Spatiotemporal Modeling And Analysis Of Ionospheric Electron Content Based On GNSS

With the continuous progress and development of earth observation technology,ionospheric soundings have also changed from the single station and small-scale measurement of ground-based sounding technologies to the long-term continuous observation covering the whole earth with the com-bination of ground-based and space-based sounding technologies.The emergence of new ionospheric sounding technologies promote the research and development of ionospheric related scientific theory,reveal the physical formation mechanism of ionospheric phenomenon,and deepen the understanding of the earth's space environment,which has the practical value and important scientific research sig-nificance.The temporal and spatial variations of electron ionospheric electron content are interrelated in time dimension and space dimension.In this thesis,based on the data from traditional ionospheric sounding and GNSS observation,There models,an ionospheric tomography model,a prediction model of ionospheric TEC,and an accurate ionospheric anomalies detection model,are proposed in space-time dimension to solve the problems in the research of computerized ionospheric tomography,ionospheric TEC prediction and ionospheric anomalies detection respectively.The major contribu-tions of this study are summarized as follows:(1)Aiming at the complicated solution process and the non-unique solution to projection matrix generation algorithm of the traditional pixel-based ionosphere tomographic.This thesis proposes an improved ionospheric tomographic projection matrix generation algorithm based on the geometric vector method.This algorithm makes use of the geometric relationships that exist between vec-tors.It simplifies the solution of the ternary quadratic equation in the traditional method into a one-dimensional equation,reduces the number of geometric parameters,decreases the computational complexity,and improves the calculation efficiency.(2)Aiming at the problems of pixel-based ionospheric tomography,an ionospheric tomography algorithm based on Chapman function and spherical harmonic function was proposed.This algorithm proposes a continuous integrable approximation Chapman function to fit the ionospheric TEC in the vertical direction.A spherical harmonic function is used to fit the Nm F2 and hm F2 of the Chapman functions in the horizontal direction.The new method can not only reflect the spatial distribution of ionospheric electrons,but also obtain the horizontal distribution of ionospheric Nm F2 and hm F2,which expands the application of GNSS in ionospheric research.Simulation experiments and actual experiments confirm the existing problems in the pixel-based ionospheric tomography algorithm,and verify the reliability and effectiveness of the new algorithm.The new algorithm has obvious advantages over traditional algorithm.used ionospheric Ionosonde in the experiment of measured data,the proposed algorithm offers an improvement of 27%in 100km~200km,17%over 300km,compared with the traditional pixel-based ionosphere tomographic algorithms,and the coefficient of determination R~2=0.986.(3)Aiming at the problems in traditional ionospheric TEC prediction methods,a new method for global prediction of ionospheric TEC based on recurrent neural network was proposed.This method utilizes the advantages of convolutional long short term memory(CLSTM)network to pro-cess ionospheric map sequences,which could makes long-term continuous predictions of global iono-spheric TEC.The new method makes full use of the effective information of ionospheric TEC in both space and time.The input sample data and the output sample data are both the ionospheric map se-quences.Besides,the original ionospheric map sequences is also subjected to first-difference and second-difference operation in the time dimension as the other two input sample data.Experiments show that the modeling sample duration will affect the accuracy of the prediction model.When the modeling sample duration exceeds a certain length,the accuracy of the new method prediction model will not be affected by the prediction duration.The prediction results of the new method can describe the distribution of the ionospheric TEC in more detail.The new method are highly sensitive to ex-ternal factors,and external factors will affect its prediction accuracy.Compared with CODE 1-Day Predicted GIM data,the proposed method offers an average improvement of 22.1%,the maximum improved up to 44.6%.(4)Aiming at the influence of uncertainty in the traditional ionospheric TEC anomalies detection,a new detecting ionospheric TEC anomalies method based on the CLSTM network to ionospheric TEC prediction is proposed.Comparing with the sliding window method,a traditional method,the new ionospheric anomalies detection method can detect ionospheric anomalies more accurately and reliablyin the static geomagnetic environment and the complex space environment before the 2016Kumamoto earthquake in Japan.By analyzing the coupling of the 2016 Kumamoto earthquake and ionosphere,it is found that the persistent ionospheric anomalies may be related to the occurrence of the earthquake in the epicenter area on the day before and on the day of the earthquake.