使用Sentinel-2图像在小规模水库中估计有色溶解有机物质:基于分类和回归的数据重新采样方法
Jinuk Kim1, Jin Hwi Kim1, Wonjin Jang2
1Future and Fusion Lab of Architectural, Civil and Environmental Engineering, Korea University, Seoul, 02841, Republic of Korea.
Environmental research
|December 14, 2025
概括
机器学习模型与合成数据相结合,有效估计有色溶解有机物质 (CDOM) 度. 使用SMOTE-SVM和SmoteR的深度神经网络显著提高了准确性,提高了水质监测.
科学领域:
- 环境科学 环境科学
- 遥感 遥感 遥感 遥感
- 机器学习 机器学习
背景情况:
- 不平衡的数据集对在水生环境中准确估计有色溶解有机物 (CDOM) 构成挑战.
- 准确的CDOM估计对于水质监测和环境管理至关重要.
研究的目的:
- 将机器学习模型与合成数据集成,以克服CDOM度估计中的数据不平衡问题.
- 评估各种机器学习模型和用于CDOM估计的合成数据生成技术的性能.
主要方法:
- 使用 Sentinel-2 图像,通过 HSIC-Lasso 模型推导出五个输入变量 (带比).
- 使用基于分类 (SMOTE-SVM,SMOTE-Tomek) 和基于回归 (SmoteR) 的技术生成合成数据,以解决数据不平衡.
- 训练并评估组合模型,包括随机森林 (RFR),极端梯度增强 (XGB),支持矢量机 (SVR) 和深度神经网络 (DNN).
主要成果:
- 深度神经网络 (DNN) 与SMOTE-SVM和SmoteR相结合,实现了最高的预测准确度 (分别为0.88和0.89).
- 这些先进的模型显著减少了错误,特别是在高CDOM度地区.
- 基于分类的过量采样技术在解决数据不平衡方面被证明是有效的,特别是在高度边界.
结论:
- 合成数据生成和机器学习为稳健的水质监测提供了强大的方法.
- 拟议的框架为估计CDOM度提供了一个可扩展的解决方案,支持环境管理和政策.
- 与特定的过量采样技术集成的DNN模型在处理不平衡的水生数据集方面表现出卓越的性能.
相关概念视频
Precipitation and Co-precipitation
Precipitation and coprecipitation methods can be used to separate a mixture of ions in a solution. In qualitative inorganic analysis, ions that form sparingly soluble precipitates with the same reagent are separated based on the differences in solubility products. For example, consider the separation of Cu(II) and Fe(II) ions by precipitation as insoluble sulfides. First, copper(II) sulfide is precipitated by the addition of acidic H2S, where the dissociation of H2S is suppressed. Adding H2S...
Sampling Methods: Sample Types
Sampling materials are classified into three main types: solid, liquid, and gas.
Solid samples include a variety of substances, such as sediments from water bodies, soil, metals, and biological tissues. Two standard methods for extracting sediments from water bodies are grab sampling and piston coring. Grab sampling involves using a device to collect a discrete sediment sample from the bottom of a water body with minimal disturbance. Grab samples do not always represent the entire area due to...
Solid samples include a variety of substances, such as sediments from water bodies, soil, metals, and biological tissues. Two standard methods for extracting sediments from water bodies are grab sampling and piston coring. Grab sampling involves using a device to collect a discrete sediment sample from the bottom of a water body with minimal disturbance. Grab samples do not always represent the entire area due to...


