Recent Advances in Flares Prediction and Validation
|
|
- Melina Jefferson
- 5 years ago
- Views:
Transcription
1 Recent Advances in Flares Prediction and Validation R. Qahwaji 1, O. Ahmed 1, T. Colak 1, P. Higgins 2, P. Gallagher 2 and S. Bloomfield 2 1 University of Bradford, UK 2 Trinity College Dublin, Ireland r.s.r.qahwaji@bradford.ac.uk
2 Organisation Introduction ASAP SMART-ASAP Evaluation Final Words
3 Automated Flares Prediction Real time performance More feasible as the size of images and datasets increases (The three solar instruments of SDO, AIA, HMI, and EVE, produces about 1-2 TB Data/day exceeding the SOHO data volume by more than a thousand time) Consistent objective performance Ability to associate different data sets Ability to extract knowledge from historical data
4 Challenges for computer-based predictions:? How to represent human s? knowledge and experiences using computerised rules? How to accurately predict flares if the flaring mechanism is still not totally understood? How to make computers learn and become better with time?
5 About this Work Recent advances in solar physics are integrated with recent advances in machine learning and feature selection to study the feasibility of developing the nextgeneration of automated flares prediction technology. This new technology could improve our understanding of the flaring mechanism by highlighting the AR properties that are most related to flare occurrences.
6 Organisation Introduction ASAP SMART-ASAP Evaluation Final Words
7 What is ASAP? Automated Solar Activity Prediction (ASAP) is a web-compliant, fully automated computer to predict solar flares in near real-time. Several solar imaging algorithms and 5 Neural Network Agents. Fully implemented in C++. Available publically at
8 ASAP Features ASAP is always downloading latest images from SOHO MDI (GIF images) and SDO HMI It takes around seconds to detect, group and classify sunspots and then to generate flare predictions. It is written in C++ and can be modified to run under different platforms (currently runs under Windows and Mac)
9 ASAP Modules and Timeline Colak & Qahwaji 2009
10 Automated predictions of solar flares are provided daily in our website
11 ASAP is integrated with SWENET COST ES0803 Alcala 2011
12 ASAP is also integrated with the Integrated Space Weather Analysis System of NASA
13
14
15 Colak T and Qahwaji R (2009): "ASAP: A Hybrid Computer Platform Using Machine Learning and Solar Imaging for Automated Prediction of Significant Solar Flares" Space Weather, 2009
16 Comparison of QR measures for 24, 48, and 72 hour time windows from ASAP and SWPC. SWPC results can be accessed from their website
17 Solar Monitor Active Region Tracker (SMART) Magnetic feature (MF) detections were generated using SMART ( Higgins et al., 2010) SMART is a recently developed feature extraction algorithm to detect, characterise, and catalogue MFs using 96 minute SoHO/MDI images. MFs are detected in magnetograms by segmenting quiet-sun and feature pixels using a combination of image processing techniques. SMART detects MFs automatically and is independent from NOAA s ARs. AR records were generated by SMART [Higgins et al. 2010]. To minimise the error caused by projection effects, only MF detections located within 45 from solar disk centre are considered for this work. Flares catalogues were obtained from the National Geographical Data Centre (NGDC) database. Period: from April Dec 2010
18 MF properties generated by SMART No AR Properties Description 1 Type-Polarity Unipolar/Multipolar 2 Type-Size Big/Small 3 Type-Evolution Emerging/Decaying 4 Area_Mmsq Area of the region [Megameters squared]. 5 Bflux_Mx 6 Bfluxp_Mx 7 Bfluxn_Mx 8 Bfluximb 9 DBfluxDt_Mx 10 Bmin_G Minimum B value in the region [Gauss]. 11 Bmax_G Maximum B value in the region [Gauss]. 12 Bmean_G Mean B value in the region [Gauss]. 13 Lnl_Mm Neutral Line Length in the region [Mega meters]. 14 Lsg_Mm High Gradient Neutral Line Length in the region [Mega meters]. 15 MxGrad_GpMm Maximum Gradient along the Neutral Line [Gauss / Megameter]. 16 MeanGrad Mean Gradient along the Neutral Line [Gauss / Megameter]. 17 MednGrad Median Gradient along the Neutral Line [Gauss / Megameter]. 18 Rval_Mx Schrijver R-Value [Maxwells], (Schrijver, 2007). 19 WLsg_GpMm Falconer WLsg value [Gauss / Megameter], (Falconer et al., 2008). 20 R_Str Schrijver R-Value with a lower threshold for summing flux [Maxwells]. 21 WLsg_Str A modified version of WLsg.
19 Generation of Associated Data Sets Algorithms are developed to associate SMART MF detections with flares from the NGDC catalogue for the period April 1996 till December The purpose of this association is to classify MF detections as flaring or non-flaring. For every MF, the association algorithm determines the times of all C-, M-, and X-class flares in the following 24-hour period. Flare locations provided in the NGDC catalogue are remapped back to the time of the MF detection (Colak and Qahwaji, 2010) and the location compared to the MF spatial coverage. MFs are defined as flaring if the remapped flare location falls within the boundary of the SMART MF contour.
20 Two forms of association used: segmented and operational. Different criteria are used for non-flaring MF detection. In the segmented set, a non-flaring MF is not associated with any C-, M-, or X-class flares in a +/-48-hour period around its observation time, while a flaring MF is associated with a flare within 24 hour period after observation. In the operational set, a non-flaring detection is defined as a MF with no C-, M-, or X-class flare associations in the 24-hour period following its observation. Only MFs, which are located within 45 from the solar disk centre are considered.
21 NGDC Flares Data
22 The total number of flare events and MF detections in the input records (pre-association) during Input Data Total Number of MF Detections Total Number of Flares C M X 663,340 4, Total: 4,915 Total number of flaring and non-flaring MF detections in the segmented and operational sets. Segmented Association Output Flaring MF Detections 27,428 (Within +24 hrs.) Non-Flaring MF Detection 432,930 (Within 48 hrs.) Operational Association Output Flaring MF Detections 27,428 (Within +24 hrs.) Non-Flaring MF Detection 491,259 (Within +24 hrs.)
23 Total Flare Events (with Location) SMART AR Catalogue Input 4,915 Total MF Samples 663,340 C 4,301 M 562 X 52 Flare Catalogue AR Properties Flare Details. No. of Flares Samples No. of AR Samples Associating ARs to Flares Associated Data AR Properties Flare/ No-Flare No. of AR Samples Association Output (Segmented Set) Total MF to Flare Associations 27,428 Total MF to No-Flare Associations 432,930 Total Associated Data 460,358
24 Feature Selection The Feature Selection (FS) process aims to study the significance of the input features (MF properties in our case) with respect to the prediction classes (flaring probability here). FS select features that are highly correlated with the class but uncorrelated with each other. FS enhances the efficiency and usability of a data set by removing features that are irrelevant, redundant, and/or noisy FS is widely used with machine learning to enhance data analysis and prediction. Here, the FS process is coupled with machine learning to highlight the AR properties that are more relevant to flaring/no-flaring. No. of AR Samples Associated Data AR Properties Feature Selection MF Properties that are mostly correlated with Flares/No-Flares Flare/ No-Flare
25 Advantages of the FS process 1. Reduce the number of input features by removing features that are irrelevant, redundant, and/or noisy. 2. Enable machine learning or predictors to be cost effective and faster. 3. Could improve the prediction accuracy of machine learning systems. 4. Provide a physical insight onto the importance of input features.
26 Feature Selection COST ES0803 Alcala 2011 Two different FS algorithms applied: Correlation-based Feature Selection (CFS) and Minimum Redundancy Maximum Relevance (MRMR). We consider all possible combinations of both to create 11 different FS processes.. Initially, cross-validation is applied to the segmented data set (consisting of 21 MF properties). 50% of the data is selected randomly and then FS is applied. This is repeated 20. CFS determines the best (but un-ranked) MF properties while MRMR ranks the most significant MF properties. The predictive capabilities of the selected MFs are compared against the full set (21 MF properties) using CCNN.
27 SMART MF properties selected by each feature selection process. COST ES0803 Alcala 2011 Evaluation Output Search Method Data Type Method Type Selected SMART MF Property IDs CFS BestFirst Backward Normalised Subset v7, v9, v13 CFS BestFirst Bidirectional Normalised Subset v7, v9, v13 CFS BestFirst Forward Normalised Subset v7, v9, v13 CFS Greedy Stepwise Backward Normalised Subset v5, v6, v7, v13, v14, v15, v19, v20, v21 CFS Greedy Stepwise Forward Normalised Subset v7, v9, v13 CFS BestFirst Backward Discretised Subset v9, v13, v14, v18, v19, v20 CFS BestFirst Bidirectional Discretised Subset v9, v13, v14, v18, v19, v20 CFS BestFirst Forward Discretised Subset v9, v13, v14, v18, v19, v20 MRMR-MIQ Forward Discretised Weighted v13, v21, v20, v19, v18, v17, v16, v15, v14, v11 MRMR-FCQ Forward Normalised Weighted v13, v4, v14, v15, v21, v7, v6, v20, v11, v19 MRMR-FCD Forward Normalised Weighted v13, v21, v14, v20, v15, v4, v5, v19, v7, v18 SMART MF properties selected by feature selection processes
28 SMART MF properties most relevant to flaring COST ES0803 Alcala 2011 The SMART MF properties most relevant to flaring Property ID SMART Property Description v4 A Area of the region v5 Total unsigned magnetic flux of the region v6 + Total positive flux in the region v7 Total negative flux in the region v11 B MAX Maximum magnetic field in the region v13 L NL Neutral line length in the region v14 L SG High gradient neutral line length in the region v15 MAX Maximum gradient along the neutral line v18 R Schrijver R value v19 WL SG Falconer WL SG value v20 R * Schrijver R value with a lower threshold v21 WL SG * A modified version of WL SG
29 The Evaluation Strategy Segmented Data Set Operational Data Set Evaluation using CCNN and Cross- Validation Evaluation using CCNN and Timedependant data Evaluation using CCNN and Cross- Validation Evaluation using CCNN and Timedependant data Benchmark Metrics
30 Machine Learning Machine learning enables the creation of computerised learning rules. Cascade Correlation Neural Networks (CCNN) [Qahwaji & Colak, 2007] were applied as our major learning technique. No. of AR Samples Associated Data 21 AR Properties Flare/ NoFlare No. of AR Samples Associated Data 8 AR Properties Flare/ NoFlare Machine learning was applied to the full-set of MF properties (21 properties) and the selected MF properties (12 properties), to compare their prediction performance. CCNN ML Evaluate Prediction Performance CCNN ML Evaluate Prediction Performance Training and testing vectors were organized based on the statistical Jack-Knife technique [Fukunaga 1990]. Compare
31 Machine Learning For all the experiments, 80% of the input samples were randomly selected for training, and the remaining 20% were used for testing. Every experiment was repeated 10 times and the average performance was recorded. CCNN s performance is evaluated by comparing the actual target values with the generated values. The metrics proposed in Fawcett (2006) and Balch (2008) were used. The evaluation strategy is applied on the full MF data-sets (segmented and operational) and FS data sets.
32 Benchmark Metrics: Fawcett (2006) & Balch (2008) TPR=TP/(TP+FN); FPR=FP/(FP+TN); TNR=TN/(TN+FP); FNR=FN/(TP+FN); FAR=FP/(FP+TP); ACC=(TP+TN)/((TP+FN)+(FP+TN)); SPC=TN/(FP+TN); PPV=TP/(TP+FP); NPV=TN/(TN+FN); (Negative Prediction Value) FDR=FP/(FP+TP); (False Discovery Rate) HSS= (2*((TP*TN)-(FP*FN))) / (((TP+FN)*(FN+TN)) + ((TP+FP)*(FP+TN))); (Heidke Skill Score) MCC= ( (TP*TN)-(FP*FN) ) / ((TP+FP)*(TP+FN)*(TN+FP)*(TN+FN)) 2 ; (Mathews Correlation Coefficient)
33 Input using CV Cross-Validation on the Segmented Dataset MSE TPR FPR TNR FNR FAR ACC HSS Cross-Validation on the Operational Dataset Input using CV MSE TPR FPR TNR FNR FAR ACC HSS
34 Train: Segmented Test: Segmented Input/Test Train MSE TPR FPR TNR FNR FAR ACC HSS , , Train: Operational Test: Operational Input/Test Train MSE TPR FPR TNR FNR FAR ACC HSS , , Train: Segmented Test: Operational Input/Test Train MSE TPR FPR TNR FNR FAR ACC HSS , ,
35 Training Data Comparison with ASAP COST ES0803 Alcala 2011 ASAP was trained on a dataset covering a longer time period, it used a different association technique to compare ARs and flares, and did not discard ARs above 45 from the solar centre. It therefore has more data samples in the training and testing sets. Jan1982 Dec2006 (1Feb Dec2002) ASAP Flare Prediction Capability No. of Flaring Testing Data ARs in the TPR ACC FAR MSE HSS Testing Set 1Feb Dec2002 5,
36 Evaluation of FS data sets Benchmark Machine Learning using Cross Validation on the 21 MF Properties MSE TPR FPR TNR FNR FAR ACC HSS Machine Learning using Cross Validation on the Selected MF Properties Group No MSE TPR FPR TNR FNR FAR ACC HSS G G G G
37 Organisation Introduction ASAP SMART-ASAP Evaluation Final Words
38 Early days still, but results so far seems very promising FS not applied to enhance the flaring prediction but to provide better physical insight. SMART and ASAP are both automated. The execution time of SMART is about 20-60, while its about 10 seconds for ASAP on a computer with 2.66 GHz Intel core 2 duo with 2GB of 800MHz DDR2 SDRAM. Hence, both systems can be run integrated to run on realtime.
39 But, Can the prediction performance be improved further (other technologies could be investigated such as HMM, SVM, RBF)? More MFs properties needed? All clear prediction test? SDO s Challenge: ASAP s technology can handle SDO. SMART is to be updated. Both to be tested again!
40 MDI 1K 1K HMI 1K 1K HMI 2K 2K HMI 4K 4K
41
Automated Solar Flare Prediction: Is it a myth?
Automated Solar Flare Prediction: Is it a myth? Tufan Colak, t.colak@bradford.ac.uk Rami Qahwaji, Omar W. Ahmed, Paul Higgins* University of Bradford, U.K.,Trinity Collage Dublin, Ireland* European Space
More informationAutomated Solar Activity Prediction: A hybrid computer platform using machine learning and solar imaging for automated prediction of solar flares
SPACE WEATHER, VOL. 7,, doi:10.1029/2008sw000401, 2009 Automated Solar Activity Prediction: A hybrid computer platform using machine learning and solar imaging for automated prediction of solar flares
More informationThe University of Bradford Institutional Repository
The University of Bradford Institutional Repository http://bradscholars.brad.ac.uk This work is made available online in accordance with publisher policies. Please refer to the repository record for this
More informationAutomated Prediction of Solar Flares Using Neural Networks and Sunspots Associations. EIMC, University of Bradford BD71DP, U.K.
Automated Prediction of Solar Flares Using Neural Networks and Sunspots Associations Tufan Colak t.colak@bradford.ac.uk & Rami Qahwaji r.s.r.qahwaji@bradford.ac.uk EIMC, University of Bradford BD71DP,
More informationDeep Learning Technology for Predicting Solar Flares from (Geostationary Operational Environmental Satellite) Data
Deep Learning Technology for Predicting Solar s from (Geostationary Operational Environmental Satellite) Data Tarek A M Hamad Nagem, Rami Qahwaji, Stan Ipson School of Electrical Engineering and Computer
More informationMachine Learning for Solar Flare Prediction
Machine Learning for Solar Flare Prediction Solar Flare Prediction Continuum or white light image Use Machine Learning to predict solar flare phenomena Magnetogram at surface of Sun Image of solar atmosphere
More informationarxiv: v1 [astro-ph.sr] 22 Aug 2016
Draft version August 24, 2016 Preprint typeset using L A TEX style AASTeX6 v. A COMPARISON OF FLARE FORECASTING METHODS, I: RESULTS FROM THE ALL-CLEAR WORKSHOP G. Barnes and K.D. Leka NWRA, 3380 Mitchell
More informationShort-Term Solar Flare Prediction Using Predictor
Solar Phys (2010) 263: 175 184 DOI 10.1007/s11207-010-9542-3 Short-Term Solar Flare Prediction Using Predictor Teams Xin Huang Daren Yu Qinghua Hu Huaning Wang Yanmei Cui Received: 18 November 2009 / Accepted:
More informationISSI International Team on Mining and exploiting the NASA Solar Dynamics Observatory data in Europe
ISSI International Team on Mining and exploiting the NASA Solar Dynamics Observatory data in Europe Minutes of Meeting ISSI, 30 May -1 June 2012, 3rd meeting Participants: Karel Schrijver, Paul Higgins,
More informationThe Forecasting Solar Particle Events and Flares (FORSPEF) Tool
.. The Forecasting Solar Particle Events and Flares (FORSPEF) Tool A. Anastasiadis 1, I. Sandberg 1, A. Papaioannou 1, M. K. Georgoulis 2, K. Tziotziou 1, G. Tsiropoula 1, D. Paronis 1, P. Jiggens 3, A.
More informationPredicting Solar Flares by Converting GOES X-ray Data to Gramian Angular Fields (GAF) Images
, July 4-6, 2018, London, U.K. Predicting Solar Flares by Converting GOES X-ray Data to Gramian Angular Fields (GAF) Images Tarek Nagem, Rami Qahwaji, Stanley Ipson, Amro Alasta Abstract: Predicting solar
More informationA new technique for the calculation and 3D visualisation of magnetic complexities on solar satellite images
Vis Comput DOI 10.1007/s00371-010-0418-1 ORIGINAL ARTICLE A new technique for the calculation and 3D visualisation of magnetic complexities on solar satellite images O.W. Ahmed R. Qahwaji T. Colak T. Dudok
More informationSide Note on HMI Distortion
Residual Distortion - using Venus transit data of June 5-6, 2012. - level 1 data with distortion removed and PSF corrected - for side camera and only along the path of Venus - least-squares fit of CDELT1
More informationClassification of Hand-Written Digits Using Scattering Convolutional Network
Mid-year Progress Report Classification of Hand-Written Digits Using Scattering Convolutional Network Dongmian Zou Advisor: Professor Radu Balan Co-Advisor: Dr. Maneesh Singh (SRI) Background Overview
More informationQuantitative Forecasting of Major Solar Flares
Quantitative Forecasting of Major Solar Flares Manolis K. Georgoulis 1 & David M. Rust 1 ABSTRACT We define the effective connected magnetic field, B eff, a single metric of the flaring potential in solar
More informationarxiv: v1 [astro-ph.sr] 1 Apr 2015
JOURNAL OF GEOPHYSICAL RESEARCH, VOL.???, XXXX, DOI:10.1029/, Using the Maximum X-ray Flux Ratio and X-ray Background to Predict Solar Flare Class L.M. Winter 1 and K. Balasubramaniam 2 arxiv:1504.00294v1
More informationPerformance Measures. Sören Sonnenburg. Fraunhofer FIRST.IDA, Kekuléstr. 7, Berlin, Germany
Sören Sonnenburg Fraunhofer FIRST.IDA, Kekuléstr. 7, 2489 Berlin, Germany Roadmap: Contingency Table Scores from the Contingency Table Curves from the Contingency Table Discussion Sören Sonnenburg Contingency
More informationEvaluation & Credibility Issues
Evaluation & Credibility Issues What measure should we use? accuracy might not be enough. How reliable are the predicted results? How much should we believe in what was learned? Error on the training data
More informationPrediction of Solar Flare Size and Time-to-Flare Using Support Vector Machine Regression
Prediction of Solar Flare Size and Time-to-Flare Using Support Vector Machine Regression Laura E. Boucheron and Amani Al-Ghraibah arxiv:1511.1941v1 [astro-ph.sr] 5 Nov 215 Klipsch School of Electrical
More informationUsing Deep Learning to Plug an Observational Gap in EUV Irradiance Observations
Using Deep Learning to Plug an Observational Gap in EUV Irradiance Observations Andrés Muñoz-Jaramillo www.solardynamo.org David Fouhey, Richard Galvez, Alexander Szenicer, Rajat Thomas, Paul Wright, Jin
More informationSolar Cycle Propagation, Memory, and Prediction Insights from a Century of Magnetic Proxies
Solar Cycle Propagation, Memory, and Prediction Insights from a Century of Magnetic Proxies Neil R. Sheeley Jr. Jie Zhang Andrés Muñoz-Jaramillo Edward E. DeLuca Work performed in collaboration with: Maria
More informationIdentification of photospheric activity features from SOHO/MDI data using the ASAP tool
Identification of photospheric activity features from SOHO/MDI data using the ASAP tool Omar Ashamari, Rami Qahwaji, Stan Ipson, Micha Scholl, Omar Nibouche, Margit Haberreiter To cite this version: Omar
More informationLinear Methods for Regression. Lijun Zhang
Linear Methods for Regression Lijun Zhang zlj@nju.edu.cn http://cs.nju.edu.cn/zlj Outline Introduction Linear Regression Models and Least Squares Subset Selection Shrinkage Methods Methods Using Derived
More informationREPORT DOCUMENTATION PAGE Form Approved OMB No
REPORT DOCUMENTATION PAGE Form Approved OMB No. 0704-0188 Public reporting burden for this collection of information is estimated to average 1 hour per response, including the time for reviewing instructions,
More informationEvaluation Metrics for Intrusion Detection Systems - A Study
Evaluation Metrics for Intrusion Detection Systems - A Study Gulshan Kumar Assistant Professor, Shaheed Bhagat Singh State Technical Campus, Ferozepur (Punjab)-India 152004 Email: gulshanahuja@gmail.com
More informationSolar Flare Durations
Solar Flare Durations Whitham D. Reeve 1. Introduction Scientific investigation of solar flares is an ongoing pursuit by researchers around the world. Flares are described by their intensity, duration
More informationSolar EUV Spectral Irradiance: Measurements. Frank Eparvier
Solar EUV Spectral Irradiance: Measurements Frank Eparvier eparvier@colorado.edu Outline Introduction to Solar EUV Irradiance TIMED-SEE and SDO-EVE New Insights into EUV Sun from EVE The Future of EUV
More informationTime-Distance Imaging of Solar Far-Side Active Regions
Time-Distance Imaging of Solar Far-Side Active Regions Junwei Zhao W. W. Hansen Experimental Physics Laboratory, Stanford University, Stanford, CA94305-4085 ABSTRACT It is of great importance to monitor
More informationKanzelhöhe Observatory Flare observations and real-time detections Astrid Veronig & Werner Pötzi
Kanzelhöhe Observatory Flare observations and real-time detections Astrid Veronig & Werner Pötzi Kanzelhöhe Observatory / Institute of Physics University of Graz, Austria Institute of Physics Kanzelhöhe
More informationSpace Weather Prediction using Soft Computing Techniques
Space Weather Prediction using Soft Computing Techniques José Manuel Costa Casas Fernandes jose.c.casas.fernandes@tecnico.ulisboa.pt Instituto Superior Técnico, Lisboa, Portugal June 2015 Abstract Nowadays,
More informationTime Distance Study of Isolated Sunspots
Astron. Nachr. / AN 999, No. 88, 789 792 (2006) / DOI please set DOI! Time Distance Study of Isolated Sunspots Sergei Zharkov, Christopher J. Nicholas, and Michael J. Thompson Solar Physics and upper-atmosphere
More informationDevelopment and Evaluation of an Automated System for Empirical Forecasting of Flares
Development and Evaluation of an Automated System for Empirical Forecasting of Flares Sung-Hong Park1 Collaborators: J. Baek2, S. Bong2, Y. Yuan3, J. Jing3 1 National Observatory of Athens (NOA) 2 Korea
More information4+ YEARS OF SCIENTIFIC RESULTS WITH SDO/HMI
4+ YEARS OF SCIENTIFIC RESULTS WITH SDO/HMI Sebastien Couvidat and the HMI team Solar Metrology Symposium, October 2014 The HMI Instrument HMI Science Goals Evidence of Double-Cell Meridional Circulation
More informationStyle-aware Mid-level Representation for Discovering Visual Connections in Space and Time
Style-aware Mid-level Representation for Discovering Visual Connections in Space and Time Experiment presentation for CS3710:Visual Recognition Presenter: Zitao Liu University of Pittsburgh ztliu@cs.pitt.edu
More informationMonthly Geomagnetic Bulletin
HARTLAND OBSERVATORY Monthly Geomagnetic Bulletin BRISTOL CHANNEL December 2002 02/12/HA Hartland NERC 2002 1. HARTLAND OBSERVATORY MAGNETIC DATA 1.1 Introduction This bulletin is published to meet the
More informationFlare Forecasting Using the Evolution of McIntosh Sunspot Classifications
submitted to Journal of Space Weather and Space Climate c The author(s) under the Creative Commons Attribution License Flare Forecasting Using the Evolution of McIntosh Sunspot Classifications A. E. McCloskey
More informationFORECAST OF SOLAR PROTON EVENTS WITH NOAA SCALES BASED ON SOLAR X-RAY FLARE DATA USING NEURAL NETWORK
Journal of the Korean Astronomical Society http://dx.doi.org/10.5303/jkas.2014.47.6.209 47: 209 214, 2014 December pissn: 1225-4614 eissn: 2288-890X c 2014. The Korean Astronomical Society. All rights
More informationThe information you need will be on the internet. Please label your data with the link you used, in case we need to look at the data again.
Solar Activity in Many Wavelengths In this lab you will be finding the sidereal rotation period of the Sun from observations of sunspots, you will compare the lifetimes of larger and smaller sunspots,
More informationAnomaly Detection and Categorization Using Unsupervised Deep Learning
Anomaly Detection and Categorization Using Unsupervised S6340 Thursday 7 th April 2016 GPU Technology Conference A. Stephen McGough, Noura Al Moubayed, Jonathan Cumming, Eduardo Cabrera, Peter Matthews,
More informationMachine Learning to Automatically Detect Human Development from Satellite Imagery
Technical Disclosure Commons Defensive Publications Series April 24, 2017 Machine Learning to Automatically Detect Human Development from Satellite Imagery Matthew Manolides Follow this and additional
More informationNew COST Action: Towards a European Network on Chemical Weather Forecasting and Information Systems
New COST Action: Towards a European Network on Chemical Weather Forecasting and Information Systems Proposer: Mikhail Sofiev Finnish Meteorological Institute Historical background EUMETNET Workshop on
More informationVariable Selection in Data Mining Project
Variable Selection Variable Selection in Data Mining Project Gilles Godbout IFT 6266 - Algorithmes d Apprentissage Session Project Dept. Informatique et Recherche Opérationnelle Université de Montréal
More informationCISM Model Timeline, Years 6-10
Solar-Heliosphere Models 1 Black = capabilities to incoporate in coupled model Green = evaluations, test CORHEL 3.4 MAS,, CONE models. Inputs from full set of observatories (NSO/KP, Wilcox, Mt. Wilson,
More informationSolar Activity during the Rising Phase of Solar Cycle 24
International Journal of Astronomy and Astrophysics, 213, 3, 212-216 http://dx.doi.org/1.4236/ijaa.213.3325 Published Online September 213 (http://www.scirp.org/journal/ijaa) Solar Activity during the
More informationAUTOMATED SUNSPOT DETECTION AND CLASSIFICATION USING SOHO/MDI IMAGERY THESIS. Samantha R. Howard, 1st Lieutenant, USAF AFIT-ENP-MS-15-M-078
AUTOMATED SUNSPOT DETECTION AND CLASSIFICATION USING SOHO/MDI IMAGERY THESIS Samantha R. Howard, 1st Lieutenant, USAF AFIT-ENP-MS-15-M-078 DEPARTMENT OF THE AIR FORCE AIR UNIVERSITY AIR FORCE INSTITUTE
More informationImproved Kalman Filter Initialisation using Neurofuzzy Estimation
Improved Kalman Filter Initialisation using Neurofuzzy Estimation J. M. Roberts, D. J. Mills, D. Charnley and C. J. Harris Introduction It is traditional to initialise Kalman filters and extended Kalman
More informationAIA DATA ANALYSIS OVERVIEW OF THE AIA INSTRUMENT
AIA DATA ANALYSIS OVERVIEW OF THE AIA INSTRUMENT SDO SUMMER SCHOOL ~ August 2010 ~ Yunnan, China Marc DeRosa (LMSAL) ~ derosa@lmsal.com WHAT IS SDO? The goal of Solar Dynamics Observatory (SDO) is to understand:
More informationDM-Group Meeting. Subhodip Biswas 10/16/2014
DM-Group Meeting Subhodip Biswas 10/16/2014 Papers to be discussed 1. Crowdsourcing Land Use Maps via Twitter Vanessa Frias-Martinez and Enrique Frias-Martinez in KDD 2014 2. Tracking Climate Change Opinions
More informationMachine Learning Concepts in Chemoinformatics
Machine Learning Concepts in Chemoinformatics Martin Vogt B-IT Life Science Informatics Rheinische Friedrich-Wilhelms-Universität Bonn BigChem Winter School 2017 25. October Data Mining in Chemoinformatics
More informationNumerical processing of sunspot images using the digitized Royal Greenwich Observatory Archive
Numerical processing of sunspot images using the digitized Royal Greenwich Observatory Archive Andrey Tlatov 1 and Vladimir Ershov 2 1 Kislovodsk Mountain Solar Station,Pulkovo observatory 2 Mullard Space
More informationActivities of NOAA s NWS Climate Prediction Center (CPC)
Activities of NOAA s NWS Climate Prediction Center (CPC) Jon Gottschalck and Dave DeWitt Improving Sub-Seasonal and Seasonal Precipitation Forecasting for Drought Preparedness May 27-29, 2015 San Diego,
More informationLearning theory. Ensemble methods. Boosting. Boosting: history
Learning theory Probability distribution P over X {0, 1}; let (X, Y ) P. We get S := {(x i, y i )} n i=1, an iid sample from P. Ensemble methods Goal: Fix ɛ, δ (0, 1). With probability at least 1 δ (over
More informationPredicting New Search-Query Cluster Volume
Predicting New Search-Query Cluster Volume Jacob Sisk, Cory Barr December 14, 2007 1 Problem Statement Search engines allow people to find information important to them, and search engine companies derive
More informationForecasting Solar Flare Events: A Critical Review
Forecasting Solar Flare Events: A Critical Review K D Leka NorthWest Research Associates (NWRA) Boulder, CO, USA Acknowledging funding from AFOSR, NASA, and NOAA. Why Important? Time-of-flight = c Space
More informationObserving Dark Worlds (Final Report)
Observing Dark Worlds (Final Report) Bingrui Joel Li (0009) Abstract Dark matter is hypothesized to account for a large proportion of the universe s total mass. It does not emit or absorb light, making
More informationrising phase of solar cycle 24 based on
11th European Space Weather Week, November 17-21 2014, Liège, Belgium Analysis of solar wind sources during the rising phase of solar cycle 24 based on the AIA/SDO EUV images Yulia Shugay 1 and Vladimir
More informationmichele piana dipartimento di matematica, universita di genova cnr spin, genova
michele piana dipartimento di matematica, universita di genova cnr spin, genova first question why so many space instruments since we may have telescopes on earth? atmospheric blurring if you want to
More informationPerformance Evaluation
Performance Evaluation David S. Rosenberg Bloomberg ML EDU October 26, 2017 David S. Rosenberg (Bloomberg ML EDU) October 26, 2017 1 / 36 Baseline Models David S. Rosenberg (Bloomberg ML EDU) October 26,
More informationSTCE Newsletter. 6 Jan Jan 2014
Published by the STCE - this issue : 16 Jan 2014. Available online at http://www.stce.be/newsletter/. The Solar-Terrestrial Centre of Excellence (STCE) is a collaborative network of the Belgian Institute
More informationLeast Mean Squares Regression
Least Mean Squares Regression Machine Learning Spring 2018 The slides are mainly from Vivek Srikumar 1 Lecture Overview Linear classifiers What functions do linear classifiers express? Least Squares Method
More informationStatistical Prediction of Solar Flares Using Line of Sight (LOS) Magnetogram Data
Statistical Prediction of Solar Flares Using Line of Sight (LOS) Magnetogram Data Jacinda Knoll Mentors: K.D. Leka and Graham Barnes Outline Importance of Solar Flare Prediction Data and Method Used Special
More informationForecas(ng the Magne(c Field Configura(on of CMEs
Volker Bothmer University of Göttingen Institute for Astrophysics 26 October 2015 ISEST Workshop, UNAM, Mexico City Forecas(ng the Magne(c Field Configura(on of CMEs Outline 1. Magnetic field configuration
More informationAbout Nnergix +2, More than 2,5 GW forecasted. Forecasting in 5 countries. 4 predictive technologies. More than power facilities
About Nnergix +2,5 5 4 +20.000 More than 2,5 GW forecasted Forecasting in 5 countries 4 predictive technologies More than 20.000 power facilities Nnergix s Timeline 2012 First Solar Photovoltaic energy
More informationEvaluation requires to define performance measures to be optimized
Evaluation Basic concepts Evaluation requires to define performance measures to be optimized Performance of learning algorithms cannot be evaluated on entire domain (generalization error) approximation
More informationPredictive Analytics on Accident Data Using Rule Based and Discriminative Classifiers
Advances in Computational Sciences and Technology ISSN 0973-6107 Volume 10, Number 3 (2017) pp. 461-469 Research India Publications http://www.ripublication.com Predictive Analytics on Accident Data Using
More informationarxiv: v1 [astro-ph.sr] 8 Dec 2016
Research in Astron. Astrophys. Vol. No. XX, 000 000 http://www.raa-journal.org http://www.iop.org/journals/raa (L A TEX: BMRs.tex; printed on December 9, 2016; 1:17) Research in Astronomy and Astrophysics
More informationAutomatic vs. Human detection of Bipolar Magnetic Regions: Using the best of both worlds. Michael DeLuca Adviser: Dr. Andrés Muñoz-Jaramillo
Automatic vs. Human detection of Bipolar Magnetic Regions: Using the best of both worlds Michael DeLuca Adviser: Dr. Andrés Muñoz-Jaramillo Magnetic fields emerge from the Sun Pairs of positive and negative
More informationAnalysis on selected geo-effective events using observations and models at Space Environment Prediction Center
Analysis on selected geo-effective events using observations and models at Space Environment Prediction Center Siqing Liu, Ercha Aa, Qiuzhen Zhong, Bingxian Luo, Zhitao Li, Jingjing Wang, and Jiancun Gong
More informationAUTOMATED MACHINE LEARNING-BASED PREDICTION OF CMES BASED ON FLARES ASSOCIATIONS
AUTOMATED MACHINE LEARNING-BASED PREDICTION OF CMES BASED ON FLARES ASSOCIATIONS R. QAHWAJI, T. COLAK, M. AL-OMARI, and S. IPSON Department of Electronic Imaging and Media Communications University of
More informationAUTOMATIC PREDICTION OF SOLAR FLARES USING A NEURAL NETWORK. James Negus University of Chicago Mentor: Andrew Jones LASP
AUTOMATIC PREDICTION OF SOLAR FLARES USING A NEURAL NETWORK James Negus University of Chicago Mentor: Andrew Jones LASP SOLAR FLARE A flare is defined as a sudden, rapid, and intense variation in brightness.
More informationCoronal Holes. Detection in STEREO/EUVI and SDO/AIA data and comparison to a PFSS model. Elizabeth M. Dahlburg
Coronal Holes Detection in STEREO/EUVI and SDO/AIA data and comparison to a PFSS model Elizabeth M. Dahlburg Montana State University Solar Physics REU 2011 August 3, 2011 Outline Background Coronal Holes
More informationAn Automatic Segmentation Algorithm for Solar Filaments in H-Alpha Images using a Context-based Sliding Window
An Automatic Segmentation Algorithm for Solar Filaments in H-Alpha Images using a Context-based Sliding Window Ibrahim A. Atoum College of Applied Sciences Al Maarefa Colleges for Science and Technology
More informationMAGNETOGRAM MEASURES OF TOTAL NONPOTENTIALITY FOR PREDICTION OF SOLAR CORONAL MASS EJECTIONS FROM ACTIVE REGIONS OF ANY DEGREE OF MAGNETIC COMPLEXITY
The Astrophysical Journal, 689:1433 1442, 2008 December 20 # 2008. The American Astronomical Society. All rights reserved. Printed in U.S.A. MAGNETOGRAM MEASURES OF TOTAL NONPOTENTIALITY FOR PREDICTION
More informationComparison of Predicted and Actual Rail Temperature. Matthew Dick, P.E. Radim Bruzek ENSCO Rail May 5, 2016
Comparison of Predicted and Actual Rail Temperature Matthew Dick, P.E. Radim Bruzek ENSCO Rail May 5, 2016 1 Executive Summary A rail temperature prediction system was deployed by ENSCO within a program
More informationSTCE Newsletter. 28 Dec Jan 2016
Published by the STCE - this issue : 8 Jan 2016. Available online at http://www.stce.be/newsletter/. The Solar-Terrestrial Centre of Excellence (STCE) is a collaborative network of the Belgian Institute
More informationStatistical properties of the Bipolar Magnetic Regions
Research in Astron. Astrophys. Vol. No. XX, 000 000 http://www.raa-journal.org http://www.iop.org/journals/raa Research in Astronomy and Astrophysics Statistical properties of the Bipolar Magnetic Regions
More informationFARSIDE HELIOSEISMIC HOLOGRAPHY: RECENT ADVANCES
FARSIDE HELIOSEISMIC HOLOGRAPHY: RECENT ADVANCES I. González Hernández 1, F. Hill 1, C. Lindsey 2, D. Braun 2, P. Scherrer 3, and S.M. Hanasoge 3 1 National Solar Observatory, Tucson, Arizona, USA 2 NorthWest
More informationThe COST733class circulation type software: An example for surface ozone concentrations in Central Europe. Demuzere, M., Kassomenos, P., Philipp, A.
The COST733class circulation type software: An example for surface ozone concentrations in Central Europe Demuzere, M., Kassomenos, P., Philipp, A. Goal of this work To provide insight the functionalities
More informationUnderstanding Solar Indices
Understanding Solar Indices By Ken Larson KJ6RZ Long distance HF radio communications is made possible by a region of charged particles in the Earth s upper atmosphere, 30 to 200 miles above the Earth
More informationFirst European Space Weather Week. Space weather - atmospheres, drag, global change future needs. 29 November-3 December 2004
First European Space Weather Week Space weather - atmospheres, drag, global change future needs 29 November-3 December 2004 Timescales of important phenomena Weather Climate No single statement of requirement
More informationVerification of Space Weather Forecasts issued by the Met Office Space Weather Operations Centre
Verification of Space Weather Forecasts issued by the Met Office Space Weather Operations Centre M. A. Sharpe 1, S. A. Murray 2 1 Met Office, UK. 2 Trinity College Dublin, Ireland. (michael.sharpe@metoffice.gov.uk)
More informationMachine Learning for OR & FE
Machine Learning for OR & FE Regression II: Regularization and Shrinkage Methods Martin Haugh Department of Industrial Engineering and Operations Research Columbia University Email: martin.b.haugh@gmail.com
More informationSparse Approximation and Variable Selection
Sparse Approximation and Variable Selection Lorenzo Rosasco 9.520 Class 07 February 26, 2007 About this class Goal To introduce the problem of variable selection, discuss its connection to sparse approximation
More informationVirtual Beach Building a GBM Model
Virtual Beach 3.0.6 Building a GBM Model Building, Evaluating and Validating Anytime Nowcast Models In this module you will learn how to: A. Build and evaluate an anytime GBM model B. Optimize a GBM model
More informationSTCE Newsletter. 7 Dec Dec 2015
Published by the STCE - this issue : 18 Dec 2015. Available online at http://www.stce.be/newsletter/. The Solar-Terrestrial Centre of Excellence (STCE) is a collaborative network of the Belgian Institute
More informationAutomatic Classification of Sunspot Groups Using SOHO/MDI Magnetogram and White Light Images
Automatic Classification of Sunspot Groups Using SOHO/MDI Magnetogram and White Light Images David Stenning March 29, 2011 1 Introduction and Motivation Although solar data is being generated at an unprecedented
More informationUsing This Flip Chart
Using This Flip Chart Sunspots are the first indicators that a storm from the Sun is a possibility. However, not all sunspots cause problems for Earth. By following the steps in this flip chart you will
More informationProbing Magnetic Fields in the Solar Convection Zone with Meridional Flow
Probing Magnetic Fields in the Solar Convection Zone with Meridional Flow Zhi-Chao Liang 22 Dec. 2015, MPS, Göttingen Liang, Z.-C. & Chou, D.-Y., 2015, Probing Magnetic Fields at the Base of the Solar
More informationThis wind energy forecasting capability relies on an automated, desktop PC-based system which uses the Eta forecast model as the primary input.
A Simple Method of Forecasting Wind Energy Production at a Complex Terrain Site: An Experiment in Forecasting Using Historical Data Lubitz, W. David and White, Bruce R. Department of Mechanical & Aeronautical
More informationModel Accuracy Measures
Model Accuracy Measures Master in Bioinformatics UPF 2017-2018 Eduardo Eyras Computational Genomics Pompeu Fabra University - ICREA Barcelona, Spain Variables What we can measure (attributes) Hypotheses
More informationForecasting demand in the National Electricity Market. October 2017
Forecasting demand in the National Electricity Market October 2017 Agenda Trends in the National Electricity Market A review of AEMO s forecasting methods Long short-term memory (LSTM) neural networks
More informationPLANETARY DEFENSE: RADAR 3D SHAPE MODELING. NASA FDL OVERVIEW- DR LIKA GUHATHAKURTA (on behalf of the FDL team.)
PLANETARY DEFENSE: RADAR 3D SHAPE MODELING NASA FDL OVERVIEW- DR LIKA GUHATHAKURTA (on behalf of the FDL team.) An AI R&D accelerator that tackles knowledge gaps useful to the space program. The program
More informationDetermination of Linear Force- Free Magnetic Field Constant αα Using Deep Learning
Determination of Linear Force- Free Magnetic Field Constant αα Using Deep Learning Bernard Benson, Zhuocheng Jiang, W. David Pan Dept. of Electrical and Computer Engineering (Dept. of ECE) G. Allen Gary
More informationSNOW COVER MAPPING USING METOP/AVHRR AND MSG/SEVIRI
SNOW COVER MAPPING USING METOP/AVHRR AND MSG/SEVIRI Niilo Siljamo, Markku Suomalainen, Otto Hyvärinen Finnish Meteorological Institute, P.O.Box 503, FI-00101 Helsinki, Finland Abstract Weather and meteorological
More informationSupplementary Figure 1. Summer mesoscale convective systems rainfall climatology and trends. Mesoscale convective system (MCS) (a) mean total
Supplementary Figure 1. Summer mesoscale convective systems rainfall climatology and trends. Mesoscale convective system (MCS) (a) mean total rainfall and (b) total rainfall trend from 1979-2014. Total
More informationA New Equatorial Plasma Bubble Prediction Capability
A New Equatorial Plasma Bubble Prediction Capability Brett A. Carter Institute for Scientific Research, Boston College, USA, http://www.bc.edu/research/isr/, RMIT University, Australia, www.rmit.edu.au/space
More informationExploiting the Magnetic Origin of Solar Activity in Forecasting Thermospheric Density Variations
Exploiting the Magnetic Origin of Solar Activity in Forecasting Thermospheric Density Variations Harry Warren Naval Research Laboratory, Space Science Division, Washington, DC John Emmert Naval Research
More informationAngelika Dehn Rob Koopman 10 Years GOME on ERS-2 Workshop
Angelika Dehn (ADehn@serco.it), Rob Koopman (Rob.Koopman@esa.int), Overview I. ERS-2 Mission History 1. Mission Plan Highlights 2. GOME Special Operations II. GOME-1 Engineering Performance 1. Routine
More informationHierarchical models for the rainfall forecast DATA MINING APPROACH
Hierarchical models for the rainfall forecast DATA MINING APPROACH Thanh-Nghi Do dtnghi@cit.ctu.edu.vn June - 2014 Introduction Problem large scale GCM small scale models Aim Statistical downscaling local
More informationOPSIAL Manual. v Xiaofeng Tan. All Rights Reserved
OPSIAL Manual v1.0 2016 Xiaofeng Tan. All Rights Reserved 1. Introduction... 3 1.1 Spectral Calculator & Fitter (SCF)... 3 1.2 Automated Analyzer (AA)... 3 2. Working Principles and Workflows of OPSIAL...
More information