| DC 欄位 |
值 |
語言 |
| DC.contributor | 環境科技博士學位學程 | zh_TW |
| DC.creator | 雅瓜納 | zh_TW |
| DC.creator | Tri Wandi Januar | en_US |
| dc.date.accessioned | 2023-8-17T07:39:07Z | |
| dc.date.available | 2023-8-17T07:39:07Z | |
| dc.date.issued | 2023 | |
| dc.identifier.uri | http://ir.lib.ncu.edu.tw:88/thesis/view_etd.asp?URN=108083604 | |
| dc.contributor.department | 環境科技博士學位學程 | zh_TW |
| DC.description | 國立中央大學 | zh_TW |
| DC.description | National Central University | en_US |
| dc.description.abstract | 摘要
本研究旨在通過利用蘭薩特和日本向日葵衛星影像的融合，提升臺灣地區高時空解析度的地表溫度（LST）估算。采用機器學習回歸方法以提高LST估算的準確性。本研究利用蘭薩特和日本向日葵的熱紅外（TIR）影像結合，估算每小時的時序LST數據。對發射率、高程和植被指數（如NDVI）進行特徵工程，創建結合特徵進行回歸分析。利用從特徵工程獲得的分類圖進行分層隨機抽樣，得到7000個樣本的數據集，將其分為80％的訓練數據和20％的測試數據。採用支持向量回歸（SVR）算法對融合的LST進行校正，顯著提高了準確性，特別是在山區地區。結果表明，蘭薩特和日本向日葵影像的融合，結合機器學習回歸，能夠有效提升臺灣地區高時空解析度的LST估算。本研究有助於了解地表溫度變化並為氣候模擬、環境監測和城市規劃等應用提供了機。 | zh_TW |
| dc.description.abstract | Abstract
This research aims to enhance the estimation of high-spatiotemporal Land Surface Temperature (LST) in Taiwan by leveraging the fusion of Landsat and Himawari satellite imagery. A machine learning regression approach namely Support Vector Regression (SVR) is employed to improve the accuracy of the LST estimation. The study utilizes the combination of Landsat and Himawari thermal infrared (TIR) images to estimate hourly timeseries LST data. Feature engineering is performed on emissivity, elevation, and vegetation indices (such as NDVI) to create combined features for regression analysis. Stratified random sampling is employed using a classification map obtained from feature engineering, resulting in a dataset of 7,000 samples divided into 80% training and 20% testing data. The SVR algorithm is applied to correct the fused LST, resulting in a significant improvement in accuracy, especially in mountainous areas. The results demonstrate that the fusion of Landsat and Himawari imagery, combined with machine learning regression, can effectively enhance the estimation of high-spatiotemporal LST in Taiwan. This research contributes to the understanding of surface temperature variations and offers opportunities for applications in climate modeling, environmental monitoring, and urban planning.
. | en_US |
| DC.subject | 時空圖像融合 | zh_TW |
| DC.subject | 機器學習 | zh_TW |
| DC.subject | 支持向量回歸 | zh_TW |
| DC.subject | 地表溫度 | zh_TW |
| DC.subject | Landsat-8 | zh_TW |
| DC.subject | Himawari-8 | zh_TW |
| DC.subject | Spatiotemporal image fusion | en_US |
| DC.subject | machine learning | en_US |
| DC.subject | support vector regression | en_US |
| DC.subject | Land surface temperature | en_US |
| DC.subject | Landsat-8 | en_US |
| DC.subject | Himawari-8 | en_US |
| DC.title | 運用機器學習回歸方法改善臺灣地區從Landsat-8與Himawari-8融合影像中的高時空地表溫度 | zh_TW |
| dc.language.iso | zh-TW | zh-TW |
| DC.title | A Machine Learning Regression Approach for Improving High-Spatiotemporal LST from Landsat and Himawari Fused Imagery in Taiwan | en_US |
| DC.type | 博碩士論文 | zh_TW |
| DC.type | thesis | en_US |
| DC.publisher | National Central University | en_US |