Adaptive coloured noise multirate Kalman filter and its application in coseismic deformations

Changxin Chen,Xu Lin,Wei Li,Lin Cheng,Hongyue Wang,Qingqing Zhang,Zhen Wang

DOI: https://doi.org/10.1093/gji/ggad117

IF: 3.352

2023-03-17

Geophysical Journal International

Abstract:SUMMARY The accuracy and sampling rate of a coseismic displacement and velocity waveform can be improved by fusing Global Navigation Satellite System (GNSS) and strong motion (SM) sensor data with a multirate Kalman filter. However, many studies have shown that the noise in GNSS coseismic waveforms includes coloured noise, and it is challenging to obtain proper GNSS and SM fusion results if the GNSS coloured noise is not accurately represented. In this paper, we propose a coloured noise multirate Kalman filter, which uses a stochastic model for modelling the GNSS coloured noise to achieve an accurate fusion of data from GNSS and SM for different sampling rates. In addition, we also propose an estimation method for the multirate Kalman filter stochastic model under coloured noise to achieve the optimal adaptive fusion of GNSS and SM data. After the reliability of the proposed method was confirmed using Monte Carlo simulations and earthquake engineering data tests, the proposed method was applied to data collected from the 2019 Mw 7.1 Ridgecrest earthquake and 2016 Mw 7.8 Kaikoura earthquake. The test results show that the proposed method can effectively fuse GNSS and SM data and accurately obtain broadband coseismic displacement and velocity waveforms.

geochemistry & geophysics

What problem does this paper attempt to address?

This paper attempts to address the problem of how to effectively integrate Global Navigation Satellite System (GNSS) and Strong Motion (SM) sensor data in seismic monitoring to improve the accuracy and sampling rate of coseismic displacement and velocity waveforms. Specifically, the paper focuses on how to handle the colored noise issue in GNSS data, as this noise affects the fusion of GNSS and SM data. To solve this problem, the authors propose a Colored Noise Multi-rate Kalman Filter (CNMKF) and its adaptive version (ACNMKF), which accurately represent the colored noise in GNSS through a stochastic model and achieve optimal fusion of GNSS and SM data.

### Main Issues
1. **Colored Noise in GNSS Data**: Existing research shows that the noise in GNSS coseismic waveforms includes colored noise, and the traditional white noise model cannot accurately represent this noise, leading to suboptimal data fusion results.
2. **Multi-rate Data Fusion**: The sampling rates of GNSS and SM are different, requiring a filter capable of handling multi-rate data for effective fusion.

### Solutions
1. **Colored Noise Multi-rate Kalman Filter (CNMKF)**: By introducing a stochastic model to represent the colored noise in GNSS, it achieves accurate fusion of GNSS and SM data at different sampling rates.
2. **Adaptive Colored Noise Multi-rate Kalman Filter (ACNMKF)**: An adaptive method is further proposed to estimate the stochastic model parameters in CNMKF, achieving optimal data fusion results.

### Experimental Validation
1. **Numerical Simulation**: The effectiveness of the proposed method is validated through Monte Carlo simulations and Large High-Performance Outdoor Shake Table (LHPOST) tests.
2. **Application to Real Earthquake Data**: The method is applied to the actual data of the 2019 Mw7.1 Ridgecrest earthquake and the 2016 Mw7.8 Kaikoura earthquake. The results show that the method can effectively fuse GNSS and SM data to obtain high-accuracy coseismic displacement and velocity waveforms.

### Conclusion
The ACNMKF method proposed in the paper can effectively handle the colored noise issue in GNSS data and achieve high-accuracy, high-sampling-rate fusion of GNSS and SM data, making it suitable for coseismic waveform analysis in seismic monitoring.