從太空看家鄉：GeoAI 輔助高中生進行遙測影像分析與環境議題探究
講者/Speaker： 楊嵐雅, Lan-Ya Yang
本教學設計以「從太空看家鄉」為核心理念，結合 GeoAI與遙測教育，引導高中生運用 Sentinel-2 衛星影像分析臺灣水資源與環境變遷議題。課程起源於 2023 年卡努颱風後學生對災害與水庫淤積問題的關注，教師進一步導入 ChatGPT、Google Colab 與 Python 自動化流程，協助學生突破傳統 GIS 軟體高門檻與繁瑣前處理的限制。教學中透過 NDWI 水體指數、Otsu 動態閾值分類，以及改良自 Chen 與 Wang（2025）之 LogiTide2DEM 模型，讓學生能以多時期衛星影像與水位資料重建水庫地形變遷，並估算卡努颱風造成之淤積量。課程強調 AI 作為「科學運算引擎」與「人機協作夥伴」，使學生從工具使用者轉化為模型建構者，培養證據導向的空間分析能力與批判性數位判讀力。實施成果顯示，學生僅於一個月內即完成超過 60 幅衛星影像分析，效率較傳統流程大幅提升，並榮獲 2025 年國家太空中心衛星影像小論文競賽全國第一名。此教學模式展現 AI 融入地理教育後，促進教育平權、跨域探究與高階科學素養培育的實踐潛力。
This instructional design integrates GeoAI (Geospatial Artificial Intelligence) and remote sensing inquiry to support senior high school students in investigating environmental change and water-resource issues through Sentinel-2 satellite imagery. The course was inspired by students’ observations of environmental impacts following Typhoon Khanun in 2023, which led them to explore reservoir sedimentation and watershed change in Taiwan. To reduce the technical barriers commonly associated with traditional GIS workflows, the course incorporated ChatGPT, Google Colab, and Python-based automation as learning supports for spatial analysis and scientific inquiry.
Students applied the Normalized Difference Water Index (NDWI), Otsu dynamic threshold classification, and an adapted LogiTide2DEM model based on Chen and Wang (2025) to reconstruct topographic changes and estimate sedimentation volume using multi-temporal satellite imagery and reservoir water-level data. Through this process, AI functioned not merely as an information tool, but as a scientific computing and human–AI collaborative partner that enabled students to focus on evidence-based reasoning, spatial modeling, and environmental interpretation.
The instructional outcomes demonstrated substantial improvements in inquiry efficiency and scientific literacy. Students processed and analyzed more than 60 satellite images within one month, significantly reducing the time required for conventional GIS analysis. Moreover, students applying this GeoAI-based approach received first prize in the 2025 Taiwan Space Agency (TASA) Satellite Imagery Research Competition. This study highlights the potential of AI-supported geography education to democratize access to advanced geospatial technologies, foster interdisciplinary inquiry, and cultivate higher-order scientific and critical digital literacy in senior high school education.
- 論文全文：從太空看家鄉：GeoAI 輔助高中生進行遙測影像分析與環境議題探究
- 影像： 相關影片