Based on Sentinel-1A ascending and descending data and D-InSAR, the coseismic deformation field of the Luding M_S6.8 earthquake is obtained. The source parameters of the earthquake are inverted based on the elastic half-space dislocation model, the slip distribution on the fault plane is determined by using the distributed slip model, and the static Coulomb stress change of the coseismic dislocation to the surrounding faults is calculated. It is shown that the maximum deformation cause by the earthquake is about 18 cm in the direction of radar line of sight (LOS). The coseismic dislocation is mainly left-lateral strike-slip at the depth of 6-24 km. The maximum slip is about 2.5 m at the depth of 16 km and the main seismogenic fault is the Xianshuihe fault. The static Coulomb stress triggering relationship shows that the Coulomb stress of the southern segment of Xianshuihe fault and the northern end of Anninghe fault increases significantly, and the strong earthquake trend is worthy of attention.

Based on Sentinel-1A ascending and descending image data, D-InSAR is used to obtain the coseismic deformation field in line of sight direction of the 2022-09-05 Luding earthquake. First, the Bayesian method is used to search the priori geometric parameters of the fault, and the non-negative least squares principle is used to invert the fine slip distribution of the fault, then the Coulomb stress variation in the  area near the epicenter is calculated based on the parameters of the fault slip distribution, and finally the interseismic GPS velocity field data are used to calculate the interseismic strain field in the seismogenic region. The results show that: 1) The maximum deformation of the coseismic deformation field in line of sight direction of the Luding earthquake is 15 cm; 2) The Luding earthquake is a typical left-lateral strike-slip earthquake with a fault strike of NNW-SSE direction of about 167°, rupture along the strike direction is about 55 km, dip angle is about 74°, rupture depth is mainly 0-17 km, maximum slip is about 1.12 m, corresponding to a depth of 1 km, and release a total seismic moment 1.02×10¹⁹ Nm, corresponding to a moment magnitude of M_W6.64; 3) The south-east section of the Xianshuihe fault zone, the north section of the Anninghe fault zone, and the middle-north section of the Yulongxi fault are in a stress-loaded state, and there is a greater possibility of future earthquakes; 4) The source area of the Luding earthquake is located in the transition region between tensile and extrusion strains, and this strain transition region may be related to the rendezvous of several different active blocks in the area.

To eliminate the disadvantages of the BP neural network model in predicting the surface subsidence of the mining area, which is limited in accuracy stability, we take a mining area as an example. We select nine factors and the maximum subsidence value affecting the surface subsidence of the mining area, including the elastic modulus, Poisson’s ratio, cohesion, et al, as initial sample data. The BP neural network is optimized by Kalman filtering(KF), and then the constructed KF-BP model is regarded as a weak predictor in adaptive boosting(AdaBoost) algorithm, and each weak predictor is weighted and combined into a strong predictor through the final weight distribution. This study uses MATLAB to establish BP neural network model, KF-BP model, AdaBoost-BP model and AdaBoost-KF-BP model to train and predict the actual settlement monitoring data of the mining area. The results show that AdaBoost-KF-BP model has the highest stability, and its accuracy is significantly improved compared with other models.

Based on sequential adjustment, we use progressive SBAS-InSAR to process the new images. Using the obtained results derived from SBAS-InSAR, only a small part of the existing images is combined with the newly acquired data to update corresponding time series deformation of all SAR images. Taking the surface deformation in Beijing plain from June 2019 to April 2020 as an example, the experimental results show that the mean and RSME of the surface deformation rate difference between the progressive and the traditional SBAS-InSAR are 0.01 mm/a and 0.1 mm/a, respectively. Moreover, the processing time of progressive SBAS-InSAR is just 1/5-1/2 times that of SBAS-InSAR.

Combined with single-day vertical displacement from GPS and environmental load deformation data, this paper analyzes the surface deformation characteristics of typhoon Mangkhut at continuously operating reference stations in Hong Kong from September 9 to 23, 2018. Both data processing results show that the ground surface of Hong Kong has a certain degree of settlement during the typhoon. The overall variation is that as vertical distance from Hong Kong is closer, the impact of the typhoon on surface subsidence in Hong Kong area increases. The two methods are more consistent in the trend of vertical surface deformation, but there are differences in the values, mainly for two reasons: 1) The surface vertical displacement calculated by GPS in a short period of time includes the influence factors of temperature, latitude, and solution error; 2) Due to the lack of relevant information on surface water and groundwater, there are large errors when the model simulates the influence of various factors on surface water and groundwater.

Taking the Qinghai Gashari superlarge landslide as an example, we use the transfer coefficient method to carry out the conventional analysis of landslide stability, and then using cusp mutation analysis and BAS-NARX-ARIMA model, we carry out the corroborative evaluation and development trend analysis of landslide stability. The analysis shows: 1) In the stability analysis results of the transfer coefficient method, the stability coefficient Fs of the main sliding surface under three working conditions ranges from 1.148 to 1.697, which is basically stable to stable; 2) The cusp mutation analysis shows that the Δ values of each monitoring point are greater than 0 and they are in a stable state, proving the accuracy of the above conventional stability calculation results. However, there are certain differences in the degree of stability at different locations; 3) The BAS-NARX-ARIMA model has good robustness and prediction effect in predicting of landslide deformation. According to the extrapolation prediction results, the landslide stability tends to weaken over time. We suggest that it is necessary to take practical measures to avoid disaster losses.

This paper proposes an adaptive UKF algorithm for GNSS/SINS integrated navigation system, aiming at the lack of adaptive adjustment ability of UKF algorithm to measurement noise anomalies. Firstly, this paper carries out the nonlinear modeling of GNSS/SINS integrated navigation system. Then, based on the variational Bayesian principle, the prediction model of measurement noise variance is introduced in the process of time update and measurement update of UKF algorithm. Finally, the GNSS/SINS integrated navigation system based on this adaptive UKF algorithm is simulated and verified. The results show that the proposed algorithm can accurately track the sudden change or slow change of the measurement noise variance in real time, and obviously improve the accuracy of the integrated navigation system compared with the classical UKF algorithm.

We compare the difference of BDS attitude models published by different analysis centers, and find that the difference of the yaw angle can be nearly 180° at noon and earth shadow time. Furthermore, we compare the effects of three attitude strategies, namely quaternion attitude, model attitude and nominal attitude, on the phase wind-up correction and the final positioning results of precise point positioning(PPP). The results show that different attitude processing models will bring phase wind-up correction difference of nearly 1 cycle. Compared with the nominal attitude RMS, the PPP positioning results of quaternion attitude decreased by 17%, 25%, 34%, respectively, and compared with the model attitude RMS, decreased by 11%, 7%, 7%, respectively. Keeping the consistency of products of the analysis center can improve the precision of PPP.

Aiming at the problem of deterministic errors in the output of the IMU array, we propose an online calibration method based on the Levenberg-Marquardt(LM) optimization algorithm. It calibrates the deterministic errors of accelerometers and gyroscopes by rotating and standing IMU array without any external reference devices. Firstly, we calibrate the accelerometers by static data from multiple positions, then we calibrate the gyroscopes by the calibrated accelerometers and the rotation data between positions. The simulation result shows that the residual accuracy is close to the theoretical value. Three sets of vehicle experiments show that the dynamic navigation accuracy is increased by an average of 19.1% after calibration, indicating that the method can calibrate errors and improve the navigation accuracy effectively.

To solve the problems that support vector regression(SVR) models cannot actively select optimal parameters and kernel functions, we optimize the support vector regression model by the aquila optimizer(AO) and construct the aquila optimized support vector regression (AO-SVR) model. The four models, AO-SVR and SVR, gray wolf optimized support vector regression(GWO-SVR), and whale optimization algorithm support vector regression(WOA-SVR), are combined with atmospheric pollutants, meteorological factors and hourly zenith tropospheric delay(ZTD) data in five cities of Lhasa, Urumqi, Changchun, Wuhan, and Shanghai from 2020-01-01 to 30 to predict the changes of PM_2.5 concentrations in the five cities on 2020-01-31, respectively. The results show that the AO-SVR model has better applicability; the predicted values in Shanghai are the closest to the actual observed values.

Based on the processing, interpretation, and comprehensive study of the two-dimensional seismic survey data in the key survey area of Xiongan New Area, we summarize the buried depth of the top and bottom of the main strata, the contact relationship, and the spatial distribution characteristics of fault structure within the depth of 3 000 m. According to the study, 16 sets of strata, including the Jixian system of the Middle Proterozoic, and 9 faults including Rongcheng, Rongnan and Xushui faults, developed within 3 000 m depth in the key investigation area of Xiongan New Area. The faults are hidden under the surface, all of which are high-angle normal faults, mainly in a north-eastern direction, controlling the uplifts and sags pattern and the spatial distribution of the Quaternary Lower Pleistocene, Neogene, Paleogene, Mesozoic, Qingbaikou and Jixian systems in the study area.

The 2022-04-06 Xingwen M_S5.1 earthquake triggered three seismic monitoring instruments laid on the shatter slope in Gongxian county, Yibin city, and the seismic wave propagation data was completely recorded. Through data filtering and baseline correction, we analyze the time history curve, Fourier spectrum and acceleration response spectrum. The results show that:  1) The holding time of this earthquake is short, the amplitude is small, and the S-wave holding time of the 2# and 3# monitoring points is 2-5 s longer than that of the 1# monitoring point.  2) Fourier spectrum shows low frequency attenuation, medium and high frequency amplification, and frequency band widening in the horizontal direction, but the change in the vertical direction is not obvious.  3) Take the 1# monitoring point as reference, the peak acceleration(PGA) amplification coefficient of 2# and 3# monitoring points is 1.17-3.05, the Arias intensity(AI) amplification coefficient is 4.43-7.59 in the EW and UD directions, and the NS direction of the 3# monitoring point can reach 18.86. 4) Compared with the monitoring profile of Renjia village(non-shatter slope) in Lushan, we believe that there is an amplification effect of the Gongxian shatter slope.

We design and implement a nondestructive measurement method for the spatial distribution of coseismic surface rupture and an automatic measurement tool based on the Python platform; the method can obtain the spatial distribution and width of surface fractures with the same degree of precision as the original data. We take the Maduo M_W7.4 earthquake in 2021 and the Idaho Lost River M_W6.9 earthquake in 1983 as examples. The results show that this method can measure and map the spatial distribution of surface rupture automatically and flexibly.

Using data from the China continental tectonic environment monitoring network, we obtain the vertical displacement data of 26 GNSS stations in Yunnan province from 2011 to 2019, and calculate the drought severity index based on GNSS vertical displacement (GNSS-DSI) to analyze hydrological drought events in Yunnan region. The results show that there is good correlation between GNSS-DSI and the drought severity index based on GRACE observations (GRACE-DSI), and 81% of the stations have a moderate-strong correlation between the two DSI data(0.45-0.78). GNSS-DSI can detect 7 moderate to extreme drought events in Yunnan from 2010 to 2019, and provide quantitative characterization of these hydrological drought events.

Using the relative gravity observation data of Hebei and surrounding areas from 2015 to 2020, we obtain the dynamic variation of the gravity field. With the polynomial fitting method to calculate annual variability, we also analyze the characteristics of the gravity field changes in Hebei. The results show that the dynamic gravity change presents obvious regional characteristics. The gravity variation is stable in the northwest, within -40~30 μGal; while in the southeast there exists a slight positive accumulation from 2015 to 2016, which spreads to the northeast, forming a NE-SW elliptical anomaly between the Tianjin-Handan area in 2020. The anomaly ranges from Yanshan in the north to Taihang mountain in the west, with a maximum difference of 120 μGal, also with a high average change rate. Combined with the water level and surface subsidence data, the gravity variation in the southeastern high-value area is mostly caused by surface subsidence, the influence of groundwater loss is covered relatively.

We propose a single station prediction model of geomagnetic variation based on chaos theory and RBF neural network. We analyze the chaotic characteristics of magnetic field data, and obtain the embedding dimension m  and time delay τ.Based on this, we reconstruct the phase space. The sample set optimized by chaos theory is used as the training and test set of the neural network for simulation experiment. The results show that the RBF neural network model improved by chaos theory can accurately predict the change trend of geomagnetic field and has good applicability to geomagnetic field in China.

Based on the basic contents of finding the mean value, eliminating the trend change, spectrum analysis, self-defined function fitting, and drawing the groundwater temperature gradient curve from seismic underground fluid using the MATLAB software, we calculate the tidal factor and barometric factor of groundwater level, and discuss the elimination of tide and pressure components of groundwater level.

Three typical groundwater level data of nine solid tide observation wells are selected and interpolated with five interpolation methods. The results show when the number of missing values is small, the cubic polynomial interpolation is the best. For groundwater level with large trend change, and the solid tide is suppressed, the linear interpolation effect is the best. For groundwater level with obvious solid tidal effect and gentle change, the ARMA model prediction method is better than other methods. For groundwater level with clear solid tide and short-term fluctuation, linear interpolation and ARMA model prediction method have their own advantages.

We propose an intelligent fault identification method for the VP-type tiltmeter. Using empirical mode decomposition(EMD), we decompose the normalized fault signal into six intrinsic mode functions(IMF), and calculate the approximate entropy respectively to construct the EMD multiscale approximate entropy input matrix. Combined with the grasshopper optimization algorithm(GOA), we optimize the parameters of the self-organizing feature map(SOM) neural network. Then, we embed the obtained GOA optimal value, and form the GOA-SOM identification model. We use the identification test set to obtain the cluster label value of the target, and compare it with the cluster label of the training set and the real fault type, to obtain the fault identification result. The experiments show that the GOA-SOM model has a mean and standard deviation of the identification accuracy under 100 random sampling conditions of 99.329 7% and 1.218 8. These are better than traditional models.

This paper establishes the linearity models based on least squares method and terminal-based straight line method of seismic data acquisition device respectively. It analyzes the source and gives the calculation formula and the evaluation process of uncertainry. It compares and analyzes the differences between the least squares method and the terminal-based straight line method in the evaluation of linearity uncertainty and the influence of the number of input points on the evaluation of linearity uncertainty. The results show that the linearity uncertainty evaluation of the methods are on the same order of magnitude, and the terminal-based straight line method is less affected by the number of input points.