A New Method to Obtain 3-D Surface Deformations From InSAR and GNSS Data With Genetic Algorithm and Support Vector Machine

• Published in: IEEE geoscience and remote sensing letters Vol. 19; pp. 1 - 5
• Main Authors: Ji, Panfeng, Lv, Xiaolei, Yao, Jingchuan, Sun, Guangcai
• Format: Journal Article
• Language: English
• Published: Piscataway IEEE 2022; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: 3-D deformation; Algorithms; Deformation; Deformation effects; Earth surface; Estimation; Gene mapping; genetic algorithm (GA); Genetic algorithms; Global navigation satellite system; global navigation satellite system (GNSS); Interferometric synthetic aperture radar; interferometric synthetic aperture radar (InSAR); Learning algorithms; Learning theory; Machine learning; Navigation; Navigation satellites; Navigation systems; Navigational satellites; Parameters; SAR (radar); Satellites; Solid modeling; Strain; Support vector machines; supported vector machine (SVM); Synthetic aperture radar; Synthetic aperture radar interferometry; Training
• ISSN: 1545-598X; 1558-0571
• DOI: 10.1109/LGRS.2021.3049128

In this letter, a new technique based on genetic algorithm and support vector machine (GA-SVM) is proposed to effectively estimate the 3-D deformations of the earth's surface by integrating sparse global navigation satellite system (GNSS) deformation measurements and interferometric synthetic aperture radar (InSAR) maps. The genetic algorithm (GA) is used to search the optimal supported vector machine (SVM) control parameters, considering the control parameters have an important influence on the prediction. Based on advanced machine learning theory, the proposed method has at least two main advantages over traditional methods: 1) it does not need to preinterpolate the displacements of GNSS points, and 2) it does not need to estimate the variance components of GNSS and InSAR point by point. Both the simulated and real experiments are implemented to prove the effectiveness of GA-SVM. In the real case of the Los Angeles, the root mean square errors of GA-SVM at 14 checkpoints are 7.92, 2.05, and 5.43 mm/a in the east-west, north-south, and vertical directions, respectively.