一种气溶胶光学特性多参数多波段联合遥感反演方法、装置及存储介质
By employing a two-stage learning inversion structure and a multi-task joint inversion method, the problem of incomplete aerosol parameter acquisition in satellite remote sensing technology was solved, enabling accurate identification of aerosol types and effective monitoring of extreme events, thereby improving the model's stability and generalization ability.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- PEKING UNIV
- Filing Date
- 2026-04-22
- Publication Date
- 2026-07-17
AI Technical Summary
Existing satellite remote sensing technologies struggle to simultaneously acquire key parameters of aerosols, such as multi-band AOD, SSA, AF, and fine-mode AOD/coarse-mode AOD. This makes it difficult to distinguish aerosol types and sources, and data loss and noise are prone to occur under bright or complex ground surfaces. Furthermore, data-driven single-model schemes lack generalization ability.
A two-stage learning inversion structure is adopted. In the first stage, a gradient boosting decision tree model is used for dynamic strong nonlinear fitting. In the second stage, a neural network model based on self-attention mechanism is used for residual correction and information fusion. The atmospheric reanalysis dataset and aerosol climate dataset are used as prior background inputs to output multi-parameter aerosol characteristics.
It improves the ability to identify aerosol types and resolve sources, reduces overfitting, enhances the generalization ability across time and sites, is suitable for monitoring aerosol changes under extreme events, and reduces missing data and noise on bright surfaces and complex underlying surfaces.
Smart Images
Figure CN122090283B_ABST