We analyze the quality and noise characteristics of BDS triple-frequency observation of Huawei P40 mobile phone. Comparing with the measurement receiver, we find that the Hisilicon chip of P40 mobile phone has good capability of capturing BDS-3 signal. The signal-to-noise ratios of B1I, B1C and B2a signals are slightly lower than that of the measurement antenna, and there are system errors related to the terminal chip for the three signals. The external antenna zero-baseline results show that the zero-baseline positioning accuracy of Hisilicon chip of P40 mobile phone reaches mm level and the pseudorange and carrier phase noise have reached the level of measurement receiver noise accuracy. The pseudorange noise of B2a is smaller than 0.5 m, and the carrier phase noise of three signals are smaller than 2 mm.

In view of the nonlinear and high noise characteristics of ionospheric total electron content(TEC), we establish a short-term ionospheric combination prediction model based on empirical wavelet transform(EWT) and Elman neural network. We use the model to forecast the ionospheric TEC time series in different geomagnetic environments. The results show that EWT-Elman combination model can reflect the variation characteristics of ionospheric TEC. The average relative accuracy of the combination model during geomagnetic quiescence is 93%, and the root mean square error is 1.04 TECu. During geomagnetic disturbance, the average relative accuracy is 92.4%, and the root mean square error is 2.18 TECu. The highest average relative accuracy of the single Elman model, EMD-Elman model and EWT-BP model is 90.7%, and the minimum root mean square error is 1.33 TECu during geomagnetic quiescence. The highest average relative accuracy is 90.7%, and the minimum root mean square error is 2.57 TECu during geomagnetic disturbance. Compared with other models, the method in this paper has the best prediction effect.

In order to establish a high-precision BDS clock error prediction model, we propose a SAFA-FDGM(1,1) clock error prediction model based on the fractional-order discrete gray system optimized by the improved firefly algorithm. In order to avoid the firefly algorithm falling into the local optimal solution, and improve the firefly algorithm optimization ability. We introduce the inertia weight factor and improve the attractiveness factor and step factor, and use the improved firefly algorithm to automatically optimize the selection of the fractional-order factor of FDGM(1,1) model to improve the accuracy of FDGM(1,1) data fitting. We apply the clock error data of three different types of satellites, C02(GEO), C09(IGSO) and C12(MEO), for test analysis. The results show that the proposed clock error prediction model is superior to the traditional quadratic polynomial model and the GM(1,1) model, 3~6 h prediction error is less than 1 ns, and the 9~12 h prediction error is better than 2 ns, which has important reference value to establish BDS satellite universal clock error prediction model.

The integrated navigation system composed of strapdown inertial navigation system/celestial navigation system/global navigation satellite system(SINS/CNS/GNSS) can effectively improve the positioning accuracy of missile and other aircraft, which often use the launch point inertial coordinate system of launch time as the datum. Firstly, we deduce and establish a simple integrated navigation mathematical model of SINS/CNS/GNSS under the launch inertial coordinate system, and take the attitude quaternion error of the model as an estimation to avoid the conversion between the attitude error angle and the misalignment angle of the mathematical platform. Then, we analyze and reasonably simplify the magnitude of gravity error caused by positioning error, and an accurate measurement noise model is derived. Finally, we realize the effective integration of SINS/CNS/GNSS by using extended Kalman filter(EKF). Simulation results show that the proposed method is effective and beneficial to engineering implementation.

We analyze the multi-GNSS precise point positioning(PPP) strategy in the polar region based on GNSS and PPP technology. Through one-station solution per day and multi-station solution per day, we analyze the PPP results under 11 different navigation satellite system combinations. We conclude that the positioning results of BDS and Galileo combination is the worst among all the combinations, and the difference in positioning accuracy between the combination and the average value is 0.4-1.3 cm. In the polar region, the mean of convergence time in the N direction of each combination is 7-11 min, and the mean of positioning accuracy is better than 1.3 cm, which is significantly better than that in the E and U directions. In the same condition, the convergence time of the three-system combination in the E, N and U is reduced by 10.3%, 14.1% and 7.3%, respectively, the positioning accuracy is increased by 9.6%, 4.6% and 11.7%, respectively, compared with the two-system combination. Compared with the three-system combination, the convergence time of the four-system is shortened by 6.8%, -2.1% and 2.0%, respectively, and the positioning accuracy is improved by 4.9%, -7.1% and 5.3%, respectively. The results show that the positioning performance of the three-system combination is better than that of the two-system combination, and the positioning performance of the four-system combination is less improved than that of the three-system combination.

Aiming at the problem that the threshold model is too conservative due to the non-Gaussian features of the CNR and CCD monitoring quantities in the signal quality monitoring(SQM) process, we propose a threshold algorithm for SQM based on stable distribution. Firstly, we give two threshold models based on Gaussian distribution, and then derive the threshold model of monitoring quantities based on stable distribution. Finally, we evaluate and verify the threshold algorithm based on stable distribution by the measured data. The results show that the stable distribution can more accurately describe the CNR and CCD monitoring quantities, making the threshold envelope the monitoring quantity parameters tighter, solving the problem of the practical threshold model being too conservative and insufficiently accurate, reducing the risk and improving the integrity of the GBAS system.

In order to weaken the influence of non-linear changes on the velocity field model and noise model, we propose a method which combines empirical mode decomposition and environmental loading correction. To prove the practicality of this method, we choose 2 a time series from 10 CORS stations in Xuzhou region. Firstly, we compute the times series using Bernese software, and use the detrended time series to reconstruct the periodic terms and residual terms. Then, we apply the environmental loading products released by GFZ to correct the reconstructed periodic terms, and use the harmonic function fitting and environmental loading correction method as a control. The results show that empirical mode decomposition and environmental loading correction method can reduce the non-linear term about 18% in average, which is better than the harmonic function fitting and environmental loading correction method. The corrections in the north and east directions have no apparent effect on each station, but the vertical direction correction effect is significant. The integrated product correction effect is better than single product correction effect.

In view of the problem that the zenith tropospheric delay(ZTD) and precipitable water vapor(PWV) caused by the accumulated errors of the GPT2w model, we obtain actual data from seven radiosonde stations and two GNSS stations in Yangtze river delta region in 2017 to test the precision of parameters such as air pressure, temperature, water vapor pressure, weighted mean temperature (Tm) and ZTD of the GPT2w model, and combine the ZTD calculated by GNSS (GNSS-ZTD) and the Tm obtained from the GPT2w model to improve the PWV precision. The results show that the biases of air pressure, temperature and water vapor pressure near the ground are distributed at -3~4 mbar, -7~7 K and -9~2 mbar respectively, which is with high accuracy. The Tm obtained from GPT2w has good applicability in Yangtze river delta, with an annual average of -1.21 K bias and 6.89 K RMS. The mean bias and RMS of ZTD obtained from GPT2w model are 1.4 cm and 9.4 cm respectively, with precision that are significantly lower than that of GNSS-ZTD derived from actual meteorological data. The accuracy of PWV calculated by parameter fusion method is equivalent to GNSS-PWV. This method can be used to obtain PWV quickly when there is without actual meteorological parameters.

To reveal the spatio-temporal characteristics of terrestrial water storage evolution in China during 2018-06 to 2019-05, we propose the latest GRACE-FO(GRACE-Follow On) time-variable gravity field model released by CSR, JPL, and GFZ. At the same time, we analyze the effect of regional crustal structure differences on GRACE-FO gravity data to retrieve the vertical deformation of terrestrial water load by Hard and PREM models. The results show that the relative difference of radial load Love number with 60th order is -4.267%, and the largest difference of vertical load deformation with the first 60th order is located in the Lancang river area in Yunnan, with an amplitude of about 0.7 mm/a. Therefore, when using GRACE-FO gravity satellite data to estimate the vertical load deformation in China, the influence of regional crustal structure cannot be ignored.

Based on the seismic observation data from the Three Gorges digital seismic network, the cross-correlation and uncertainty between the seismicity and water level of the Three Gorges project in the impoundment process are systematically analyzed using the cross-correlation and rearrangement experiment technology, and the genetic mechanism of earthquakes in the Three Gorges reservoir is discussed. The results show that: 1)During the impoundment process, the crust of the reservoir head area experiences a complex process of “continuous loading-seepage saturation-rebound and rebalancing”. The spatial-temporal distribution of seismicity in the reservoir area is closely related to the temporal variation of reservoir water level. 2) From the beginning of impoundment to the 175 m experimental impoundment stage, the huge direct reservoir water load leads to the instability of mine caves, karst caves, shallow rock strata, faults and so on within 10 km along the river bank, thus inducing earthquakes. The shallow seismicity within 5 km of focal depth is obviously related to the elastic loading caused by the rapid rise of reservoir water level, showing an “instantaneous” response to the rise of reservoir water level. 3) With the stable operation of reservoir water in the 175 m experimental impoundment stage, reservoir water seepage gradually becomes the main influencing factor of seismic activity in the reservoir area, and the fault softening may be the reason for triggering the Zigui M4.5, M4.9 structural reservoir earthquakes in March 2014. The deep earthquake activity with focal depth greater than 5 km has a “lag” response to the reservoir water level. The Xiangxi reservoir section of the Yangtze river is a triangle structural area where Xiannüshan fault, Jiuwanxi fault and the Yangtze river intersect. There is a risk of medium-magnitude earthquake around this area.

Using GRACE(gravity recovery and climate experiment) monthly time-variable gravity field data from April 2002 to June 2017, we invert time series of equivalent water height and surface vertical and horizontal deformation in the area of Three Gorges Reservoir, Yangtze River basin and Amazon basin. A deep neural network model based on simple LSTM(long short-term memory) network is applied to predict time series data. The deep LSTM network could be extended deeper by stacking multiple LSTM hidden layers and adding linear layers in output layers. In addition, the attention mechanism is added to increase the ability of long-term characteristics extraction, and the genetic algorithm is used to select the best number of network layers and optimize some hyper parameters. In dynamic predicting mode, the NSE(Nash-Sutcliffe efficiency coefficient) is 0.907 9 at worst and 0.977 7 at best, and the values of R*(scaled root mean square error) are between 0.146 5 and 0.297 5. In static predicting mode, all of NSE values are better than 0.99, and the values of R* are less than 0.062 2, showing that the performance of deep LSTM network is very good.

On the basis of 28-scene descending orbit InSAR data of Sentinel in 2019, we monitor the average annual subsidence rate of Dongying city, and analyze and verify the time series of two interferometries and various deformation centers in different seasons. The results show that there are five subsidence areas in Dongying city, all located in the coastal area. The maximum annual rate of subsidence is nearly 300 mm/a. The subsidence characteristics are different from the types of underground fluid extraction, such as groundwater extraction, oil exploitation and brine development, and have obvious seasonal characteristics. In general, each deformation area shows a large linear rate of settlement from January to May, and rebounds to different degrees and then slowly subsides from June to December.The deformation trend of the interferometry profile also qualitatively verifies the correctness of time series deformation.

Based on the dense GNSS data from the multi-source projects, we calculate the high spatial resolution GNSS horizontal velocity field and strain rate field of the study area. Using seismic waveform data, the focal mechanism solution is solved by CAP method. On this basis, the damping regional stress inversion method is used to obtain the spatial distribution characteristics of the regional tectonic stress field in Sichuan-Yunnan block. Combined with the seismic activity characteristics of major faults, the crustal deformation characteristics in Sichuan-Yunnan region are comprehensively analyzed. The results show that: 1) The maximum principal compressive stress of the shallow part of the block and the maximum principal compressive strain rate of the surface change from SSE to near SN,which present good consistency. The deformation depth inferred from GNSS data can extend to at least 20 km. 2) The cause of eastward sliding and clockwise rotation of the Sichuan-Yunnan block are the pushing/thrust effect caused by the Tibet Plateau, the obstruction effect caused by the South China block, and the right-lateral shear dragging effect caused by northward movement decreases from west to east between the Indian plate and the Eurasian plate. 3) Under the effects of the clockwise rotation and the absorption and conversion of the strike-slip thrust fault zone caused by the comprehensive couple forces, the SE extrusion of material in the Tibet Plateau is limited.

Using the GPS data of 1999-2018 and the precise leveling data of two periods in 1974 to 2014(1974-1984, 2013-2014), we study the characteristics of present-day 3D crustal movement of Sichuan-Yunnan region. The results show that: 1) The horizontal movements are mainly inherited activities, and the rotation of southwest Yunnan has weakened obviously in recent years. The stress fields in the western Longmenshan fault and the eastern boundary of Sichuan-Yunnan rhombic block have readjusted due to the Wenchuan earthquake; the adjustments have weakened obviously in recent years, which are recovering to the pre-earthquake state. 2) The vertical movements exhibit a different uplifting on the whole, with the uplift rate in the northwest Sichuan-Yunnan obviously larger than that in the southeast. The uplift rate of the northern plateau of Longmenshan fault zone is 3-4 mm/a, the uplift rate of the plateau in the western Sichuan is 1-3 mm/a, the uplift rate of the northwest Yunnan is 1-2 mm/a, the subsidence rate of Simao and Jiangxi village in the southwest Yunnan is 2-3 mm/a, and the movement rate of other parts of Sichuan-Yunnan region is basically －1-1 mm/a. 3) There are normally some vertical differential movements on both sides of the main active faults, especially Xianshuihe and Longmenshan faults, which have segmented characteristics. 4) The horizontal movement and vertical movement in the Sichuan-Yunnan region show good consistency in geological structure. Recently, the horizontal movement of the plateau in the western Longmenshan fault has slowed down and the compression deformation has weakened. It may indicate that the region is gradually entering a new period of energy accumulation, which requires continuous attention.

We use GNSS measurements from Sichuan province to analyze the seasonal load displacement and stress-strain field. Combined with precipitation data, the effect of seasonal hydrological load on crustal deformation and stress-strain near active faults in this region are analyzed. The results show that there is a strong spatio-temporal correlation between the seasonal surface displacement and precipitation in Sichuan. The stress disturbance caused by the periodic change of hydrological load has a certain regulating effect on the stress accumulation mode of the active fault zone, which then affects the motion state of the active fault zone.

We use multi-source remote sensing data and SBAS-InSAR technology to carry out fault remote sensing interpretation and comparative analysis of the Taiyangshan fault zone and surrounding areas from multi-scale and multi-angle. The results show that there are 6 major faults in the region. Among them, the Taiyangshan fault zone has obvious structural and geomorphological characteristics, consisting of 4 main faults, namely Gangshi-Hefu fault, the west side of Taiyangshan fault, Xiaowupu fault, and Xianfengyu fault. Several main faults show abnormal geomorphological features such as linear scarps, fault valleys, lake boundaries, and mountain faults. SBAS-InSAR results show that from January 2017 to October 2019, the area has a cumulative maximum settlement of 62.59 mm and a maximum uplift of 59.42 mm. The fault controls the regional structural framework and affects the distribution pattern of ground uplift and subsidence. The deformation distribution characteristics of subsidence and uplift in the Taiyangshan area are consistent with the direction of tectonic belt, which indirectly reflects the accuracy of remote sensing interpretation.

In this paper, the typical coseismic waveforms collected by vertical pendulum broadband tiltmeter at Jixian seismic station in Tianjin and the data from the seismometer at the same site are studied. We obtain and identify the spectrum characteristics, frequency response and phase composition of coseismic waveforms of vertical pendulum broadband tiltmeter in Tianjin. Based on the precursory network observation data tracking and analysis platform, we count the amplitude and magnitude of vertical pendulum broadband tiltmeter in different regions, and fit the relationship between amplitude and magnitude by empirical formula, and realize the maximum amplitude conversion of surface wave of vertical pendulum broadband tiltmeter and broadband seismometer at Jixian seismic station.The results show that the vertical pendulum broadband tiltmeter has rich recording frequency band and high consistency with broadband seismometer in waveform.

In view of the problems of difficulty and high cost of obtaining labeled seismic data required for the training of existing models, we propose a new machine learning model CCLSN for seismic detection suitable for small samples. Through jointly using continuous wavelet time-frequency transform and redesigned lightweight convolutional neural network, the model greatly reduces the amount of labeled seismic data required for training and improves the applicability. The results show that CCLSN achieves high-accuracy and stable recognition function using a small-scale dataset containing only hundreds of samples. The accuracy and recall rate of CCLSN are both above 98%. CCLSN can provide a new technical approach for automatic seismic detection in areas with fewer or weaker earthquakes, such as central and eastern China.