Subscribe to the gold package and get unlimited access to Shamra Academy
Register a new userFast and Accurate Emulation of the SDO/HMI Stokes Inversion with Uncertainty Quantification
108
0
0.0
(
0
)
Ask ChatGPT about the research
No Arabic abstract
The Helioseismic and Magnetic Imager (HMI) onboard NASAs Solar Dynamics Observatory (SDO) produces estimates of the photospheric magnetic field which are a critical input to many space weather modelling and forecasting systems. The magnetogram products produced by HMI and its analysis pipeline are the result of a per-pixel optimization that estimates solar atmospheric parameters and minimizes disagreement between a synthesized and observed Stokes vector. In this paper, we introduce a deep learning-based approach that can emulate the existing HMI pipeline results two orders of magnitude faster than the current pipeline algorithms. Our system is a U-Net trained on input Stokes vectors and their accompanying optimization-based VFISV
rate research
Read More
Both NASAs Solar Dynamics Observatory (SDO) and the JAXA/NASA Hinode mission include spectropolarimetric instruments designed to measure the photospheric magnetic field. SDOs Helioseismic and Magnetic Imager (HMI) emphasizes full-disk high-cadence and good spatial resolution data acquisition while Hinodes Solar Optical Telescope Spectro-Polarimeter (SOT-SP) focuses on high spatial resolution and spectral sampling at the cost of a limited field of view and slower temporal cadence. This work introduces a deep-learning system named SynthIA (Synthetic Inversion Approximation), that can enhance both missions by capturing the best of each instruments characteristics. We use SynthIA to produce a new magnetogram data product, SynodeP (Synthetic Hinode Pipeline), that mimics magnetograms from the higher spectral resolution Hinode/SOT-SP pipeline, but is derived from full-disk, high-cadence, and lower spectral-resolution SDO/HMI Stokes observations. Results on held-out data show that SynodeP has good agreement with the Hinode/SOT-SP pipeline
We take advantage of the HMI/SDO instrument to study the naked emergence of active regions from the first imprints of the magnetic field on the solar surface. To this end, we followed the first 24 hours in the life of two rather isolated ARs that appeared on the surface when they were about to cross the central meridian. We analyze the correlations between Doppler velocities and the orientation of the vector magnetic field finding, consistently, that the horizontal fields connecting the main polarities are dragged to the surface by relatively-strong upflows and are associated to elongated granulation that is, on average, brighter than its surroundings. The main magnetic footpoints, on the other hand, are dominated by vertical fields and downflowing plasma. The appearance of moving dipolar features, MDFs, (of opposite polarity to that of the AR) in between the main footpoints, is a rather common occurrence once the AR reaches a certain size. The buoyancy of the fields is insufficient to lift up the magnetic arcade as a whole. Instead, weighted by the plasma that it carries, the field is pinned down to the photosphere at several places in between the main footpoints, giving life to the MDFs and enabling channels of downflowing plasma. MDF poles tend to drift towards each other, merge and disappear. This is likely to be the signature of a reconnection process in the dipped field lines, which relieves some of the weight allowing the magnetic arcade to finally rise beyond the detection layer of the HMI spectral line.
Downflows on the solar surface are suspected to play a major role in the dynamics of the convection zone. We investigate the existence of the long-lasting downflows whose effects influence the interior of the Sun and the outer layers. We study the sets of Dopplergrams and magnetograms observed with SDO and Hinode spacecrafts and a MHD simulation. All of the aligned sequences, which were corrected from the satellite motions and tracked with the differential rotation, were used to detect the long-lasting downflows in the quiet-Sun at the disc centre. To learn about the structure of the flows below the solar surface, the time-distance local helioseismology was used. The inspection of the 3D data cube (x, y, t) of the 24-hour Doppler sequence allowed us to detect 13 persistent downflows. Their lifetimes lie in the range between 3.5 and 20 hours with sizes between 2 and 3 and speeds between -0.25 and -0.72 km/s. These persistent downflows are always filled with the magnetic field with an amplitude of up to 600 G. The helioseismic inversion allows us to describe the persistent downflows and compare them to the other (non-persistent) downflows in the field of view. The persistent downflows seem to penetrate much deeper and, in the case of a well-formed vortex, the vorticity keeps its integrity to the depth of about 5 Mm. In the MHD simulation, only sub-arcsecond downflows are detected with no evidence of a vortex comparable in size to observations at the surface of the Sun. The long temporal sequences from the space-borne allow us to show the existence of long-persistent downflows together with the magnetic field. They penetrate inside the Sun but are also connected with the anchoring of coronal loops in the photosphere, indicating a link between downflows and the coronal activity. A link suggests that EUV cyclones over the quiet Sun could be an effective way to heat the corona.
Deep learning-based object pose estimators are often unreliable and overconfident especially when the input image is outside the training domain, for instance, with sim2real transfer. Efficient and robust uncertainty quantification (UQ) in pose estimators is critically needed in many robotic tasks. In this work, we propose a simple, efficient, and plug-and-play UQ method for 6-DoF object pose estimation. We ensemble 2-3 pre-trained models with different neural network architectures and/or training data sources, and compute their average pairwise disagreement against one another to obtain the uncertainty quantification. We propose four disagreement metrics, including a learned metric, and show that the average distance (ADD) is the best learning-free metric and it is only slightly worse than the learned metric, which requires labeled target data. Our method has several advantages compared to the prior art: 1) our method does not require any modification of the training process or the model inputs; and 2) it needs only one forward pass for each model. We evaluate the proposed UQ method on three tasks where our uncertainty quantification yields much stronger correlations with pose estimation errors than the baselines. Moreover, in a real robot grasping task, our method increases the grasping success rate from 35% to 90%.
Using data from the Helioseismic Magnetic Imager, we report on the amplitudes and phase relations of oscillations in quiet-Sun, plage, umbra and the polarity inversion line (PIL) of an active region NOAA$#$11158. We employ Fourier, wavelet and cross correlation spectra analysis. Waves with 5-minute periods are observed in umbra, PIL and plage with common phase values of ${phi}(v,I)=frac{pi}{2}$, ${phi}(v,B_{los})=-frac{pi}{2}$. In addition, ${phi}(I,B_{los})=pi$ in plage are observed. These phase values are consistent with slow standing or fast standing surface sausage wave modes. The line width variations, and their phase relations with intensity and magnetic oscillations, show different values within the plage and PIL regions, which may offer a way to further differentiate wave mode mechanics. Significant Doppler velocity oscillations are present along the PIL, meaning that plasma motion is perpendicular to the magnetic field lines, a signature of Alv`enic waves. A time-distance diagram along a section of the PIL shows Eastward propagating Doppler oscillations converting into magnetic oscillations; the propagation speeds range between 2$-$6 km s$^{-1}$. Lastly, a 3-minute wave is observed in select regions of the umbra in the magnetogram data.
Log in to be able to interact and post comments
comments
Fetching comments
Sign in to be able to follow your search criteria
