4 2.1 The Role of Tropical Forests in Climate Change 4 2.2 Remote Sensing Approaches to Estimation of AGB 6 3. Materials and Methods 11 3.1. Description of the Study Area 11 3.2. Remote Sensing Datasets 12 3.3. Field Biomass Measurement 16 3.4. AGB Estimated from Field data 27 3.5. Generating Regression Model for RS Biomass Estimation18 3.5.1. Multiple Linear Regression Model (MLR) 18 3.5.2. Principal Component Analysis (PCA) 19 3.6. Model Validation 20 4. Results and Discussion 21 4.1. Forest Characteristics 21 4.2. AGB Estimated from Field Data 25 4.3. Regression Models for AGB Estimation 26 4.4. AGB Estimation Maps from Two Models 30 5. Conclusion 34 Literature Cited 36 Appendix 1 48 Appendix 2 53 Abstract in Korean 57 Acknowledgement 59Maste