COSMIC-2 provides continuous atmospheric data for improving weather forecasts, which are helpful for monitoring and warning of extreme weather. We divide Chinese mainland into five regions according to climate type and analyze the accuracy of COSMIC-2 PWV in Chinese mainland based on the water vapor sequence of CMONOC station. From the perspectives of distance and elevation between stations, we design the PWV matching scheme of COSMIC-2 occultation points and GNSS stations. We evaluate the accuracy of COSMIC-2 for different climate types in China. The results show that the RMSE of plateau mountain, temperate continental, temperate monsoon, subtropical monsoon, and tropical monsoon climate types are 1.40 mm, 1.49 mm, 2.68 mm, 3.11 mm, and 3.16 mm, respectively. The accuracy in inland area is better than that in coastal area.

Using multi-layer perceptron(MLP) on sounding data from 2015 to 2017 in southwest China as the experimental data, we establish the weighted mean temperature (T_m) model for the area. We use meteorological parameters (surface temperature, water vapor pressure) and non-meteorological parameters (elevation, latitude, and day of year) as model input factors, and use the T_m calculated by the numerical integration method as the learning target. We apply the neural network model for iterative training to obtain the T_m in southwest China. Using the T_m data of the sounding stations in 2018 as reference values, we verify the accuracy of the MLP model and compare it with the Bevis and GPT3 models. The results show that the mean annual RMSE and annual bias of the MLP model are 1.99 K and 0.15 K, respectively. Compared with the Bevis model and the GPT3 model, the mean annual RMSE is reduced by 1.36 K (40.6%) and 1.51 K (43.1%), respectively. The bias drops by 0.70 K(82.4%) and 1.04 K(87.4%), respectively, and the accuracy and stability of the model in different elevations, latitudes and seasons in southwest China are better than the Bevis and GPT3 models.

Using GFZ and WHU precise products, we compare and analyze the accuracy of GNSS precise products. We then select 18 MGEX stations distributed in the eastern hemisphere for BDS/GPS/GLONASS/Galileo PPP experiments based on WHU precise products and investigate the PPP performances of the BDS B1C/B2a and B1I/B3I ionosphere-free combinations. The results show that the accuracy of GNSS orbit products is at the cm level, and the STD of GNSS precise clock is better than 0.1 ns. We find that BDS-3 satellite clock accuracy is significantly improved compared to BDS-2. The PPP experiments show that the PPP performance of BDS is the best in the Asia-Pacific region, and in non-Asia-Pacific region, GPS positioning accuracy is better. The convergence times of BDS static and kinematic PPP are 50 mins and 78 mins, which are lightly slower than that of GPS and Galileo, and better than that of GLONASS. The comparison shows that the PPP performance of the B1C/B2a and B1I/B3I dual-frequency combinations is basically comparable.

To solve the problem of the slow convergence speed of the existing single-system, GPS PPP, we select the LEO constellation system composed of different numbers of satellites. According to the observational value simulation principle, we obtain the pseudorange observations and phase observations of the points to be estimated from low-orbit satellites. To analyze the enhancement effect of LEO constellation, we select CPVG, WUH2, KOUG, IISC, FUCN stations, and use GPS and LEO constellation observation data with additional orbital error to perform static precise single-point positioning. The results show that: 1) After adding the LEO constellation to the original GPS system, the average number of visible satellites increased by 9.7, 19.4 and 31.3, and the PDOP value decreased by 0.7, 0.9 and 1.1, respectively; 2) The convergence time was shortened rapidly, the single system of GPS satellites were reduced from 17.1 min to 4.7 min, 2.1 min, and 1.5 min, respectively, and the convergence time was shortened by 67.3%, 84.4%, and 89.5%, respectively; 3) The RMS value in the N direction was increased by 37.4%, 60.3%, and 67.3%, respectively, the E direction increased by 34.4%, 59.4%, and 72.9%, respectively, and the U direction increased by 55.9%, 71.7%, and 79.6%. LEO constellation greatly improves the performance of GPS PPP, which is of great significance for guiding the design of LEO constellation enhanced GPS system and the expansion of application scenarios for precise single-point positioning.

We propose a real-time anomaly detection method for BDS broadcast ephemeris. First, by checking whether the broadcast ephemeris parameters are exceeded, we judge the anomalous broadcast ephemeris. Second, to realize the consistency-based real-time anomaly detection of the broadcast ephemeris, we calculate the SISRD of broadcast ephemeris of continuous epochs at intermediate time. Finally, we carry out the anomaly detection with BDS broadcast ephemeris data of the ALIC, AREG, ABMF and GENO station, and use the anomaly results detected by the final precision ephemeris as a reference for verification. The results show that the proposed method accurately detects the anomalies of broadcast ephemeris.

This paper focusses on southeast Gansu and its adjacent areas(N32°-36°, E102°-106°) from January 1, 2013, to November 1, 2021. We use the epicenter location data of 11 659 earthquakes in the earthquake catalogue, the K-means in the hard clustering method and the Gaussian mixture model GMM clustering method of the soft clustering method to cluster the spatial location of the original earthquake catalogue and the earthquake fine positioning catalogue. To determine the optimal cluster number, we use AIC and BIC model selection methods. Finally, we divide the original catalogue into 6 seismic clusters, and the fine positioning catalogue into 14 seismic clusters. The results show that the combination of fine positioning seismic catalogue and GMM clustering method can better find seismic clusters with different spatial distribution characteristics from seismic big data.

We calculate the apparent stress and b value using waveform data and earthquake catalogue respectively to analyze the stress change process of 2021 Luxian M_S6.0 earthquake. Using Schuster statistical test method and statistical parameter P, we statistically test the correlation between small and medium earthquakes and the change of earth rotation rate before the earthquake. We analyze the instability of the focal area. The results show that the apparent stress of M_L3.2-3.9 and M_L3.5-3.9 shows an upward trend for more than 7 years before the earthquake, and the apparent stress before the main earthquake reaches twice the early level. Before the earthquake, the b value experienced a decline of 7 years, from 1.2 in 2014 to 0.8 in 2020. There is a strong negative correlation between apparent stress and b value before the earthquake, reflecting the stress enhancement process near the epicenter. The P value of earthquakes with M_L≥2.0 near the epicenter is less than 1% from August 2020 to February 2021, indicating that the seismicity in the focal area is significantly related to the earth’s rotation about half a year before the Luxian earthquake during which time the crustal medium in the focal area was in an extremely unstable state.

Based on the seismic data for Luxian M_S6.0 earthquake provided by Sichuan and surrounding provinces’ seismic network, using the CAP inversion method, we calculate source parameters, like focal depth, moment magnitude and focal mechanism. From the focal mechanism of 155 events occurring in those regions, using the MSATSI code, we derive the tectonic stress field in the Sichuan basin before the earthquake. The results indicate that the focal depth of the Luxian earthquake is 3 km, and the moment magnitude is M_W5.36 with pure thrust faulting rupture. Tectonic stress before the event shows that the directions of both maximum and the intermediate principal compressive stress are almost horizontal, and the strike angle of the maximum principal compressive stress is about 101° near the epicenter. Under such stress framework, the optimal sliding fault is pure thrust, completely consistent with the focal mechanism of the Luxian earthquake. This event faulting is mainly controlled by the lateral force, suggesting that the southwestern part of the Sichuan basin is involved in the eastward expansion of the Tibetan plateau. This event is the result of activation of the shallow detachment fault under this tectonic background. Based on the source properties and background stress, we infer that the seismogenic fault of the Luxian earthquake may not be the Huayingshan fault, but a very shallow caprock slip-type fault between its branches.

Based on the digital waveform data of the regional seismograph network in Hubei province, using seismological and geological methods, we obtain the present crustal tectonic stress field in the northeastern margin of Jianghan basin. The results show that the crust in the studied area is moving in a SEE direction under the influence of the horizontal force source in the NWW direction, leading to the strike-slip shear failure with a small amount of thrust between the Zaoshi and Changjiangbu faults within the depth of 4-6 km. The seismic swarm activity is related to the fault activity since the Paleogene.

By limiting the diameter of the earthquake swarm concentration area, and deducing the same ray path of two earthquakes at similar distances recorded by a single seismic station, using the method of double difference wave velocity ratio, we carry out characteristic analysis of the wave velocity ratio of the two earthquake swarms in Changdao area from 2017 to 2018. Through the sensitivity analysis of seismic phase data, we conclude that the method of double-difference wave velocity ratio is comparatively sensitive to the noise in arrival time, especially for Pg-wave. The phase errors of Pg-wave and Sg-wave should be controlled within 0.02 to 0.20 s. The final analysis results show that the wave velocity ratio of Changdao earthquake swarm fluctuates in the range of 1.69-1.78, and the change of wave velocity ratio is closely related to the activity process of earthquake swarm. Before the occurrence of several large aftershocks, the wave velocity ratio showed a low state. The wave velocity ratio between main shock and aftershock varies, which may reflect differing seismogenic mechanisms.

To analyze the temporal and spatial evolution characteristics of wave velocity ratio before and after the January 2, 2022 Ninglang M_S5.5 earthquake, we use the Wadati method of single station and multi-earthquake, and calculate the phase arrival time data of earthquakes that meet the calculation conditions recorded by Yunnan digital seismic network from January 1, 2020 to January 10, 2022. We analyze the wave velocity ratios of 59 stations of Yunnan and the wave velocity ratio of post-earthquake sequence recorded by the LGH and NLA stations closest to the epicenter. The results show that: 1)Ninglang earthquake occurred in the area with high wave velocity ratio in northwest Yunnan, and the high value area of wave velocity ratio is tending to expand with time. It may be that in the process of earthquake preparation, there is fluid infiltration in the cracks near the epicenter to reach the state of water saturation, resulting in the increase of wave velocity ratio. 2)The wave velocity ratios of the 4 stations within 100 km of the epicenter show the evolutionary process of continuous descent and earthquake, the time of wave velocity ratio begins to decline at the LGH station closest to the epicenter later than that of the other three stations. The preliminary analysis shows that the occurrence time of wave velocity ratio is negatively correlated with the epicentral distance. 3)There is better correspondence between the wave velocity ratio change of post-earthquake sequence and the activity of aftershocks, the wave velocity ratio sudden drops before the largest aftershock. When the earthquake sequence shows normal attenuation trend, the wave velocity ratio maintains stable low value.

To carry out statistical analysis of the spatio-temporal evolution of the electric field during the M_S5.1 earthquakes in Songyuan, Jilin province and Tangshan, Hebei province, we use four frequency bands of the power spectrum density (PSD) data in the ULF/ELF frequency band recorded by the electric field detector (EFD) of the first seismo-electromagnetic satellite-CSES in China with geomagnetic active indexes Dst≥-30nT and K_p<3. The analysis shows that the perturbation amplitudes in the four frequency bands all increased to the high perturbation stage, which reached more than 2 times the standard deviation before the earthquake, and the earthquake occurred during the descent stage during the M_S5.1 earthquake in Tangshan, Hebei province. Then, the perturbation amplitudes recovered, and it was obviously abnormal in the second frequency band. The perturbation amplitudes in the four frequency bands are very similar, and they all reached more than 2 times the standard deviation before the earthquake. The perturbation amplitudes in the four frequency bands are different during the M_S5.1 earthquake in Songyuan, Jilin province. There, only the perturbation amplitude in the second frequency band reached more than 2 times the standard deviation. The perturbation amplitudes of the both earthquakes were obviously abnormal in the second frequency band (371-879 Hz) before both earthquakes, which is considered to be the dominant frequency band for seismic anomaly.

Based on the catalogue of small earthquakes in China, using the temporal and spatial scanning method of earth tide modulation ratio, we analyze the temporal and spatial variation characteristics of earth tide modulation ratio before moderate-strong earthquakes in the capital circle area of China since 1987. The results show that seven of the nine earthquakes (M_S≥5.0) present abnormal modulation ratios, which occurred within 1 a before the main shock and within 100 km from the epicenter. The seismic correspondence ratio is 78%. The abnormal high value intensity usually lasts about 3~5 months. Most disappear in the months before the earthquakes. The analysis of the earth tide modulation ratio shows a good earthquake reflecting ability, and it offers a useful reference for the prediction of moderate-strong earthquakes in the capital circle area of China.

To comprehensively and accurately identify the potential landslide hazards on both sides of the main stream of the Xiaojinchuan Banfenmen-Dawei reach, based on the SBAS-InSAR technology, we use the data analysis results of Sentinel-1 lift rails combined with complementary methods to identify the potential landslide hazards in this area at an early stage. The analysis and processing results show that: 1) The combination of the ascending and descending orbit data effectively avoids the shadow overlapping area and can comprehensively identify the hidden danger points of long-term creep of the bank slopes on both sides. 2) The LOS direction deformation rate is -120-10 mm/a from the Sentinel-1 descending orbit data, and -131-10 mm/a from the ascending orbit data. 3) Based on the deformation results, 35 landslide hidden danger points are identified, including 8 typical landslide hidden danger points. The deformation characteristics of the landslide hidden danger points are reviewed on site. The road on the slope is obviously broken, with developed cracks in the deformation area. These led to the dislocation of the platform, which verifies that the slope was in a long-term creep state.

The Yellow river delta region has many wetlands, farmlands and large areas, making it difficult for PS-InSAR technology to obtain high-density surface deformation information. In this paper, we study a surface deformation monitoring method of Yellow river delta based on distributed scatterers InSAR(DS-InSAR). In this method, we select homogeneous pixel points through confidence interval estimation, then we extract the corresponding phase value of dominant scatterers by eigenvalue decomposition method to achieve phase optimization. We determine distributed scatterers according to spatio-temporal coherence, and finally solve time-series surface deformation information. Using 26 Sentinel-1A images as data sources, we extract the surface subsidence information of the Yellow river delta from December 2019 to December 2020. Compared with the results of the PS-InSAR method, the point density increases by 5.56 times. The correlation coefficient between the points with the same name and the deformation rate obtained by this method is 0.727, indicating that the two methods are in good agreement. The experimental results showed that there are four obvious subsidence areas in the studied area, and the maximum subsidence rate was -238 mm/a. Analysis and field investigation show that the main influencing factors are underground brine and oil and gas exploitation.

We use SBAS-InSAR technology to process 65-view Sentinel-1A SAR data of Heze to obtain the subsidence results of the city from 2017-05-20 to 2021-05-23. We finely analyze the ground subsidence of each imaging period in combination with the underground coal mining, and the actual level data are used to SBAS-InSAR monitoring results for accuracy verification. The results show that the ground subsidence in Heze accelerates continuously during studying period, and the subsidence in Yuncheng area is more serious, with the maximum annual average subsidence rate reaching -311 mm/a and the maximum cumulative subsidence amount reaching -1 269 mm. The subsidence location and subsidence change trend monitored by SBAS-InSAR are consistent with leveling results, but in the area of severe settlement, the amount of settlement monitored by SBAS-InSAR is somewhat different from leveling data.

Settlement data is easily disturbed by external environmental noise in the process of safety monitoring of steel frame structure, so we propose a denoising method with improved wavelet threshold function based on the theory of complementary ensemble empirical mode decomposition(CEEMD). First, the noisy signal is decomposed by CEEMD to obtain various eigenmode functions(IMF), then we select the IMF components for the improved wavelet threshold to denoise, and finally we reconstruct the IMF information to obtain the denoised signal. The simulation results indicate that the hybrid method besed on CEEMD and the improved wavelet threshold method  can effectively filter the noise components in the signal and improve the accuracy of settlement data monitoring.

Using minute, 1 sps, 10 sps and 100 sps sampling rate observation data from a 4-component borehole strainmeter at Huangyuan station, we study the co-seismic variation of Maduo M7.4 earthquake in Qinghai by self-check analysis, co-seismic strain step analysis and spectrum analysis. The results show that the seismic wave information recorded by minute sampling is seriously lacking. If we use the minute value data to study the initial motion of seismic wave and the amplitude of co-seismic variation, we obtain a conclusion of low reliability. The higher the sampling rate, the stronger the ability to record strain seismic wave information, but the difference between 100 sps sampling and 10 sps sampling is not significant, and 10 sps sampling basically records more comprehensive strain seismic wave information. The variation property and amplitude of co-seismic strain steps are independent of sampling rate, and the minute sampling data can be used for related research. When the 4-component borehole strainmeter is set up in the future, the sampling rate should be increased to at least 1 sps.

We propose a seismic signal denoising method combining adaptive noise complete ensemble empirical mode decomposition(CEEMDAN) and wavelet transform(WT). First, using CEEMDAN, we  adaptively decompose the seismic signal into several intrinsic mode functions(IMFs) and margins, and calculate the Pearson correlation coefficient between each component and the original signal. We process the components in the threshold interval by wavelet filtering, maintaining the original state, and directly eliminating, and then we carry out linear reconstruction. Finally, to quantitatively evaluate the denoising effect, we construct the index system of sample entropy change, mutual information, and signal-to-noise ratio. The simulated experiment and measured data (Maduo earthquake in Qinghai) results show that, compared with EMD, EEMD and other methods, the CEEMDAN-WT method can effectively suppress the influence of random noise and improve the signal-to-noise ratio. The refined reconstruction effect of the seismic signal is better, and the effective components of the signal are largely preserved.

In this paper, we study the feasibility of support vector machine in small sample seismic signal recognition. The results show that with the increase of sample size, the recognition rate of this method increases first and then decreases. Using the seismic data of Shandong and Jiangsu in 2006-2017, only 30 training samples per class can achieve the correct recognition rate of about 85%. The improvement of recognition rate does not depend on the addition of a large number of samples. It is not only suitable for the study of seismic signal recognition in areas with few seismic data samples, but also provides a new idea for simplifying the sample library and reducing the operation cost.