A5024352463- Solar and Space Plasma Dynamics
- Stellar, planetary, and galactic studies
- Solar Radiation and Photovoltaics
- Astro and Planetary Science
- Ionosphere and magnetosphere dynamics
- Hydrocarbon exploration and reservoir analysis
- Oil, Gas, and Environmental Issues
- GNSS positioning and interference
- Wireless Communication Security Techniques
- Distributed Sensor Networks and Detection Algorithms
- Atmospheric and Environmental Gas Dynamics
- Computational Physics and Python Applications
- Seismic Imaging and Inversion Techniques
- Geomagnetism and Paleomagnetism Studies
- Geophysics and Gravity Measurements
- Laser-induced spectroscopy and plasma
- Sparse and Compressive Sensing Techniques
- Plasma Diagnostics and Applications
- Dust and Plasma Wave Phenomena
- Anomaly Detection Techniques and Applications
University of Bern
2023
University of Geneva
2018-2023
FHNW University of Applied Sciences and Arts
2018-2021
IRIS performs solar observations over a large range of atmospheric heights, including the chromosphere where majority flare energy is dissipated. The strong Mg II h&k spectral lines are capable providing excellent diagnostics, but have not been fully utilized for flaring atmospheres. We aim to investigate whether physics identical all by analyzing if there certain spectra that occur in flares. To achieve this, we automatically analyze hundreds thousands line profiles from set 33 flares, and...
Abstract Spectral lines allow us to probe the thermodynamics of solar atmosphere, but shape a single spectral line may be similar for different thermodynamic solutions. Multiline analyses are therefore crucial, computationally cumbersome. We investigate correlations between several chromospheric and transition region restrain solutions atmosphere during flares. used machine-learning methods capture statistical dependencies six sourced from 21 large flares observed by NASA’s Interface Region...
The prediction of solar flares is practical and scientific interest; however, many machine learning methods used for this task do not provide the physical explanations behind a model's performance. We made use two recently developed explainable artificial intelligence techniques called gradient-weighted class activation mapping (Grad-CAM) expected gradients (EG) to reveal decision-making process high-performance neural network that has been trained distinguish between MgII spectra derived...
Abstract A three-dimensional picture of the solar atmosphere’s thermodynamics can be obtained by jointly analyzing multiple spectral lines that span many formation heights. In Paper I, we found strong correlations between shapes from a variety different ions during flares in comparison to quiet Sun. We extend these techniques address following questions: which regions atmosphere are most connected flare, and what likely responses across several windows based on observation single Mg ii...
Context. The origin of the slow solar wind is still an open issue. It has been suggested that upflows at edge active regions are a possible source plasma outflow and therefore contribute to wind. Aims. We investigate morphology upflow compare region core properties. Methods. studied how properties flux, Doppler velocity, non-thermal velocity change throughout atmosphere, from chromosphere via transition corona in region. limb-to-limb observations (NOAA 12687) obtained 14 25 November 2017....
Abstract With machine learning entering into the awareness of heliophysics community, solar flare prediction has become a topic increased interest. Although machine-learning models have advanced with each successive publication, input data remained largely fixed on magnetic features. Despite this model complexity, results seem to indicate that photospheric field alone may not be wholly sufficient source for prediction. For first time, we extended study spectral data. In work, use Deep Neural...
Small reconnection events in the lower solar atmosphere can lead to its heating, but whether such heating propagate into higher atmospheric layers and potentially contribute coronal is an open question. We carry out a large statistical analysis of all IRIS observations from 2013 2014. identified "IRIS burst" (IB) spectra via k-means by classifying selecting Si IV with superimposed blend lines on top bursts, which indicate low heating. found that ~8% show IBs about 0.01% recorded being IB...
We present a new decentralized classification system based on distributed architecture. This consists of nodes, each possessing their own datasets and computing modules, along with centralized server, which provides probes to aggregates the responses nodes for final decision. Each node, access its training dataset given class, is trained an auto-encoder consisting fixed data-independent encoder, pre-trained quantizer class-dependent decoder. Hence, these auto-encoders are highly dependent...
Context. Reliably predicting solar flares can mitigate the risks of technological damage and enhance scientific output by providing reliable pointings for observational campaigns. Flare precursors in spectral line Mg II have been identified. Aims. We extend previous studies examining presence flare additional lines, such as Si IV C , over longer time windows, more observations. Methods. trained neural networks XGBoost decision trees to distinguish spectra observed from active regions that...
The Spectrometer/Telescope for Imaging X-rays (STIX) on-board the ESA Solar Orbiter mission retrieves coordinates of solar flare locations by means a specific sub-collimator, named Coarse Flare Locator (CFL). When occurs on Sun, emitted X-ray radiation casts shadow peculiar "H-shaped" tungsten grid over CFL detector. From measurements areas detector that are illuminated radiation, it is possible to retrieve $(x,y)$ location disk. In this paper, we train neural network dataset real...
Whitepaper #012 in the Decadal Survey for Solar and Space Physics (Heliophysics) 2024-2033. Main topics: basic research. Additional solar physics; space-based missions/projects; emerging opportunities.
Flares are an eruptive phenomenon observed on the sun, which major protagonists in space weather and can cause adverse effects such as disruptions communication, power grid failure damage of satellites. Our method answers importance time component some scientific video observations, especially for flare detection study is based NASA's Interface Region Imaging Spectrograph (IRIS) observations sun since 2013, consists a very asymmetrical unlabeled big data. For detecting analyzing flares our...
The prediction of solar flares is practical and scientific interest; however, many machine learning methods used for this task do not provide the physical explanations behind a model's performance. We made use two recently developed explainable artificial intelligence techniques called gradient-weighted class activation mapping (Grad-CAM) expected gradients (EG) to reveal decision-making process high-performance neural network that has been trained distinguish between MgII spectra derived...