水稻是台灣最重要的作物，在農業種植面積和產量上都佔有相當大的比例，同時也是水資源用量最多的作物。然而隨著氣候變遷加劇，乾旱、水災等天然災害發生的頻率越來越高，這對台灣的經濟和糧食安全造成巨大的威脅。遙測的發展提供了在大範圍的情況下監測水稻的生長狀況，如何及時且準確地監測稻田土壤水分狀況能夠更好地配置水資源確保食物供應的穩定性，使有效的農業水資源應用成為桃園灌區的重要議題。 估計植被覆蓋下的土壤濕度狀況是目前遙測在農業應用上的重要難題，本研究使用水雲模型 (Water-Cloud Model) 來降低植被覆蓋對於土壤溼度估計的影響，該模型透過結合合成孔徑雷達（Synthetic Aperture Radar, SAR）和光學資訊來模擬不同植被覆蓋下的雷達散射狀況。水雲模型的應用需要根據分析SAR回波、光學植被指數和現場量測的地表土壤濕度之間的關係進行校準，因此本研究蒐集了台灣桃園市桃園灌區2018年至2022年的Sentinel-1雷達影像和Sentinel-2光學影像和中央大學大氣水文觀測站所量測的土壤濕度，並考量了模型變數的影響分別測試了不同類型的雷達極化和針對不同環境變化進行校正的植生指數對於水雲模型校準的性能差異，最後利用精度最高的模型監測桃園灌區水稻田的土壤濕度狀況。 研究結果顯示，在不受到降雨所影響的條件下使用同極化VV搭配耐大氣植生指數 (Atmospherically Resistant Vegetation Index, ARVI) 時的效果最好 (R2=0.55, RMSE=4.16) ，然而，由於缺乏不同條件的土壤濕度量測資料和農業活動的不確定性，本研究所使用的方法需要進一步的分析和實驗以改進模型的性能。 ;Paddy rice is the crop with the largest planting area and the most significant irrigation water demand in Taiwan, and timely and accurate monitoring of soil moisture in paddy rice fields can make a better allocation of water resources and secure the stability of the food supply. Previous studies have focused on rice mapping using remote sensing techniques. However, irrigation and soil moisture which also significantly influence the rice yield has not yet been fully considered in the rice production estimation. In this study, to monitor soil moisture in paddy rice fields, this study uses the modified water-cloud model (WCM) which is able to estimate surface moisture in different vegetation covers by integrating Synthetic Aperture Radar (SAR) and optical information. The model application requires calibrating work that is based on analyzing the relationships between SAR backscattering, optical vegetation index, and measured surface moisture content. Specifically, this study collected Sentinel-1 and Sentinel-2 satellite images to perform the soil estimation, and ground measurements from 2018 to 2022 were also obtained for calibrating and validating estimation results. The soil moisture content is measured and recorded using Frequency-domain sensors (FDR) by National Central University (NCU) Atmosphere and Hydrology Observation station. To calibrate the model, this study used the SAR backscattering data from Sentinel-1 and tested various vegetation indexes calculated by using Sentinel-2 imagery. Among them, the highest R2 value of 0.55 can be obtained when the Atmospherically Resistant Vegetation Index (ARVI) is applied. However, the model can be further improved by calibrating the model with more soil moisture observations from various conditions of vegetation covers.
