Solar-wind forecasting is critical for predicting events which can affect Earth's technological systems.  Typically, forecasts combine coronal model outputs with heliospheric models to predict near-Earth conditions. Ensemble forecasting generates sets of outputs to create probabilistic forecasts which quantify forecast uncertainty, vital for reliable/actionable forecasts. We adapt meteorological methods to create a calibrated solar-wind ensemble and probabilistic forecast for ambient solar wind, a prerequisite for accurate coronal mass ejection (CME) forecasting. Calibration is achieved by adjusting ensemble inputs/outputs to align the ensemble spread with observed event frequencies. We produce hindcasts in near-Earth space using coronal-model output over Solar Cycle 24, as input to Heliospheric Upwind eXtrapolation with time dependence (HUXt) solar-wind model. Making spatial perturbations to the coronal model output at 0.1~AU, we produce ensembles of inner-boundary conditions for HUXt, evaluating how forecast accuracy was impacted by the scales of perturbations applied. We found optimal spatial perturbations described by Gaussian distributions with variances of 20 degrees latitude and 10 degrees longitude; these might represent spatial uncertainty within the coronal model. This produced probabilistic forecasts better matching observed frequencies.  Calibration improved forecast reliability, reducing the Brier score by 9% and forecast decisiveness increasing AUC ROC score by 2.5%. Improvements were subtle but systematic. Additionally, we explored statistical post-processing to correct over-confidence bias, improving forecast actionability. However, this method, applied post-run, does not affect the solar-wind state used to propagate CMEs. This work represents the first formal calibration of solar-wind ensembles, laying groundwork for comprehensive forecasting systems like a calibrated multi-model ensemble.

A document by Harriet Turner. Click on the document to view its contents.

A document by Michael Lockwood. Click on the document to view its contents.

Knowledge of the ambient solar wind is important for accurate space-weather forecasting. A simple-but-effective method of forecasting near-Earth solar-wind conditions is “corotation”, wherein solar-wind structure is assumed to be fixed in the reference frame rotating with the Sun. Under this approximation, observations at a source spacecraft can be rotated to a target location, such as Earth. Forecast accuracy depends upon the rate of solar-wind evolution, longitudinal and latitudinal separation between the source and target, and latitudinal structure in the solar wind itself. The time-evolution rate and latitudinal structure of the solar wind are both strongly influenced by the solar cycle, though in opposing ways. A latitudinal offset is typically present, introducing an error to corotation forecasts. In this study, we use observations from the STEREO and near-Earth spacecraft to quantify the latitudinal error. Aliasing between the solar cycle and STEREO orbits means that individual contributions to the forecast error are difficult to isolate. However, by considering an 18-month interval near the end of solar minimum, we find that the latitudinal-offset contribution to corotation-forecast error cannot be directly detected for offsets <6º, but is increasingly important as offsets increase. This result can be used to improve solar-wind data assimilation, allowing representivity errors in solar-wind observations to be correctly specified. Furthermore, as the maximum latitudinal offset between L5 and Earth is ≈5º, corotation forecasts from a future L5 spacecraft should not be greatly affected by latitudinal offset.

Accurate forecasting of the arrival time and arrival speed of coronal mass ejections (CMEs) is a unsolved problem in space weather research. In this study, a comparison of the predicted arrival times and speeds for each CME based, independently, on the inputs from the two STEREO vantage points is carried out. We perform hindcasts using ELlipse Evolution model based on Heliospheric Imager observations (ELEvoHI) ensemble modelling. An estimate of the ambient solar wind conditions is obtained by the Wang-Sheeley-Arge/Heliospheric Upwind eXtrapolation (WSA/HUX) model combination that serves as input to ELEvoHI. We carefully select 12 CMEs between February 2010 and July 2012 that show clear signatures in both STEREO-A and STEREO-B HI time-elongation maps, that propagate close to the ecliptic plane, and that have corresponding in situ signatures at Earth. We find a mean arrival time difference of 6.5 hrs between predictions from the two different viewpoints, which can reach up to 9.5 hrs for individual CMEs, while the mean arrival speed difference is 63 km s$^{-1}$. An ambient solar wind with a large speed variance leads to larger differences in the STEREO-A and STEREO-B CME arrival time predictions ($cc = 0.92$). Additionally, we compare the predicted arrivals, from both spacecraft, to the actual in situ arrivals at Earth and find a mean absolute error of 7.5 $\pm$ 9.5 hrs for the arrival time and 87 $\pm$ 111 km s$^{-1}$ for the arrival speed. There is no tendency for one spacecraft to provide more accurate arrival predictions than the other.

Accurate space weather forecasting requires advanced knowledge of the solar wind conditions in near-Earth space. Data assimilation (DA) combines model output and observations to find an optimum estimation of reality and has led to large advances in terrestrial weather forecasting. It is now being applied to space weather forecasting. Here, we use solar wind DA with in-situ observations to reconstruct solar wind speed in the ecliptic plane between 30 solar radii and Earth’s orbital radius. This is used to provide solar wind speed hindcasts. Here, we assimilate observations from the Solar Terrestrial Relations Observatory (STEREO) and the near-Earth dataset, OMNI. Analysis of two periods of time, one in solar minimum and one in solar maximum, reveals that assimilating observations from multiple spacecraft provides a more accurate forecast than using any one spacecraft individually. The age of the observations also has a significant impact on forecast error, whereby the mean absolute error (MAE) sharply increases by up to 23% when the forecast lead time first exceeds the corotation time associated with the longitudinal separation between the observing spacecraft and the forecast location. It was also found that removing coronal mass ejections from the DA input and verification time series reduces the forecast MAE by up to 10% as it removes false streams from the forecast time series. This work highlights the importance of an L5 space weather monitoring mission for near-Earth solar wind forecasting and suggests that an additional mission to L4 would further improve future solar wind DA forecasting capabilities.

The magnetosphere, ionosphere and thermosphere (MIT) act as a coherently integrated system (geospace), driven in part by solar influences and characterized by variability and complexity. Among the most important and yet uncertain aspects of the geospace system is energy and momentum coupling between regions, which is, in part, accomplished by the transfer of charged particles from the magnetosphere to the ionosphere in a process known as particle precipitation, and in the opposite direction by ion outflow. Both processes are inherently multiscale and manifest the variabilities and complexities of the geospace system. Despite the importance of the transfer of particles, existing models are increasingly ill-equipped to provide the specification necessary for the growing demand for geospace now- and forecasts. Due to recent trends in the availability of data, we now face an exciting opportunity to progress particle transfer in geospace through the intersection of traditional approaches and state-of-the-art data-driven sciences. We reveal novel particle transfer models utilizing machine learning (ML), present results from the models, and provide an evaluation of their capabilities including comparisons with observations and the current ’state-of-the-art’ models (e.g., OVATION Prime for particle precipitation and the Gamera-Ionosphere Polar Wind Model for ion outflow). We detail the data wrangling required to utilize the available geospace observations to make progress on the long-standing challenge of particle transfer and place specific emphasis on the discovery possible when ML models are appropriate and robustly interrogated in the context of physical understanding. Our presentation helps illustrate the trends in the application of data science in space science.

Characterising the location of the outer boundary of the outer radiation belt is a key aspect of improving radiation belt models and helps to constrain our understanding of the mechanisms by which the source and seed electron populations are transported into the radiation belts. In this paper, we hypothesise that there are statistical differences in the electron distribution function across the radiation belt outer boundary, and thus analyse electron flux data from the THEMIS (Time History of Events and Macroscale Interactions during Substorms) satellites to identify this location. We validate our hypothesis by using modelled electron L* values to approximately characterise the differences between electron distribution functions inside and outside of the radiation belts. Initially, we perform a simple statistical analysis by studying the radial evolution of the electron distribution functions. This approach does not yield a clear discontinuity, thus highlighting the need for more complex statistical treatment of the data. Subsequently, we employ machine learning (with no dependence on L*) to test a range of candidate outer boundary locations. By analysing the performance of the models at each candidate location, we identify a statistical boundary at ≈8 Earth radii, with results suggesting some variability. This statistical boundary is typically further out than those used in current radiation belt models.

Accurately forecasting the arrival of coronal mass ejections (CMEs) at Earth is important to enabling mitigation of the associated space weather risks to society. This is only possible with accurate modelling of the event. To do so, we must understand the propagation of a CME through the heliosphere and quantify the performance of models through comparison with spacecraft observations. For the December 12th 2008 Earth-directed CME event, we compute ensembles using the HUXt solar wind model to analyse CME distortion with a structured solar wind and explore hindcast arrival time error (ATE). By highlighting the impact CME shape has on Root-Mean-Square-Error (RMSE) values, we show that time-elongation profiles of fronts captured by the Heliospheric Imager (HI) instruments onboard NASA’s STEREO mission match those of the modelled CME nose and flank and can therefore be used to infer details of the longitudinal extent of the CME. We then show that accounting for CME distortion is important to enable accurate estimates of the CME arrival at Earth. This can be achieved by either using observations of multiple features in HI data to infer CME evolution or mapping the solar wind back to a lower inner boundary to allow CMEs to be distorted close to the Sun. For the event studied we show that these approaches resulted in reduced RMSEs of 0.726˚and 0.638˚ with an ATE of one hour and three hours respectively.

Geometric modelling of Coronal Mass Ejections (CMEs) is a widely used tool for assessing their kinematic evolution. Furthermore, techniques based on geometric modelling, such as ELEvoHI, are being developed into forecast tools for space weather prediction. These models assume that solar wind structure does not affect the evolution of the CME, which is an unquantified source of uncertainty. We use a large number of Cone CME simulations with the HUXt solar wind model to quantify the scale of uncertainty introduced into geometric modelling and the ELEvoHI CME arrival times by solar wind structure. We produce a database of simulations, representing an average, a fast, and an extreme CME scenario, each independently propagating through 100 different ambient solar wind environments. Synthetic heliospheric imager observations of these simulations are then used with a range of geometric models to estimate the CME kinematics. The errors of geometric modelling depend on the location of the observer, but do not seem to depend on the CME scenario. In general, geometric models are biased towards predicting CME apex distances that are larger than the true value. For these CME scenarios, geometric modelling errors are minimised for an observer in the L5 region. Furthermore, geometric modelling errors increase with the level of solar wind structure in the path of the CME. The ELEvoHI arrival time errors are minimised for an observer in the L5 region, with mean absolute arrival time errors of 8.2±1.2h, 8.3±1.0h, and 5.8±0.9h for the average, fast, and extreme CME scenarios

We introduce the inductive effects of polar cap motions towards and way from the Sun into magnetospheric electrodynamics and show how this explains observed Universal Time variations in hemispheric geomagnetic indices. The large (and growing) hemispheric asymmetry in the offsets of the geomagnetic (dip or eccentric dipole) poles from Earth’s rotational axis means that the effect is not cancelled out in global indices. By adding this effect to that of the Russell-McPherron effect on solar wind-magnetosphere coupling, that of ionospheric conductivities, and that of the solar wind dynamic pressure and dipole tilt on the near-Earth tail lobe field and cross-tail current sheet, we are able to model the persistent “equinoctial” time-of-day/time-of-year pattern (with additional net Universal time variations) observed in the an, as and am geomagnetic indices since 1959. We discuss the implications for the longitudinal dependence of the effects of extreme space weather events