The urban population is the social subject of the city and the most active factor in urban development. The urban management needs to understand the distribution of urban population in different areas of the city at different time points, which is collectively known as the spatio-temporal distribution of urban population. Accurate and refined urban population distribution data is of great significance to urban management and planning, urban economic development and people's life. Therefore, the estimation of urban population spatio-temporal distribution is one of the hot issues in urban geography. This paper reviews from the following key aspects:① The construction method of spatial distribution units; ② The main models or methods are summarized and described in detail from the perspective of the development process of model ideas and estimation objects; ③ The main application of estimation results in urban development. On this basis, this paper points out the existing problems in the construction of spatial units, modeling data, modeling ideas and results verification in the current research on the spatio-temporal distribution of population, and discusses future research directions.

Remote sensing image matching is the key foundation of remote sensing image processing, which has been a research hotspot for scholars at home and abroad. Due to the characteristics of multi-modal images such as radiation difference, geometric difference, scale difference, viewpoint difference and dimensional difference, a universal matching method with strong universality has not yet appeared. With the continuous development of remote sensing, artificial intelligence, big data, and other technologies and the continuous expansion of application fields, the image matching technology system is also developing and evolving. This paper summarizes the multi-modal remote sensing image matching classification system based on the development history of image matching technology, discusses the latest progress of multi-modal image matching technology from the perspective of feature-driven and data-driven, and points out the core difficulties and future development trend in order to promote the deeper development of multi-modal image matching research.

The non-conservative force (NCF) model is one of the main factors that restrict orbit accuracy of BDS-3 satellites. In this paper, several NCF models for BDS-3 MEO satellites were built by compensating typical empirical solar radiation pressure (SRP) models, i.e., the ECOM1 and ECOM2, with earth radiation pressure (ERP), antenna thrust (AT) and the box-wing SRP model. Experiments were carried out with globally distributed stations to compare different NCF models by analyzing the overlap orbit differences and SLR residuals. It's shown that the empirical SRP modes dominate the orbit precision and the ECOM2 is superior to ECOM1 under the nominal yaw steering (YS) mode, while the ECOM1 is less sensitive to different YS modes. The ERP and AT together cause a systematic deviation of about 3 cm in the radial direction of orbit. By modeling the ERP and AT, the SLR residuals bias of satellites C29 and C30 can be almost completely eliminated, but the bias of satellites C20 and C21 increases. In addition, introducing the box-wing model is also beneficial to improve the orbit accuracy.

The GNSS regional reference station network can provide fast and precise positioning services for PPP and RTK users over the wide range. However, for different technical systems, the error influencing factors and service modes are different, the data processing methods at the server and the performance at positioning terminal will also vary. After fixing non-difference ambiguity of the reference station, this paper presents an integrated service method of PPP and RTK based on the regional reference station network, and comprehensively evaluates the terminal positioning performance for these two technologies. Experimental analysis are conducted by using CORS station network data with a station spacing of about 100km in a northwestern province. Results shows the positioning accuracy of PPP fixed solution using the integer clock/UPD product solved by the regional reference station network are very high, and the positioning error in horizontal direction can be less than 0.5 cm. The regional modeled atmospheric can significantly increase the initialization speed of the positioning terminal. For PPP and RTK, a fixed solution can be obtained in a single epoch of 60.0% and 87.7%. It should be noted that the RTK positioning based on the VRS mode is equivalent to strong atmospheric constraints. When the accuracy of modeling atmospheric delay is poor, the positioning accuracy will be significantly reduced, while the PPP-RTK positioning accuracy using virtual atmospheric constraints is almost unaffected.

This study proposes the EMD-HD&WD algorithm to solve the limitations of screening rules and mode mixing in the denoising analysis of the GNSS coordinate time series when using empirical mode decomposition (EMD). The improved algorithm introduces the Hausdorff distance (HD) as a screening criterion for EMD and combines wavelet decomposition (WD). The reliability and universality of EMD-HD&WD algorithm were verified by 149 GNSS vertical time series in the crustal movement observation network of China. The results show that the HD is better than the existing screening criteria; the average correction rates of the EMD-HD&WD for the velocity uncertainty and the amplitude of flicker noise of GNSS stations are 88.4%. The algorithm can effectively identify the observational noise, reduce the mode aliasing of EMD, and improve the model accuracy of GNSS vertical coordinate time series.

In the simulation system for testing the polar performance of inertial navigation at middle and low latitudes, the IMU conversion error caused by reference error is an important factor affecting the simulation test accuracy. This paper discusses and analyzes its simplified calculation method. Firstly, based on the trajectory transfer rule with constant navigation parameters in the transverse coordinate system of the earth sphere model, the simulation test method and IMU conversion formula are introduced. Secondly, the calculation method of IMU conversion error is studied, and compared with the complete calculation formula, the approximate calculation formula suitable for marine is given, and the coefficient, the magnitude and variation form of each component are analyzed. Finally, using the navigation parameters of the actual voyage, the correctness of the formula is verified. The maximum calculation error of the simplified formula is about 10%, which can meet the requirements of IMU error analysis in the subsequent simulation test.

The gestation and occurrence of earthquakes are essentially the inevitable result of the gradual accumulation of stress and strain energy in the crust and their sudden or slow release. Studying the changing process of strain is of great significance to the determination of earthquake risk. Based on the strain time series of GNSS grids in Yunnan region from 2013 to 2019, the Hilbert-Huang transform (HHT) and ensemble empirical mode decomposition (EEMD) analysis method was used to explore the time-frequency characteristics of GNSS strain time series before earthquakes in Yunnan region which was a different method from traditional time-frequency analysis, it was specially suitable for nonlinear and non-stationary signals, and try to mine the seismogenic information carried in the strain time-frequency signal. The paper summarizes the earthquakes corresponding to grids No.23 and No.42 as earthquake examples. The results show that EEMD can decompose signal according to the time characteristic of the data, and can fully retain the characteristics of the data itself in the decomposition process. Its decomposition is objective and adaptive, and can better analyze the change characteristics of seismic signals at different scales. In addition, the Hilbert transform can describe the subtle changes of the signal over time, reflect the instantaneous characteristics of the signal, and has applicability for the abnormal identification. Through the method of EEMD, IMF component anomaly recognition, Hilbert transform method and comprehensive dynamic analysis of strain time series, it can find some potential information on the eve of earthquakes, and provide reference for the determination of dangerous locations of strong earthquakes in Yunnan area in the future.

To solve the problem that the accurate estimation of the zenith tropospheric delay (ZTD) obtained from the numerical weather model (NWM) depends on external benchmarks, a ZTD accuracy estimation model coupled with particle swarm algorithm and extended RBF neural network is constructed. The model sample is built using NWM's meteorological and terrain feature data. The target set is constructed using GNSS ZTD products as reference values. The model scale structure is determined by hierarchical clustering and fuzzy C-mean clustering, and the particle swarm algorithm optimizes the model parameters. The ERA5 pressure stratification product provided by the European Centre for Medium-Range Weather Forecasts (ECMWF) is used to train the model and verify the results for the NWM particular case. The results show that the model has good estimation accuracy and generalization capability. The average estimation accuracy is better than 4 mm and can achieve ZTD accuracy estimation at any location without relying on an external reference frame.

Significant wave height (SWH) is an important parameter to describe the sea state and plays an important role in the prediction of wave and ocean dynamics. However, traditional methods such as satellite radar altimeter and buoy are difficult to achieve high spatial-temporal resolution SWH estimation. As a novel remote sensing technology, spaceborne Global Navigation Satellite System-Reflectometry (GNSS-R) provides a new opportunity to estimate SWH. This paper presents a new method for retrieving SWH of spaceborne GNSS-R waves based on normalized integral delay waveform (NIDW). Firstly, the spaceborne GNSS-R delay Doppler maps (DDMs) are denoised and filtered in order to strictly control the quality of DDM data. Then, NIDW is extracted from DDM, and four GNSS-R observables are calculated based on NIDW (i.e. leading edge slope of NIDW, trailing edge slope of NIDW, leading edge waveform summation of NIDW and trailing edge waveform summation of NIDW). Subsequently, based on these four observables, an empirical SWH retrieval model of spaceborne GNSS-R is established. Finally, ERA5 and AVISO SWH data products are used as validation data respectively, and the SWH estimation results of the retrieval model are compared and analyzed with ERA5 and AVISO SWH data products. The experimental results show that when the EAR5 SWH data is used as the verification data, the root mean square error (RMSE) and correlation coefficient (CC) of the four observables are better than 0.66 m and 0.65, respectively. When the AVISO SWH data is used as the verification data, the RMSE and CC of the four observables are better than 0.68 m and 0.70, respectively. It further shows that the modeling method proposed in this paper is feasible and reliable in spaceborne GNSS-R SWH estimation.

Model-errors and inability of adapting to different forest scenes appear when the SINC function model simplified from RVoG model is applied to retrieve forest height with TanDEM-X interferometric SAR data. To solve the problem, a method of forest height inversion based on TanDEM-X InSAR coherence with the aid of a subset of LiDAR data is proposed, which is independent on global high-precision DEM products to provide ground phase information and uses coherence to obtain forest height only. In order to verify the feasibility of the method, TanDEM-X InSAR acquisitions obtained in two typical study sites in Spain are experimented, while LiDAR data is used for validation. The results show forest height is obtainable based on the single-baseline TanDEM-X InSAR coherence, whose accuracy are 2.34 m and 1.74 m respectively.

This paper proposes a spaceborne GNSS-R soil moisture retrieval method based on CYGNSS data. Firstly, the theoretical model of soil moisture retrieval is constructed by combining the surface reflectance parameters extracted from CYGNSS data and the auxiliary information of vegetation optical depth, surface roughness and temperature extracted from SMAP data. The fine mathematical model of soil moisture retrieval is determined by using the neural network model. Then, the soil moisture obtained by the proposed model is processed at an interval of 0.35, and the stage model proposed in this paper is used to improve the soil moisture retrieval accuracy, and the spaceborne GNSS-R soil moisture is obtained globally by using the CYGNSS data from October 2018 to May 2019. Finally, the effectiveness of the spaceborne GNSS-R soil moisture retrieval method proposed in this paper is evaluated through comparing with the soil moisture data provided by SMAP, and the time series of spaceborne GNSS-R soil moisture is analyzed. The results show that the soil moisture obtained by the method proposed in this paper is in good agreement with the soil moisture obtained by SMAP, and the trend of variation with time is also consistent with the actual situation, which provides a new idea for high-precision soil moisture retrieval.

Aiming at the problem that the current multi-beam and side-scan sonar image registration methods fail to take into account the details of image deformation, the scale differences and exist local texture distortion, a iterative adaptive registration method combining wavelet transform, affine transform and Demons registration algorithm was proposed. By using wavelet transformation to extract the low frequency information from side scan sonar image and reconstruction images, using affine transformation and Demons algorithm to reconstruct image with iterative adaptive multi-beam image registration, obtain registration transformation model. The model is used to transform the original side-scan sonar image to obtain the side-scan sonar image constrained by the geographic coordinates of the multi-beam image. The results of example verification show that this method can effectively realize the registration of multi-beam and side-scan sonar images, and obtain the fusion sonar images with accurate position and rich texture.

The existing line-element-complexity-measurement methods are usually to calculate the overall complexity of a single line element, which cannot consider the differences of line element in different regions. Therefore, we present a complexity measurement method of line element on nautical chart considering local difference. Firstly, the correlation between the length of edge in triangle network and the complexity of the line elements is analyzed. Then, the complex field of line elements on nautical chart is obtained by the steps of constructing constrained Delaunay triangulation, calculating the data point complexity of line elements and interpolating. Finally, we choose two groups of typical nautical charts as experimental data, and verify the rationality of this method by several statistical experiments. The experimental results show that:the proposed method, considering local differences, can define the complexity of line elements on nautical chart better, which also can accurately and quantitatively evaluate the complexity of line elements in each region.

Shape recognition is one of the important contents of map spatial cognition, and neural network combined with spatial cognitive experiment and its effective shape feature vector extraction are effective ways to improve shape recognition. In this paper, a neural network building-polygon shape recognizer is constructed, which integrates the Fourier descriptors of macro shape parameters such as roundness, eccentricity and rectangularity as shape feature vectors. Firstly, the Fourier shape descriptors, circularity, eccentricity and rectangularity parameters of building polygons are extracted by Fourier transform and computational geometry methods, and the shape feature vectors are formed. Then, the neural network recognizer matching between building polygon and shape template is realized through the training of sample data. The results show that this method greatly improves the accuracy (98.7%) compared with the previous methods, and the feature extraction algorithm is not limited by the inconsistency of polygon points. The shape recognition of real building data in Wuhan and Zhengzhou is carried out, and its information entropy is calculated. This method has good recognition effect.