Search

Coronal mass ejections (CMEs) are the main drivers of geomagnetic storms at Earth and are capable of impacting power grids on the planet’s surface as well as satellites in orbit. Developing models to accurately predict the arrival time and speed of CMEs is a necessary step towards ensuring that we are capable of providing advance warning to mitigate potential severe space weather effects. We use the Ellipse Evolution model based on Heliospheric Imager observations (ELEvoHl) to predict the arrival of CMEs at Earth using data from STEREO-A’s heliospheric imagers. The HI instruments provide both real-time, lower-quality beacon data as well as higher- quality science data which is downlinked to Earth with some delay. Issuing alerts for approaching CMEs before they arrive at Earth necessitates the utilization of real-time data. HI data for each event and data type is compiled into a time-elongation plot (Jmap) in which every CME’s trajectory is tracked manually. We predict the arrival time and speed of 10 Earth-directed CMEs from 2010 to 2020 using both science and beacon data with the ELEvoHI model and compare the results in terms of accuracy across both data types. We find that predictions made using beacon data are generally worse than those made using science data and give an outlook on possible future perspectives for improving the real-time prediction of CMEs.

Planetary magnetospheres create multiple sharp boundaries, such as the bow shock, where the solar wind plasma is decelerated and compressed, or the magnetopause, a transition between solar wind field and planetary field.We attempt to use convolutional neural networks (CNNs) to identify magnetospheric boundaries, i.e. planetary and interplanetary shocks crossings and magnetopause crossings in spacecraft in situ data. The boundaries are identified by a discontinuity in a magnetic field, plasma density, and in the spectrum of high-frequency waves. These measurements are available on many planetary missions. Data from Earth's missions Cluster and THEMIS are used for CNN training. We ultimately strive for successful classification of boundaries (shock, magnetopause, inbound, outbound) and the correct handling of multiple crossings.

Interplanetary coronal mass ejections (ICMEs) are one of the main drivers for space weather disturbances. In the past,different machine learning approaches have been used to automatically detect events in existing time series resulting fromsolar wind in situ data. However, classification, early detection and ultimately forecasting still remain challenges when facingthe large amount of data from different instruments. We propose a pipeline using a Network similar to the ResUNet++ (Jha et al. (2019)), for the automatic detection of ICMEs. Comparing it to an existing method, we find that while achieving similar results, our model outperforms the baseline regarding GPU usage, training time and robustness to missing features, thus making it more usable for other datasets.The method has been tested on in situ data from WIND. Additionally, it produced reasonable results on STEREO A and STEREO B datasetswith less input parameters. The relatively fast training allows straightforward tuning of hyperparameters and could therefore easily be used to detect other structures and phenomena in solar wind data, such as corotating interaction regions.

During the last years, we focused on developing a prediction tool that utilizes the wide-angle observations of STEREO's heliospheric imagers. The unsurpassable advantage of these imagers is the possibility to observe the evolution and propagation of a coronal mass ejection (CME) from close to the Sun up to 1 AU and beyond. We believe that using this advantage instead of relying on coronagraph observations that are limited to observe only 14% of the Sun-Earth line, it is possible to improve today's CME arrival time predictions.The ELlipse Evolution model based on HI observations (ELEvoHI) assumes an elliptic frontal shape within the ecliptic plane and allows the CME to adjust to the ambient solar wind speed, i.e. it is drag-based. ELEvoHI is used as an ensemble simulation by varying the CME frontal shape within given boundary values. The results include a frequency distrubution of predicted arrival time and arrival speed and an estimation of the arrival probability. ELEvoHI can be operated using several kinds of inputs. In this study we investigate 15 well-defined single CMEs when STEREO was around L4/5 between the end of 2009 and the beginning of 2011. Three different sources of input propagation directions (and shapes) are used together with three different sources of ambient solar wind speed and two different ways of defining the most appropriate fit to the HI data. The combination of these different approaches and inputs leads to 18 different model set-ups used to predict each of the 15 events in our list leading to 270 ELEvoHI ensemble predictions and all in all to almost 60000 runs. To identify the most suitable and most accurate model set-up to run ELEvoHI, we compare the predictions to the actual in situ arrival of the CMEs.This model is specified for using data from future space weather missions carrying HIs located at L5 or L1 and can also directly be used together with STEREO-A near real-time HI beacon data to provide real-time CME arrival predictions during the next 7 years when STEREO-A is observing the Sun-Earth space.

We introduce Helio4cast, an open source python package to provide real time solar wind predictions at the Sun-Earth L1 point, and to directly couple them to forecasts of the aurora oval, geomagnetically induced currents and further geomagnetic indices. We present its current status, using a combination of our PREDSTORM solar wind forecast and the real time modeling of the aurora with the OVATION model. The solar wind prediction is driven by data from either STEREO-A, a recurrence model, an empirical background solar wind model or a future L5 mission. For coronal mass ejections (CMEs), we plan to use our semi-empirical 3DCORE model to produce in situ magnetic flux rope signatures constrained by real-time solar observations, or a machine learning approach based on many previous observations of in situ CMEs. We are particularly interested in how the errors in the solar wind prediction propagate to ground-based observations. Challenges and future plans of the real-time implementation are discussed.