Space weather dependence of decameter-scale ionospheric irregularities spatial distribution

Oleg I. Berngardt; Ivan A. Lavygin

physics.ao-phOct 21, 2026

Space weather dependence of decameter-scale ionospheric irregularities spatial distribution

Authors: Oleg I. Berngardt, Ivan A. Lavygin

Organizations: ISTP SB RAS, 664033, Lermontova Str, 126A, Irkutsk, Russia

Abstract

This paper describes an approach to constructing nowcast and forecast predictive models for the occurrence of high-latitude decameter ionospheric irregularities based on SuperDARN and SECIRA radar data and space weather parameters. The method relies on self-learning pattern recognition in the dynamics of space weather parameters, that influence the irregularities formation. The proposed approach is a data-driven, multiscale method in which the spatial distribution of irregularities is approximated using a neural network of an original architecture, inspired by graph neural networks. The model was trained and tested using data from 11 radars during the period 2010-2025 (SuperDARN: CVW, ICE, ICW, INV, PGR, PYK, RKN, SAS, STO; SECIRA: EKB, MAGW). It is shown that the model achieves optimal performance when using geomagnetic coordinates MLT,MLAT. The optimal set of parameters for the best nowcast and forecast over studied set of parameters is a combination of geomagnetic and solar disturbance indices ( $a_{p}$ , $K_{p}$ , $AE$ , F10.7) from ground-based observations, and satellite-measured solar wind and interplanetary magnetic field parameters ( $P_{sw}$ , $E_{sw}$ , $\widetilde{B_{z}}$ , $T_{sw}$ ). Hourly mean values for all the parameters were used in the analysis. The results indicate that acceptable forecast quality is achieved with a lead time of up to 7-9 hours, with the highest quality reached for nowcast (zero lead time). It is shown that the parameters $\widetilde{B_{z}}$ and F10.7 are significant for nowcasting only and can be neglected for other lead times. It is shown that $AE$ and $T_{sw}$ are significant for mid-term forecast (\textgreater 2hours ahead) and can be neglected for short-term predictions. Statistical relationships between the maximum latitude of the equatorial boundary of the irregularities and space weather parameters are also discussed.

Figures & tables

Radar	Coordinates	Years
CVW SuperDARN	43N 120W	2020,2023
EKB SECIRA	56N 58E	2021, 2025
ICE SuperDARN	64N 21W	2024
ICW SuperDARN	64N 21W	2024
INV SuperDARN	68N 134W	2018,2023
PGR SuperDARN	54N 123W	2020,2023

Table 1: Radars used in the analysis, their geographical coordinates, and the years analyzed. Data not used for model training are highlighted in bold.

Figure 1: (A) Radars fields-of-view in geographic coordinates; (B) Data preparation algorithm.

Figure 2: Statistics of observations in MLAT,MLT of ionospheric scattering according to three ionospheric echo detection algorithms: (A) - FA; (B) - CA; (C) - 36C.

Parameters	Comments	Data source
$K_{p}$ , $a_{p}$	Geomagnetic disturbance indices	[ 18 ]
F10.7	Solar activity index	[ 18 ]
AE, AU, AL, AO	Auroral geomagnetic disturbance indices	[ 22 ]
ASY-H, ASY-D, SYM-D, SYM-H	Mid-latitude geomagnetic disturbance indices	[ 22 ]
$B_{x}$ , $\widetilde{B_{y}}$ , $\widetilde{B_{z}}$ , $B_{t}$ , $\psi_{CA}$	Interplanetary magnetic field parameters in GSM coordinates and IMF clock angle	[ 28 ]
$V_{x}$ , $V_{y}$ , $V_{z}$	Solar wind speed in the longitudinal (x) and transverse (y, z) directions.	[ 28 ]

Table 2: Parameters used for modeling

IS detecting algorithm	Significant parameters
FA	MLAT,MLT,AE,AL, $\widetilde{B_{z}}$
CA	MLAT,MLT,AL,GLAT, $\widetilde{B_{z}}$ ,AE, $E_{sw}$ , $B_{t}$
36C	MLAT,GLAT

Table 3: Hourly-averaged parameters most significant for nowcasting (in descending order of significance), using various algorithms of detecting ionospheric scattering from mixture with other signals (simple nowcast network).

Figure 3: (A) - dataset structure in the form of dynamic graphs, dashed lines indicates nodes, arrows - directed edges; (B-D) - regression neural network (RN-NN) for predicting the probability of ionospheric scattering and its training method in nowcasting mode (B) and in forecasting mode (C). Dashed lines indicate blocks fed into the network simultaneously. Notation: QT - quantile transformer, RN - recurrent neural network, NN - attention neural network. (D) - NN architecture. FC - standard dense (i.e.fully connected) layer, GeLU - nonlinear activation function, Dropout - standard dropout layer with dropout probability 0.3, Clip - clipping values to the interval [0,1]. (E) - RN architecture.

Figure 4: Comparison of forecast model quality at different lead times. Marked regions are: (I) - nowcast and short-term forecast; (II) - mid-term forecast; (III) - long-term forecast.

Figure 5: Importance of model parameters for different forecast lead times (from 0 hours to 192 hours)

Forecast lead time(h)	Important parameters
0	$a_{p},E_{sw},K_{p},P_{sw},\widetilde{B_{z}}$ ,F10.7
1	$a_{p},K_{p},P_{sw},E_{sw}$ ,AO
2	$a_{p},K_{p},P_{sw},E_{sw}$ ,SYM-H
3	$a_{p},K_{p},T_{sw},P_{sw}$ ,AE
4	$a_{p},K_{p}$ ,AE, $P_{sw}$ ,ASY-H
5	AE, $K_{p},P_{sw},a_{p},T_{sw}$

Table 4: Most important model parameters by forecast lead time (MLAT and MLT excluded), in descending order of importance

Figure 6: Training quality of distilled models at train and test datasetfor various lead times in comparison with initial model: initial model - 23 input parameters, final distilled model - 8 parameters: $a_{p},K_{p},P_{sw},E_{sw},\widetilde{B_{z}},T_{sw}$ ,AE,F10.7

Figure 7: Prediction statistics of $P(OBS|\theta_{s})$ . Intervals corresponds the mean results, error interval shows standard deviation over the dataset.

Figure 8: Model predictions on the validation dataset in nowcasting mode for the geomagnetic storm of August 11-13, 2024. The left panel shows the time series of the indices $a_{p}$ , $E_{sw}$ , $K_{p}$ , $P_{sw}$ , and $\widetilde{B_{z}}$ . The right panel displays model-predicted probabilities of observing ionospheric irregularities $P(IS|\theta_{s})$ as circles, and experimental observations of the probability that a scattered signal originates from ionospheric scattering $P(IS|OBS,\theta_{s})$ as squares. The line indicates the model border $P(IS|\theta_{s})=0.1$ .

Figure 9: Dependence of the maximum latitude of the equatorial boundary of the model region containing irregularities (nowcasting) on space weather parameters based on whole available dataset 2010-2025.

Spatiotemporal Forecasting

Spatiotemporal forecasting predicts how quantities change across locations and time, from traffic and weather to disease and environmental systems.

See all 429 papers

Electric Potential Patterns Forecasting in the Southern Hemisphere with Deep Learning Techniques
Sep 25, 2026Francesco Pio Ramunno, Simone Mestici, Igino Coco +5Spatiotemporal Forecasting Probabilistic Forecasting
University of Applied Sciences and Arts Nortwestern Switzerland · Department of Computer Science, University of Geneva, 1211 Geneva, Switzerland · Universit´a Roma ”La Sapienza” +3
Forecasting Ionospheric Irregularities on GNSS Lines of Sight Using Dynamic Graphs with Ephemeris Conditioning
Apr 20, 2026Mert Can Turkmen, Eng Leong Tan, Yee Hui LeeProbabilistic Forecasting Dynamic Graph Learning
Nanyang Technological University Singapore
t-STEP: An interpretable model for Total Electron Content predictions and irregularities estimations
Jun 28, 2026Stephen Tete, Carl Shneider, Maxime Cordy +3Time Series Forecasting Spatiotemporal Forecasting
Interdisciplinary Center for Security, Reliability and Trust (SnT), University of Luxembourg, Luxembourg City L-1855, Luxembourg · Istituto Nazionale di Geofisica e Vulcanologia, Via di Vigna Murata 605, Rome 00143, Italy · Mission Space | Space Weather, Luxembourg City L-1191, Luxembourg

Sep 25, 2026J/K move · Enter open · S save

Electric Potential Patterns Forecasting in the Southern Hemisphere with Deep Learning Techniques

Francesco Pio Ramunno, Simone Mestici, Igino Coco, Maria Walach+4

University of Applied Sciences and Arts Nortwestern Switzerland · Department of Computer Science, University of Geneva, 1211 Geneva, Switzerland · Universit´a Roma ”La Sapienza”+3

Spatiotemporal Forecasting Probabilistic Forecasting Diffusion Models

Open paper AskPDF arXiv

Space weather disturbances driven by the solar wind can degrade satellite navigation, disrupt radio communications, and threaten power infrastructure, making accurate forecasting of the high-latitude ionosphere's response a critical operational need. Existing approaches, empirical climatological models and physics-based magnetohydrodynamic simulations, either smooth out the ionosphere's time-dependent, non-linear response or are too computationally expensive for real-time use, while prior machine learning efforts have mostly targeted scalar indices rather than the full spatial structure of ionospheric convection. Here we train and compare three deep learning architectures, a probabilistic diffusion model conditioned on multi-variate solar wind and interplanetary magnetic field measurements at the L1 Lagrange point, an unconditioned diffusion ablation, and a deterministic U-Net baseline, to forecast Southern Hemisphere high-latitude electric potential maps derived from SuperDARN radar observations. Using five years (2020--2025) of SuperDARN data synchronised with DSCOVR L1 measurements, we evaluate the models under a single-pass regime, an extended autoregressive rollout of up to 350 frames, and an out-of-distribution case study on the intense March 2015 St.\ Patrick's Day storm, unseen during training. We find that the relative advantage of the deterministic and probabilistic approaches is not fixed: the deterministic model is competitive over short horizons and calm conditions, while the diffusion model's advantage grows and eventually dominates as the forecast horizon lengthens and the event becomes more dynamic, evidence that probabilistic, sample-based generative models are the more promising direction for operational, long-horizon space weather forecasting.

Abstract

Spatiotemporal Forecasting

Explore similar work

Electric Potential Patterns Forecasting in the Southern Hemisphere with Deep Learning Techniques

Forecasting Ionospheric Irregularities on GNSS Lines of Sight Using Dynamic Graphs with Ephemeris Conditioning

t-STEP: An interpretable model for Total Electron Content predictions and irregularities estimations