Block Optimization and Outlier Detection in GEO SAR Turbulent Tropospheric Error Compensation | IEEE Journals & Magazine | IEEE Xplore

Block Optimization and Outlier Detection in GEO SAR Turbulent Tropospheric Error Compensation


Abstract:

Geosynchronous synthetic aperture radar (GEO SAR) has important potential applications in the fields of the atmospheric water cycle and digital weather forecast, and esta...Show More

Abstract:

Geosynchronous synthetic aperture radar (GEO SAR) has important potential applications in the fields of the atmospheric water cycle and digital weather forecast, and establishing the connection between GEO SAR imaging and atmospheric water vapor inversion is a key step. The turbulent variation of atmospheric phase delay will cause the SAR image to defocus, and turbulent tropospheric delay (TTD) phase estimation based on the block autofocus method in low-orbit high-resolution space-borne SAR can be used for reference, but this method is very sensitive to the data block selection, modulation frequency (FM) outlier detection, and correction. This letter gives some consideration to these two problems of block selection and outlier correction. Based on the idea of an extreme learning machine (ELM), a mathematical model with nonlinearity as the core and a linear model as easy to solve is established. When the parameters affecting the turbulent troposphere are known, the model can be directly used to obtain the block optimization selection, namely, the number and size of blocks. When the parameters are unknown, the model can be used to roughly inverse the parameters and provide initial values for iteration. For the second problem, using the prior information on the correlation of the FM estimated by the adjacent subblocks, we propose a local outlier factor (LOF) method to identify the error point by defining the distance threshold. The effectiveness of the proposed algorithm is verified by Sentinel-1 data injected with tropospheric error.
Published in: IEEE Geoscience and Remote Sensing Letters ( Volume: 20)
Article Sequence Number: 4003805
Date of Publication: 20 March 2023

ISSN Information:

Funding Agency:


References

References is not available for this document.