| 摘要: |
| 基于卫星遥感的透明度（SD）反演方法已经在大洋、近海、湖泊等不同水体开展了广泛应用。但在富营养化的河口港湾等水体，通常具有水域面积小且受陆地影响较大等特点，此外对其开展环境监测具有一定实时性的要求。受时间空间分辨率以及云层覆盖、大气校正等影响，基于卫星遥感方法在河口港湾等小水域进行透明度反演会受到一定的应用限制。因此，在此类特征水体建立一种高效便捷的SD反演方法作为遥感方法的有益补充就十分迫切。本研究尝试通过无人机和智能手机应用程序HydroColor APP搭载的普通光学相机构建基于红（R）绿（G）蓝（B）三波段的象山港SD反演方法。结果表明，无人机和HydroColor的红蓝波段比值（R/B）和红绿波段比值（R/G）与SD具有显著负相关，相关系数R 为–0.88 至 –0.93（n = 16, p < 0.001）。在根据相关性结果构建SD反演模型后，基于独立数据库对模型进行精度评估。结果显示（1）指数反演模型要优于线性经验模型，（2）基于R/G反演模型要优于R/B模型，（3）HydroColor反演模型要优于无人机反演模型。通过以上结果分别构建基于无人机DJI-R/G和HydroColor-R/G最优SD指数反演模型。DJI-R/G模型平均相对误差和均方根误差为29%和0.3 m，HydroColor-R/G模型为21.9%和0.27 m。以上结果表明，通过无人机和手机的RGB信息均可在象山港水域对SD可进行较准确的反演，该方法的建立为快速便捷开展河流港湾等水体的水质监测和赤潮防控提供了新的技术支持。 |
| 关键词: 透明度 RGB 遥感反演 无人机 手机APP 象山港 |
| DOI： |
| 分类号: |
| 基金项目: |
|
| ESTIMATING TRANSPARENCY IN TYPICAL BAY USING THE RED (R), GREEN (G), AND BLUE (B) SPECTRAL BANDS |
|
ZHU YUANLI, MAO MING, GUO RANG, DU PING, TAO BANGYI, JIANG ZHIBING, ZENG JIANGNING
|
|
Second Institute of Oceanography
|
| Abstract: |
| Satellite remote sensing-based transparency (SD) inversion methods are prevalent for vast water bodies like oceans, offshore areas, and lakes. However, estuaries and bays—characterized by significant land influence and smaller water expanses, in addition to the time requirement—pose challenges due to constraints like temporal or spatial resolution and cloud interference. This underscores the need for a robust SD inversion approach tailored for these regions. The study aimed to devise an SD inversion method for Xiangshan bay using both a UAV and the HydroColor APP on smartphones, focusing on the Red (R), Green (G), and Blue (B) bands. We observed that the UAV and HydroColor's red-to-blue (R/B) and red-to-green (R/G) band ratios had a strong negative correlation with SD, with correlation coefficients R between -0.88 and -0.93 (n = 16, P < 0.001). After developing the SD inversion model grounded on these correlations, we assessed its accuracy using an independent dataset. The findings revealed that: the exponential inversion model was more effective than the linear empirical one. Models based on R/G outperformed those on R/B. The HydroColor inversion model surpassed the UAV-based model. The most efficient SD inversion approach were found at two platforms to be the DJI--R/G and HydroColor-R/G exponential models, boasting an average relative error of 29% and 21.9% as well as a root-mean-square error of 0.3 m and 0.27 m. In conclusion, leveraging RGB data from UAVs and smartphones can yield accurate SD inversions. This novel approach offers a fast and efficient method for water quality assessment and red tide monitoring in rivers, bays, and similar water bodies. |
| Key words: transparency RGB remote sensing inversion UAV smartphones APP Xiangshan bay |