一种基于多源遥感数据融合的水质参数反演方法及系统
By employing optimization strategies involving multi-source remote sensing data fusion and a one-dimensional convolutional neural network model, the limitations of data sources and model overfitting in remote sensing water quality monitoring were addressed. This enabled high-precision, large-scale monitoring of water quality parameters and improved the effectiveness of water quality parameter inversion.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- HEBEI UNIV OF ENVIRONMENTAL ENG
- Filing Date
- 2026-04-22
- Publication Date
- 2026-07-17
AI Technical Summary
Existing remote sensing water quality monitoring methods suffer from problems such as long sampling cycles, limited coverage, high labor costs, limitations of single remote sensing data sources, and model overfitting, making it difficult to achieve large-scale, high-frequency, multi-source data fusion and high-precision water quality parameter inversion.
A multi-source remote sensing data fusion method was adopted, combining a one-dimensional convolutional neural network model and a multilayer perceptron model. Through optimization strategies such as smoothing L1 loss function, AdamW optimizer, Dropout layer, L1 regularization and hierarchical K-fold cross-validation, a one-dimensional convolutional neural network model was constructed. By fusing UAV hyperspectral and satellite remote sensing data, local spectral features and long-range dependence features were extracted, and predicted values of water quality parameters were output.
It achieves high-precision and high-stability prediction of chlorophyll a and turbidity, improves the generalization ability of the model, overcomes the saturation effect of a single data source, and improves the accuracy and coverage of water quality parameter inversion.
Smart Images
Figure CN122409536A_ABST