Temporal extrapolation of TEC using WNN during 2007–2018 and comparison against GIM,IRI2016 and Kriging

In this paper, a new method of temporal extrapolation of the ionosphere total electron content (TEC) is proposed. Using 3-layer wavelet neural networks (WNNs) and particle swarm optimization (PSO) training algorithm, TEC time series are modeled. The TEC temporal variations for next times are extrapolated with the help of training model. To evaluate the proposed model, observations of Tehran GNSS station (35.69°N, 51.33°E) from 2007 to 2018 are used. The efficiency of the proposed model has been evaluated in both low and high solar activity periods. All observations of the 2015 and 2018 have been removed from the training step to test the proposed model. On the other hand, observations of these 2 years are not used in network training. According to the F10.7, the 2015 has high solar activity and the 2018 has quiet conditions. The results of the proposed model are compared with the global ionosphere maps (GIMs) as a traditional ionosphere model, international reference ionosphere 2016 (IRI2016), Kriging and artificial neural network (ANN) models. The root mean square error (RMSE), bias, dVTEC = |VTEC_GPS ? VTEC_Model| and correlation coefficient are used to assess the accuracy of the proposed method. Also, for more accurate evaluation, a single-frequency precise point positioning (PPP) approach is used. According to the results of 2015, the maximum values of the RMSE for the WNN, ANN, Kriging, GIM and IRI2016 models are 5.49, 6.02, 6.34, 6.19 and 13.60 TECU, respectively. Also, the maximum values of the RMSE at 2018 for the WNN, ANN, Kriging, GIM and IRI2016 models are 2.47, 2.49, 2.50, 4.36 and 6.01 TECU, respectively. Comparing the results of the bias and correlation coefficient shows the higher accuracy of the proposed model in quiet and severe solar activity periods. The PPP analysis with the WNN model also shows an improvement of 1 to 12 mm in coordinate components. The results of the analyzes of this paper show that the WNN is a reliable, accurate and fast model for predicting the behavior of the ionosphere in different solar conditions.     

A multilayer perceptron and multiclass support vector machine based high accuracy technique for daily rainfall estimation from MSG SEVIRI data

The current paper introduces a new multilayer perceptron (MLP) and support vector machine (SVM) based approach to improve daily rainfall estimation from the Meteosat Second Generation (MSG) data. In this study, the precipitation is first detected and classified into convective and stratiform rain by two MLP models, and then four multi-class SVM algorithms were used for daily rainfall estimation. Relevant spectral and textural input features of the developed algorithms were derived from the spectral MSG SEVIRI radiometer channels. The models were trained using radar rainfall data set colected over north Algeria. Validation of the proposed daily rainfall estimation technique was performed by rain gauge network data set recorded over north Algeria. Thus, several statistical scores were calculated, such as correlation coefficient (r), root mean square error (RMSE), mean error (Bias), and mean absolute error (MAE). The findings given by: (r = 0.97, bias = 0.31 mm, RMSE = 2.20 mm and MAE = 1.07 mm), showed a quite satisfactory relationship between the estimation and the respective observed daily precipitation. Moreover, the comparison of the results with those of two advanced techniques based on random forests (RF) and weighted ‘k’ nearest neighbor (WkNN) showed higher accuracy obtained by the proposed model.     

A new global F2 peak electron density model for the International Reference Ionosphere (IRI)

A new neural network (NN) based global empirical model for the F2 peak electron density (NmF2) has been developed using extended temporal and spatial geophysical relevant inputs. Measured ground based ionosonde data, from 84 global stations, spanning the period 1995 to 2005 and, for a few stations from 1976 to 1986, obtained from various resources of the World Data Centre (WDC) archives (Space Physics Interactive Data Resource SPIDR, the Digital Ionogram Database, DIDBase, and IPS Radio and Space Services) have been used for training a NN. The training data set includes all periods of quiet and disturbed magnetic activity. A comprehensive comparison for all conditions (e.g., magnetic storms, levels of solar activity, season, different regions of latitudes, etc.) between foF2 value predictions using the NN based model and International Reference Ionosphere (IRI) model (including both the International Union of Radio Science (URSI) and International Radio Consultative Committee (CCIR) coefficients) with observed values was investigated. The root-mean-square (RMS) error differences for a few selected stations are presented in this paper. The results of the foF2 NN model presented in this work successfully demonstrate that this new model can be used as a replacement option for the URSI and CCIR maps within the IRI model for the purpose of F2 peak electron density predictions.     

A comparative analysis of SLR,MLR, ANN,XGBoost and CNN for crop height estimation of sunflower using Sentinel-1 and Sentinel-2

Sustainable monitoring and determining the biophysical characteristics of crops is of global importance due to the increase in demand for food. In this context, remote sensing data provide valuable information on crops. This study investigates the relationship between the variables determined from both Synthetic Aperture Radar (SAR) and optical images and crop height. For this purpose, backscatter (σVH, σVV, σVH / σVV) and coherence (?VH, ?VV) of multi-temporal dual-polarized Sentinel-1 and vegetation indices of multi-temporal Sentinel-2 data are analyzed. Two indices, namely, Normalized Difference Vegetation Index (NDVI) and NDVI with the red-edge band (NDVIred), are interpreted to identify the contribution of the red-edge band over the near-infrared band. The Zile District of Tokat province in Turkey where dominantly sunflower cultivation is carried out, was selected as the study area. In the analysis of the data, Simple Linear Regression (SLR), Multiple Linear Regression (MLR), Artificial Neural Network (ANN), EXtreme Gradient Boosting (XGBoost), and Convolutional Neural Network (CNN) were used. In the results of the study, ANN showed the lowest RMSE = 3.083 cm (RMSE%= 11.284) in the stem elongation period. The CNN followed the lowest RMSE for the Inflorescence development and flowering stages 19.223 cm (RMSE%=15.458) and 8.731 cm (RMSE%=5.821), respectively. In the ripening period, XGBoost achieved the lowest RMSE = 8.731 cm (RMSE%=6.091). All the best models in four methods were created using common variables of σVH, σVV, ?VH, ?VV and NDVIred, except ANN which exclude coherence variables. The results concluded that NDVIred contributed more than NDVI which is widely interpreted in previous studies.     

Cross testing of ionosphere models IRI-2001 and IRI-2007, data from satellite altimeters (Topex/Poseidon and Jason-1) and global ionosphere maps

In this study, we use a great body of statistical data covering the entire 23rd solar cycle to cross test data of satellite altimeters, Global Ionosphere Maps and the International Reference Ionosphere models, IRI-2001 and IRI-2007. It is revealed that experimental TEC values of the satellite altimeters regularly exceed the model ones by ∼3 TECU (1 TECU = 10¹⁶ m⁻²). The best possible value of difference between TECs obtained from altimeter and GIM-map data significantly differs for different laboratories: the maximum for CODG data falls on 2.5 TECU, ESAG – 3 TECU, JPLG – 0 TECU, UPCG – 2 TECU. The dependence of experimental and model data root-mean-square deviation on the F10.7 index is shown to be nearly linear. IRI-2001 and IRI-2007 relative errors are characterized by considerable 11-year and annual variations. Given the geomagnetic planetary index K_p under 7, IRI-2001 and IRI-2007 reproduce TEC in the ionosphere with an accuracy of ∼30% relative to measurement data from satellite altimeters. The amplitude of absolute error variations resulting from the difference in ionization enhancement between the model and the real ionosphere during the morning solar terminator transit is ∼5 TECU.     

Ionospheric electron density over Resolute Bay according to E-CHAIM model and RISR radar measurements

In this study, predictions of the E-CHAIM ionospheric model are compared with measurements by the incoherent scatter radars RISR at Resolute Bay, Canada, in the northern polar cap. Reasonable coverage was available for all seasons except winter for which no conclusions were drawn. It is shown that ratios of the model-to measured electron densities are close to unity in the central part of the F layer, around its peak. This is particularly evident for summer daytime. Distributions of the ratios are wider for other seasons indicating larger number of cases when the model underestimates or overestimates. E-CHAIM underestimates the electron density at ionospheric topside and bottomside by ～ 10–20 %. At the bottomside, the underestimations are strongest in summer and equinoctial nighttime. At the topside, the underestimations are strongest in autumn nighttime. Model overestimations are noticeable in the middle part of the F layer during dawn hours in autumn. Overall, the model tends to not predict highest-observed peak electron densities and the largest-observed heights of the peak.     

The effect of total solar eclipse of October 3, 2005, on the total electron content over Europe

GPS observations from EUREF permanent GPS network were used to observe the response of TEC (Total Electron Content) to the total solar eclipse on October 3, 2005, under quiet geomagnetic conditions of the daytime ionosphere. The effect of the eclipse was detected in diurnal variations and more distinctly in the variations of TEC along individual satellite passes. The trough-like variations with a gradual decrease and followed by an increase of TEC at the time of the eclipse were observed over a large region. The depression of TEC amounted to 3–4 TECU. The maximum depression was observed over all stations located at the maximum path of the solar eclipse. The delay of a minimum level of TEC with respect to the maximum phase of the eclipse was about 20–30 min.     

Towards a GPS-based TEC prediction model for Southern Africa with feed forward networks

In this paper, first results from a national Global Positioning System (GPS) based total electron content (TEC) prediction model over South Africa are presented. Data for 10 GPS receiver stations distributed through out the country were used to train a feed forward neural network (NN) over an interval of at most five years. In the NN training, validating and testing processes, five factors which are well known to influence TEC variability namely diurnal variation, seasonal variation, magnetic activity, solar activity and the geographic position of the GPS receivers were included in the NN model. The database consisted of 1-min data and therefore the NN model developed can be used to forecast TEC values 1 min in advance. Results from the NN national model (NM) were compared with hourly TEC values generated by the earlier developed NN single station models (SSMs) at Sutherland (32.38°S, 20.81°E) and Springbok (29.67°S, 17.88°E), to predict TEC variations over the Cape Town (33.95°S, 18.47°E) and Upington (28.41°S, 21.26°E) stations, respectively, during equinoxes and solstices. This revealed that, on average, the NM led to an improvement in TEC prediction accuracy compared to the SSMs for the considered testing periods.     

Mitigation of receiver biases in ionospheric observables from PPP with ambiguity resolution

PPP (Precise Point Positioning) is a GNSS (Global Navigation Satellite Systems) positioning method that requires SSR (State Space Representation) corrections in order to provide solutions with an accuracy of centimetric level. The so-called RT-PPP (Real-time PPP) is possible thanks to real-time precise SSR products, for orbits and clocks, provided by IGS (International GNSS Service) and its associate analysis centers such as CNES (Centre National d'Etudes Spatiales). CNES SSR products also enable RT-PPP with integer ambiguity resolution. In GNSS related literature, PPP with ambiguity resolution (PPP-AR) in real-time is often referred as PPP-RTK (PPP – Real Time Kinematic). PPP-WIZARD (PPP - With Integer and Zero-difference Ambiguity Resolution Demonstrator) is a software that is made available by CNES. This software is capable of performing PPP-RTK. It estimates slant ionospheric delays and other GNSS positioning parameters. Since ionospheric effects are spatially correlated by GNSS data from active networks, it is possible to model and provide ionospheric delays for any position in the network coverage area. The prior knowledge ionospheric delays can reduce positioning convergence for PPP-RTK users. Real-time ionospheric models could benefit from highly precise ionospheric delays estimated in PPP-AR. In this study, we demonstrate that ionospheric delays obtained throughout PPP-AR estimation are actu ally ionospheric observables. Ionospheric observables are biased by an order of few meters caused by the receiver hardware biases. These biases prohibit the use of PPP-WIZARD ionospheric delays to produce ionospheric models. Receiver biases correction is essential to provide ionospheric delays while using PPP-AR based ionospheric observables. In this contribution, a method was implemented to estimate and mitigate receiver hardware biases influence on slant ionospheric observables from PPP-AR. In order to assess the proposed approach, PPP-AR data from 12 GNSS stations were processed over a two-month period (March and April 2018). A comparison between IGS ionospheric products and PPP-AR based ionospheric observables corrected for receiver biases, resulted in a mean of differences of −39 cm and 51 cm standard deviation. The results are consistent with the accuracy of the IGS ionospheric products, 2–8 TECU, considering that 1 TECU is ~16 cm in L1. In another analysis, a comparison of ionospheric delays from 5 pairs of short baselines GNSS stations found an agreement of 0.001 m in mean differences with 22 cm standard deviation after receiver biases were corrected. Therefore, the proposed solution is promising and could produce high quality (1–2 TECU) slant ionospheric delays. This product can be used in a large variety of modeling approaches, since ionospheric delays after correction are unbiased. These results indicate that the proposed strategy is promising, and could benefit applications that require accuracy of 1–2 TECU (~16–32 cm in L1).     

基于神经网络的磁平静期赤道电集流预测

利用BP神经网络技术分别对2008年后磁平静期印度扇区、秘鲁扇区以及CHAMP卫星的赤道电集流(EEJ)变化进行预测,其中神经网络训练数据为对应的2000-2007年磁平静期EEJ观测数据,输入参量为天数、地方时、太阳天顶角、太阳活动指数(F10.7)、太阴时以及卫星地理经度,输出参量为EEJ.对EEJ预测结果进行了统...     

Prediction of geosynchronous electron fluxes using an artificial neural network driven by solar wind parameters

There are hundreds of satellites operating at the geosynchronous (GEO) orbit where relativistic electrons can cause severe damage. Thus, predicting relativistic electron fluxes is significant for spacecraft safety. In this study, using GOES satellite data during 2011–2020, we propose two neural network models with two hidden layers to predict geosynchronous relativistic electron fluxes at two energy channels (>0.8 MeV and > 2 MeV). The number of input neurons of the two channels (>0.8 MeV and > 2 MeV) are determined to be 36 and 44, respectively. The > 0.8 MeV model has 22 and 9 neurons in the hidden layers, while the > 2 MeV model has 25 and 15 neurons in the hidden layers. The input parameters include the north–south component of the interplanetary magnetic field, solar wind speed, solar wind dynamic pressure and solar wind proton density. Through the analysis of different time delays, we determine that the optimal time delays of two energy channels (>0.8 MeV and > 2 MeV) are 8 days and 10 days, respectively. The training set and validation set (Jan 2011-Dec 2018) are divided by the 10-fold cross-validation method, and the remaining data (Jan 2019-Feb 2020) is used to analyze the model performance as a test set. The prediction results of both energy channels show good agreement with satellite observations indicated by low RMSE (～0.3 cm^-2sr^-1s^?1), high PE (～0.8) and CC (～0.9). These results suggest that only using solar wind parameters is capable of obtaining reasonable predictions of geosynchronous relativistic electron fluxes.     

BP-AdaBoost模型在光纤陀螺零偏温度补偿中的应用

针对光纤陀螺零偏漂移随温度呈复杂的非线性变化,建立了BP-AdaBoost（Back Propagation neural network,Adaptive Boosting）模型对零偏进行补偿,改善了光纤陀螺的零偏稳定性能.同时,研究了模型参数对预测精度的影响,给出了BP神经网络隐含层神经元个数的选择以及AdaBoost模型迭代次数的确定方法.运用AdaBoost算法提升单个BP神经网络的预测能力,提高了集成模型整体的预测精度.对采集的光纤陀螺输出实测数据进行了事后仿真,结果表明,BP-AdaBoost模型相比传统的线性回归模型、混合线性回归模型、单个BP神经网络模型的补偿效果更显著,验证了该模型的有效性,具有重大的工程应用参考价值.      

A neural network-based ionospheric model for Arecibo

The Arecibo Observatory (18°N, 66°W) has the world’s largest single dish antenna (300 m diameter). Beyond radio astronomy it can also operate as an incoherent scatter radar and in that mode its figure-of-merit makes it also one of the most powerful world-wide. For the present purpose all electron density data available on the web, from the beginning with the first erratic measurements in 1966 up to 2004 inclusive, were downloaded. The measurements range from about 100 km to beyond 700 km and are essentially evenly distributed, i.e. not dedicated to measure specific geophysical events. From manually edited/inspected data a neural network (NN) was established with season, hour of the day, solar activity and Kp as the input parameters. The performance of this model is checked against a – likewise NN based – global model of foF2, a measure of the maximum electron density of the ionosphere. Considering the diverse data sources and assumptions of the two models it can be concluded that they agree remarkably well.     

Estimation of GPS instrumental biases from small scale network

With 4 GPS receivers located in the equatorial anomaly region in southeast China, this paper proposes a grid-based algorithm to determine the GPS satellites and receivers biases, and at the same time to derive the total electron content (TEC) with time resolution of 15 min and spatial resolution of 1° by 3.5° in latitude and longitude. By assuming that the TEC is identical at any point within a given grid block and the biases do not vary within a day, the algorithm arranges unknown biases and TECs with slant path TEC from the 4 receivers’ observations into a set of equations. Then the instrumental biases and the TECs are determined by using the least squares fitting technique. The performance of the method is examined by applying it to the GPS receiver chain observations selected from 16 geomagnetically quiet days in four seasons of 2006. It is found that the fitting agrees with the data very well, with goodness of fit ranging from 0.452 TECU to 1.914 TECU. Having a mean of 0.9 ns, the standard deviations for most of the GPS satellite biases are less than 1.0 ns for the 16 days. The GPS receiver biases are more stable than that of the GPS satellites. The standard deviation in the 4 receiver bias is from 0.370 ns to 0.855 ns, with a mean of 0.5 ns. Moreover, the instrumental biases are highly correlated with those derived from CODE and JPL with independent methods. The typical precision of the derived TEC is 5 TECU by a conservative estimation. These results indicate that the proposed algorithm is valid and qualified for small scale GPS network.     

Validation of University of New Brunswick Ionospheric Modeling Technique with ionosonde TEC estimation over South Africa

For more than a decade, ionospheric research over South Africa has been carried out using data from ionosondes geographically located at Madimbo (28.38°S, 30.88°E), Grahamstown (33.32°S, 26.50°E), and Louisvale (28.51°S, 21.24°E). The objective has been modelling the bottomside ionospheric characteristics using neural networks. The use of Global Navigation Satellite System (GNSS) data is described as a new technique to monitor the dynamics and variations of the ionosphere over South Africa, with possible future application in high frequency radio communication. For this task, the University of New Brunswick Ionospheric Modelling Technique (UNB-IMT) was applied to compute midday (10:00 UT) GNSS-derived total electron content (GTEC). GTEC values were computed using GNSS data for stations located near ionosondes for the years 2002 and 2005 near solar maximum and minimum, respectively. The GTEC was compared with the midday ionosonde-derived TEC (ITEC) measurements to validate the UNB-IMT results. It was found that the variation trends of GTEC and ITEC over all stations are in good agreement and show a pronounced seasonal variation for the period near solar maximum, with maximum values (∼80 TECU) around autumn and spring equinoxes, and minimum values (∼22 TECU) around winter and summer. Furthermore, the residual ΔTEC = GTEC − ITEC was computed. It was evident that ΔTEC, which is believed to correspond to plasmaspheric electron content, showed a pronounced seasonal variation with maximum values (∼20 TECU) around equinoxes and minimum (∼5 TECU) around winter near solar maximum. The equivalent ionospheric and total slab thicknesses were also computed and comprehensively discussed. The results verified the use of UNB-IMT as one of the tools for future ionospheric TEC research over South Africa.     

ESA UGI (Unified-GNSS-Ionosphere): An open-source software to compute precise ionosphere estimates

Ionospheric estimation is becoming more and more important for the new multifrequency positioning algorithms, since they can help to improve greatly the convergence time for acquiring a good positioning error. In this paper, an open source tool to estimate precise ionospheric estimates is presented, namely ESA UGI (Unified-GNSS-Ionosphere). The presentation is done jointly with a methodology to test ionospheric model using a modified NeQuick to generate synthetic data. The results with different option of the ESA UGI shows that it has a good performance below 1 TECU (Total Electron Content Units) in vTEC (vertical Total Electron Content) RMS (Root Mean Squared) for European networks, around 2 TECU in a well-covered African region and between 1 and 6 TECU globally with this synthetic data. It shows as well the capability of changing between different ionosphere models (voxel, multilayer and spherical harmonics) and configuration options. Finally, a test with uncombined PPP actual data is presented showing that instantaneous convergence below 30 cm in 3D RMS position error are achievable in a well sounded area in Europe.     

基于神经网络的自由漂浮空间机械臂 自适应鲁棒控制

针对自由漂浮空间机械臂动力学模型难以精确获得,且无法表达为关于未知参数的线性形式问题,提出基于自适应神经网络的鲁棒控制方法.对于不确定性空间机械臂系统模型中存在的未知不确定部分,利用神经网络的万能逼近特性,设计神经网络控制器来补偿未知模型,避免传统控制中的保守上界估计;采用泰勒线性化技术将神经网络隐含层中的高斯函数线性化,设计包括网络权值、高斯中心及宽度在内的网络全参数自适应学习律,实现在线实时调整,提高控制精度;设计鲁棒自适应控制器来抑制外界扰动,并补偿逼近误差,提高系统鲁棒性;基于Lyapunov理论证明闭环系统的一致最终有界（UUB）.仿真试验表明所提控制方法能够获得较好控制效果,对空间机械臂控制具有一定工程应用价值.     

Precise orbit determination and baseline consistency assessment for Swarm constellation

Precise orbit determination (POD) and precise baseline determination (PBD) of Swarm satellites with 4 years of data are investigated. Ambiguity resolution (AR) plays a crucial role in achieving the best orbit accuracy. Swarm POD and PBD based on single difference (SD) AR and traditional double difference (DD) AR methods are explored separately. Swarm antenna phase center variation (PCV) corrections are developed to further improve the orbit determination accuracy. The code multipath of C1C, C1W and C2W observations is first evaluated and clear variations in code noise related to different receiver settings are observed. Carrier phase residuals of different time periods and different loop tracking settings of receiver are studied to explore the effect of ionospheric scintillation on POD. The reduction of residuals in the polar and geomagnetic equator regions confirms the positive impact of the updated carrier tracking loops (TLs) on POD performance. The SD AR orbits and orbits with float ambiguity (FA) are compared with the Swarm precise science orbits (PSOs). An average improvement of 27 %, 4 % and 16 % is gained in along-track, cross-track and radial directions by fixing the ambiguity to integer. For Swarm-A/B and Swarm-B/C formations, specific days are selected to perform the DD AR-based POD during which the average distance of the formation satellites is less than 5000 km. Satellite laser ranging (SLR) observations are employed to validate the performance of FA, SD AR and DD AR orbits. The consistency between the SD AR orbits and SLR data is at a level of 10 mm which shows an improvement of 25 % when comparing with the FA results. An SLR residuals reduction of 15 % is also achieved by the DD AR solution for the selected days. Precise relative navigation is also an essential aspect for spacecraft formation flying missions. The closure error method is proposed to evaluate the baseline precision in three dimensions. A baseline precision of 1–3 mm for Swarm-A/C formation and 3–5 mm for Swarm-A/B and Swarm-B/C satellite pairs is verified by both the consistency check and closure error method.     

A simple model to predict solar radiation under clear sky conditions

Solar radiation is one of the major factors that dominate the thermal behaviors of aerostats in the daytime and the primary energy source of high altitude long endurance aerostats. Therefore, it is necessary to propose an accurate model to predict the solar irradiances. A comprehensive review of the well-known solar radiation models is conducted to help develop the new model. Based on the analysis of the existing models and the available radiation data, the extensive computer tests of the regression and optimization are conducted, from which the new solar radiation model for direct and diffuse irradiances under clear sky conditions is proposed. The new model has excellent prediction accuracy. The coefficient of determination for direct radiation is 0.992, with the root mean square error (RMSE) of 16.9 W/m² and the mean absolute error (MAE) of 2.2%. The coefficient of determination for diffuse radiation is 0.86, with RMSE = 8.7 W/m² and MAE = 9.9%. Comparisons with the well-known existing models show that the new model is much more accurate than the best existing ones.     

基于BP神经网络的飞行绩效评价模型

为实现飞行绩效的客观评价,利用眼动数据,建立了拓扑结构为6-14-3型的BP(Back Propagation)神经网络模型.已有的实验数据和随机插值法得到的数据作为建模的数据来源,数据分为训练集和测试集并归一化.基于Matlab的神经网络工具箱,利用经验公式和实验比较法确定了BP网络模型的隐含层节点数;对BP算法的各种改进算法进行了优化选择;将训练集数据和测试集数据先后输入网络模型进行学习训练和仿真测试;对3个技术水平的飞行员的飞行绩效进行了预测和评价.研究表明,眼动数据的BP神经网络模型可以较为准确地评价飞行绩效,评价方法可以为飞行训练提供参考.