Forecasting space weather: Predicting interplanetary shocks using neural networks

We are developing a system to predict the arrival of interplanetary (IP) shocks at the Earth. These events are routinely detected by the Electron, Proton, and Alpha Monitor (EPAM) instrument aboard NASA’s ACE spacecraft, which is positioned at Lagrange Point 1 (L1). In this work, we use historical EPAM data to train an IP shock forecasting algorithm. Our approach centers on the observation that these shocks are often preceded by identifiable signatures in the energetic particle intensity data. Using EPAM data, we trained an artificial neural network to predict the time remaining until the shock arrival. After training this algorithm on 37 events, it was able to forecast the arrival time for 19 previously unseen events. The average uncertainty in the prediction 24 h in advance was 8.9 h, while the uncertainty improved to 4.6 h when the event was 12 h away. This system is accessible online, where it provides predictions of shock arrival times using real-time EPAM data.     

Using neural networks to determine the optimum solar input for the prediction of ionospheric parameters

Neural networks (NNs) are proving to be ideal tools for modeling the behaviour of the ionosphere. The NNs are trained using a database of archived data describing the relationship between the output parameter and an input space. The input space is designed from knowledge of those variables that affect the behaviour of the output parameter. For ionospheric parameters this input space would always include a solar variable due to the strong influence that the sun has on ionospheric behaviour.     

PC-index fluctuations and intermittency of the magnetospheric dynamics

One minute resolution Polar Cap (PC) index was used for the analysis of magnetospheric dynamics. The 1995–2000 time series analysis revealed that the power spectrum of the PC-index fluctuations is a power law in a wide range of frequencies. However, the obtained exponents differ for low and high frequency regions. The probability distribution functions of the PC-index fluctuations show a strong non-gaussian shape, depending on the time of increment. This indicates that the PC-index exhibits intermittency, previously detected in solar wind and auroral electrojet index fluctuations. The PC-index probability distribution functions were fitted by the functional form proposed by Castaing et al. [Velocity probability density functions of high Reynolds number turbulence. Physica D. 46, 177–200, 1990] to describe intermittency phenomena in ordinary turbulent fluid flows. The agreement between the fitting parameters obtained for the PC index and those reported before for solar wind magnetic field fluctuations is within 30%; which is noticeably less than the difference between the same parameters of solar wind and the AE-index fluctuations. This fact indicates that the PC index reflects the solar wind influence on the high-latitude magnetosphere, especially during the summer.     

Storm-time atmospheric density modeling using neural networks and its application in orbit propagation

Upper atmospheric densities during geomagnetic storms are usually poorly estimated due to a lack of clear understanding of coupling mechanisms between the thermosphere and magnetosphere. Consequently, the orbit determination and propagation for low-Earth-orbit objects during geomagnetic storms have large uncertainties. Artificial neural networks are often used to identify nonlinear systems in the absence of rigorous theory. In the present study, an attempt has been made to model the storm-time atmospheric density using neural networks. Considering the debate over the representative of geomagnetic storm effect, i.e. the geomagnetic indices ap and Dst, three neural network models (NNM) are developed with ap, Dst and a combination of ap and Dst respectively. The density data used for training the NNMs are derived from the measurements of the satellites CHAMP and GRACE. The NNMs are evaluated by looking at: (a) the mean residuals and the standard deviations with respect to the density data that are not used in training process, and (b) the accuracy of reconstructing the orbits of selected objects during storms employing each model. This empirical modeling technique and the comparisons with the models NRLMSIS-00 and Jacchia-Bowman 2008 reveal (1) the capability of neural networks to model the relationship between solar and geomagnetic activities, and density variations; and (2) the merits and demerits of ap and Dst when it comes to characterizing density variations during storms.     

Forecasting the geomagnetic activity of the Dst index using multiscale radial basis function networks

The Dst index is a key parameter which characterises the disturbance of the geomagnetic field in magnetic storms. Modelling of the Dst index is thus very important for the analysis of the geomagnetic field. A data-based modelling approach, aimed at obtaining efficient models from limited input–output observational data, provides a powerful tool for analysing and forecasting geomagnetic activities including the prediction of the Dst index. In this study, the process of the Dst index is treated to be a structure-unknown system, where the solar wind parameter (VBs) and the solar wind dynamic pressure (P) are the system inputs, and the Dst index is the system output. A novel multiscale RBF (MSRBF) network is introduced to represent such a two-input and single-output system, where the Dst index is related to the solar wind parameter and the dynamic pressure, via a hybrid network model consisting of two submodels: a linear part that reflects the linear relationship between the output and the inputs, and a nonlinear part that captures the effect of the interacting contribution of past observations of the inputs and the output, on the current output. The proposed MSRBF network can easily be converted into a linear-in-the-parameters form and the training of the linear network model can easily be implemented using a forward orthogonal regression (FOR) algorithm. One advantage of the new MSRBF network, compared with traditional single scale RBF networks, is that the new network is more flexible for describing complex nonlinear dynamical systems.     

Nighttime cloud properties retrieval using MODIS and artificial neural networks

In this work a methodology for inferring water cloud macro and microphysical properties from nighttime MODIS imagery is developed. This method is based on the inversion of a theoretical radiative transfer model that simulates the radiances detected in each of the sensor infrared bands. To accomplish this inversion, an operational technique based on Artificial Neural Networks (ANNs) is proposed, whose main characteristic is the ability to retrieve cloud properties much faster than conventional methods. Furthermore, a detailed study of input data is performed to avoid different sources of errors that appear in several MODIS infrared channels. Finally, results of applying the proposed method are compared with in-situ measurements carried out during the DYCOMS-II field experiment.     

Predicting the probability of occurrence of spread-F over Brazil using neural networks

The probability of occurrence of spread-F can be modeled and predicted using neural networks (NNs). This paper presents a feasibility study into the development of a NN based model for the prediction of the probability of occurrence of spread-F over selected equatorial stations within the Brazilian sector. The input space included the day number (seasonal variation), hour (diurnal variation), sunspot number (measure of the solar activity), magnetic index (measure of the magnetic activity) and magnetic position. Twelve years of spread-F data from Brazil (covering the period 1978–1989) measured at the equatorial site Fortaleza (3.9°S, 38.45°W) and low latitude site Cachoeira Paulista (22.6°S, 45.0°W) are used in the development of an input space and NN architecture for the model. Spread-F data that is believed to be related to plasma bubble developments (range spread-F) was used in the development of the model. The model results show the probability of spread-F occurrence as a function of local time, season and latitude. Results from the Brazilian Sector NN (BSNN) based model are presented in this paper, as well as a comparative analysis with a Brazilian model developed for the same purpose.     

An attempt to study long-term variation of sporadic E layers using neural networks

Strong positive correlation between sporadic E layers and the solar activity and the long-term declining trend of Es were found in this paper. Then the feed-forward back propagation neural networks (NNs) were used to simulate the long-term variation of Es at four stations and predict foEs yearly average values. The inputs used for NNs are the yearly mean values of foEs in the daytime of the past ten years and the yearly averaged data of solar 10.7 cm radio flux (F107) of the present year, and the output is the present yearly mean value of daytime foEs. The outputs of trained NNs have high correlation with the desired values and the foEs yearly mean values predicted by NNs have good agreement with the observed data. The results indicate that NNs can make full use of the observed data to simulate the long variation rule of Es. Also, the results confirm the effect of solar activity on Es.     

A steady-state model for the D- to F-region of the polar cap

For obvious reasons the ionosphere of the polar cap, surrounded by the auroral zone, is only poorly investigated. Even ionosonde data are very scant from geomagnetic latitudes beyond 70°. Since 1997 the European incoherent scatter radar facility EISCAT has an additional installation on Svalbard and has been providing electron density data nearly continuously ever since. These measurements which mainly cover the E- and F-regions are supplemented by rocket data from Heiss Island at a comparable magnetic latitude; these data are more sporadic, but cover lower altitudes and densities. A provisional, steady-state, neural network-based model is presented which uses the data of both sites.     

Neural network development for the forecasting of upper atmosphere parameter distributions

This paper presents a neural network modeling approach to forecast electron concentration distributions in the 150–600 km altitude range above Arecibo, Puerto Rico. The neural network was trained using incoherent scatter radar data collected at the Arecibo Observatory during the past two decades, as well as the Kp geomagnetic index provided by the National Space Science Data Center. The data set covered nearly two solar cycles, allowing the neural network to model daily, seasonal, and solar cycle variations of upper atmospheric parameter distributions. Two types of neural network architectures, feedforward and Elman recurrent, are used in this study. Topics discussed include the network design, training strategy, data analysis, as well as preliminary testing results of the networks on electron concentration distributions.     

New flare model using recent measurements of the solar ultraviolet irradiance

The solar photon output from the Sun, which was once thought to be constant, varies considerably over time scales from seconds during solar flares to years due to the solar cycle. This is especially true in the wavelengths shorter than 190 nm. These variations cause significant deviations in the Earth and space environment on similar time scales, which then affects many things including satellite drag, radio communications, atmospheric densities and composition of particular atoms, molecules, and ions of Earth and other planets, as well as the accuracy in the Global Positioning System (GPS). The Flare Irradiance Spectral Model (FISM) is an empirical model that estimates the solar irradiance at wavelengths from 0.1 to 190 nm at 1 nm resolution with a time cadence of 60 s. This is a high enough temporal resolution to model variations due to solar flares, for which few accurate measurements at these wavelengths exist. This model also captures variations on the longer time scales of solar rotation (days) and solar cycle (years). Daily average proxies used are the 0–4 nm irradiance, the Mg II c/w, F10.7, as well as the 1 nm bins centered at 30.5 nm, 121.5 (Lyman Alpha), and 36.5 nm. The GOES 0.1–0.8 nm irradiance is used as the flare proxy. The FISM algorithms are given, and results and comparisons are shown that demonstrate the FISM estimations agree within the stated uncertainties to the various measurements of the solar Vacuum Ultraviolet (VUV) irradiance.     

Application of neural networks to South African GPS TEC modelling

The propagation of radio signals in the Earth’s atmosphere is dominantly affected by the ionosphere due to its dispersive nature. Global Positioning System (GPS) data provides relevant information that leads to the derivation of total electron content (TEC) which can be considered as the ionosphere’s measure of ionisation. This paper presents part of a feasibility study for the development of a Neural Network (NN) based model for the prediction of South African GPS derived TEC. The South African GPS receiver network is operated and maintained by the Chief Directorate Surveys and Mapping (CDSM) in Cape Town, South Africa. Vertical total electron content (VTEC) was calculated for four GPS receiver stations using the Adjusted Spherical Harmonic (ASHA) model. Factors that influence TEC were then identified and used to derive input parameters for the NN. The well established factors used are seasonal variation, diurnal variation, solar activity and magnetic activity. Comparison of diurnal predicted TEC values from both the NN model and the International Reference Ionosphere (IRI-2001) with GPS TEC revealed that the IRI provides more accurate predictions than the NN model during the spring equinoxes. However, on average the NN model predicts GPS TEC more accurately than the IRI model over the GPS locations considered within South Africa.     

Effects of geomagnetic activity variations on the physiological and psychological state of functionally healthy humans: Some results of Azerbaijani studies

There are collaborative and cross-disciplinary space weather studies in the Azerbaijan National Academy of Sciences conducted with purposes of revealing possible effects of solar, geomagnetic and cosmic ray variability on certain technological, biological and ecological systems. This paper describes some results of the experimental studies of influence of the periodical and aperiodical changes of geomagnetic activity upon human brain, human health and psycho-emotional state. It also covers the conclusions of studies on influence of violent solar events and severe geomagnetic storms of the solar cycle 23 on the mentioned systems in middle-latitude location. It is experimentally established that weak and moderate geomagnetic storms do not cause significant changes in the brain’s bioelectrical activity and exert only stimulating influence while severe disturbances of geomagnetic conditions cause negative influence, seriously disintegrate brain’s functionality, activate braking processes and amplify the negative emotional background of an individual. It is concluded that geomagnetic disturbances affect mainly emotional and vegetative spheres of human beings while characteristics reflecting personality properties do not undergo significant changes.     

Effects of geomagnetic activity and atmospheric power variations on quantitative measures of brain activity: Replication of the Azerbaijani studies

This study replicates and extends the observations by Babayev and Allahveriyeva that changes in right hemispheric electroencephalographic activity are correlated with increases in geomagnetic activity. During the geomagnetically quiet interface between solar cycle 23 and 24 quantitative electroencephalographic (QEEG) measurements were completed for normal young adults in three separate experiments involving about 120 samples over 1.5 years. The most consistent, moderate strength correlations occurred for the changes in power within the gamma and theta ranges over the right frontal lobe. Real-time measures of atmospheric power obtained from polar orbiting satellites showed similar effects. The preferential involvement of the right frontal lobe and the regions subject to its inhibition with environmental energetic changes are consistent with the behavioural correlations historically associated with these conditions. They include increased incidence of emotional lability, erroneous reconstruction of experiences, social confrontations, and unusual perceptions.     

The possibilities of polar meteorology,environmental remote sensing,communications and space weather applications from Artificial Lagrange Orbit

The ability to observe meteorological events in the polar regions of the Earth from satellite celebrated an anniversary, with the launch of TIROS-I in a pseudo-polar orbit on 1 April 1960. Yet, after 50 years, polar orbiting satellites are still the best view of the polar regions of the Earth. The luxuries of geostationary satellite orbit including rapid scan operations, feature tracking, and atmospheric motion vectors (or cloud drift winds), are enjoyed only by the middle and tropical latitudes or perhaps only cover the deep polar regions in the case of satellite derived winds from polar orbit. The prospect of a solar sailing satellite system in an Artificial Lagrange Orbit (ALO, also known as “pole sitters”) offers the opportunity for polar environmental remote sensing, communications, forecasting and space weather monitoring. While there are other orbital possibilities to achieve this goal, an ALO satellite system offers one of the best analogs to the geostationary satellite system for routine polar latitude observations.     

A new global version of M(3000)F2 prediction model based on artificial neural networks

A new version of global empirical model for the ionospheric propagation factor, M(3000)F2 prediction is presented. Artificial neural network (ANN) technique was employed by considering the relevant geophysical input parameters which are known to influence the M(3000)F2 parameter. This new version is an update to the previous neural network based M(3000)F2 global model developed by Oyeyemi et al. (2007), and aims to address the inadequacy of the International Reference Ionosphere (IRI) M(3000)F2 model (the International Radio Consultative Committee (CCIR) M(3000)F2 model). The M(3000)F2 has been found to be relatively inaccurate in representing the diurnal structure of the low latitude region and the equatorial ionosphere. In particular, the existing hmF2 IRI model is unable to reproduce the sharp post-sunset drop in M(3000)F2 values, which correspond to a sharp post-sunset peak in the peak height of the F2 layer, hmF2. Data from 80 ionospheric stations globally, including a good number of stations in the low latitude region were considered for this work. M(3000)F2 hourly values from 1987 to 2008, spanning all periods of low and high solar activity were used for model development and verification process. The ability of the new model to predict the M(3000)F2 parameter especially in the low latitude and equatorial regions, which is known to be problematic for the existing IRI model is demonstrated.     

Instantaneous rainfall estimation using neural network from multispectral observations of SEVIRI radiometer and its application in estimation of daily and monthly rainfall

In the present paper, an artificial neural network (ANN) based technique has been developed to estimate instantaneous rainfall by using brightness temperature from the IR sensors of SEVIRI radiometer, onboard Meteosat Second Generation (MSG) satellite. The study is carried out over north of Algeria. For estimation of rainfall, weight matrices of two ANNs namely MLP1 and MLP2 are developed. MLP1 is to identify raining or non-raining pixels. When rainy pixels are identified, then for those pixels, instantaneous rainfall is estimated by using MLP2. For identification of raining and non raining pixels, 7 input parameters from the IR sensors are utilized. Corresponding data of raining/non-raining pixels are taken from radar. For instantaneous rainfall estimation, 14 input parameters are utilized, where 7 parameters are information about raining pixels and 7 parameters are related with cloud features. The results obtained show the neural network performs reasonably well.