• Journal of Astronomy and Space Sciences
• /
• v.26 no.3
• /
• pp.305-316
• /
• 2009

In this study, we suggest an empirical forecast of CIR (Corotating Interaction Regions) and geomagnetic storm based on the information of coronal holes (CH). For this we used CH data obtained from He I $10830{\AA}$ maps at National Solar Observatory-Kitt Peak from January 1996 to November 2003 and the CIR and storm data that Choi et al. (2009) identified. Considering the relationship among coronal holes, CIRs, and geomagnetic storms (Choi et al. 2009), we propose the criteria for geoeffective coronal holes; the center of CH is located between $N40^{\circ}$ and $S40^{\circ}$ and between $E40^{\circ}$ and $W20^{\circ}$, and its area in percentage of solar hemispheric area is larger than the following areas: (1) case 1: 0.36%, (2) case 2: 0.66%, (3) case 3: 0.36% for 1996-2000, and 0.66% for 2001-2003. Then we present contingency tables between prediction and observation for three cases and their dependence on solar cycle phase. From the contingency tables, we determined several statistical parameters for forecast evaluation such as PODy (the probability of detection yes), FAR (the false alarm ratio), Bias (the ratio of "yes" predictions to "yes" observations) and CSI (critical success index). Considering the importance of PODy and CSI, we found that the best criterion is case 3; CH-CIR: PODy=0.77, FAR=0.66, Bias=2.28, CSI=0.30. CH-storm: PODy=0.81, FAR=0.84, Bias=5.00, CSI=0.16. It is also found that the parameters after the solar maximum are much better than those before the solar maximum. Our results show that the forecasting of CIR based on coronal hole information is meaningful but the forecast of goemagnetic storm is challenging.

• The Bulletin of The Korean Astronomical Society
• /
• v.37 no.2
• /
• pp.123.1-123.1
• /
• 2012

We have compared near-real time Kp forecast models based on neural network (NN) and support vector machine (SVM) algorithms. We consider four models as follows: (1) a NN model using ACE solar wind data; (2) a SVM model using ACE solar wind data; (3) a NN model using ACE solar wind data and preliminary kp values from US ground-based magnetometers; (4) a SVM model using the same input data as model 3. For the comparison of these models, we estimate correlation coefficients and RMS errors between the observed Kp and the predicted Kp. As a result, we found that the model 3 is better than the other models. The values of correlation coefficients and RMS error of the model 3 are 0.93 and 0.48, respectively. For the forecast evaluation of models for geomagnetic storms ($Kp{\geq}6$), we present contingency tables and estimate statistical parameters such as probability of detection yes (PODy), false alarm ratio (FAR), bias, and critical success index (CSI). From a comparison of these statistical parameters, we found that the SVM models (model 2 and model 4) are better than the NN models (model 1 and model 3). The values of PODy and CSI of the model 4 are the highest among these models (PODy: 0.57 and CSI: 0.48). From these results, we suggest that the NN models are better than the SVM models for predicting Kp and the SVM models are better than the NN models for forecasting geomagnetic storms.

• Journal of The Korean Astronomical Society
• /
• v.47 no.6
• /
• pp.209-214
• /
• 2014

In this study we develop a set of solar proton event (SPE) forecast models with NOAA scales by Multi Layer Perceptron (MLP), one of neural network methods, using GOES solar X-ray flare data from 1976 to 2011. Our MLP models are the first attempt to forecast the SPE scales by the neural network method. The combinations of X-ray flare class, impulsive time, and location are used for input data. For this study we make a number of trials by changing the number of layers and nodes as well as combinations of the input data. To find the best model, we use the summation of F-scores weighted by SPE scales, where F-score is the harmonic mean of PODy (recall) and precision (positive predictive value), in order to minimize both misses and false alarms. We find that the MLP models are much better than the multiple linear regression model and one layer MLP model gives the best result.

• The Bulletin of The Korean Astronomical Society
• /
• v.38 no.1
• /
• pp.61.1-61.1
• /
• 2013

In this study we apply Support Vector Machine (SVM) to the prediction of geo-effective halo coronal mass ejections (CMEs). The SVM, which is one of machine learning algorithms, is used for the purpose of classification and regression analysis. We use halo and partial halo CMEs from January 1996 to April 2010 in the SOHO/LASCO CME Catalog for training and prediction. And we also use their associated X-ray flare classes to identify front-side halo CMEs (stronger than B1 class), and the Dst index to determine geo-effective halo CMEs (stronger than -50 nT). The combinations of the speed and the angular width of CMEs, and their associated X-ray classes are used for input features of the SVM. We make an attempt to find the best model by using cross-validation which is processed by changing kernel functions of the SVM and their parameters. As a result we obtain statistical parameters for the best model by using the speed of CME and its associated X-ray flare class as input features of the SVM: Accuracy=0.66, PODy=0.76, PODn=0.49, FAR=0.72, Bias=1.06, CSI=0.59, TSS=0.25. The performance of the statistical parameters by applying the SVM is much better than those from the simple classifications based on constant classifiers.

• Asian Journal of Atmospheric Environment
• /
• v.12 no.1
• /
• pp.1-16
• /
• 2018

Ground-level ozone ($O_3$) can be a menace for vegetation, especially in Asia where $O_3$ levels have been dramatically increased over the past decades. To ensure food security and maintain forest ecosystem services, such as nutrient cycling, carbon sequestration and functional diversity of soil biota, in the over-populated Asia, environmental standards are needed. To set proper standards, dose-response relationships should be established from which critical levels are derived. The predictor of the response in the dose-response relationship is an $O_3$ metric that indicates the dose level to which the plant has been exposed. This study aimed to review the relevant scientific literature and summarize the $O_3$ metrics used worldwide to provide insights for Asia. A variety of $O_3$ metrics have been used, for which we discuss their strengths and weaknesses. The most widely used metrics are based only on $O_3$ levels. Such metrics have been adopted by several regulatory agencies in the global. However, they are biologically irrelevant because they ignore the plant physiological capacity. Adopting AOT40 ($O_3$ mixing ratios Accumulated Over the Threshold of $40nmol\;mol^{-1}$) as the default index for setting critical levels in Asia would be a poor policy with severe consequences at national and Pan-Asian level. Asian studies should focus on flux-based $O_3$ metrics to provide relevant bases for developing proper standards. However, given the technical requirements in calculating flux-based $O_3$ metrics, which can be an important limitation in developing countries, no-threshold cumulative exposure indices like AOT0 should always accompany flux-based indices.

• Journal of The Korean Astronomical Society
• /
• v.45 no.2
• /
• pp.31-38
• /
• 2012

In this study we apply Support Vector Machine (SVM) to the prediction of geo-effective halo coronal mass ejections (CMEs). The SVM, which is one of machine learning algorithms, is used for the purpose of classification and regression analysis. We use halo and partial halo CMEs from January 1996 to April 2010 in the SOHO/LASCO CME Catalog for training and prediction. And we also use their associated X-ray flare classes to identify front-side halo CMEs (stronger than B1 class), and the Dst index to determine geo-effective halo CMEs (stronger than -50 nT). The combinations of the speed and the angular width of CMEs, and their associated X-ray classes are used for input features of the SVM. We make an attempt to find the best model by using cross-validation which is processed by changing kernel functions of the SVM and their parameters. As a result we obtain statistical parameters for the best model by using the speed of CME and its associated X-ray flare class as input features of the SVM: Accuracy=0.66, PODy=0.76, PODn=0.49, FAR=0.72, Bias=1.06, CSI=0.59, TSS=0.25. The performance of the statistical parameters by applying the SVM is much better than those from the simple classifications based on constant classifiers.

• Atmosphere
• /
• v.27 no.2
• /
• pp.119-131
• /
• 2017

We present and discuss the Tropopause Folding Turbulence Detection (TFTD) algorithm for the Korean Communication, Ocean, Meteorological Satellite (COMS) which is originally developed for the Tropopause Folding Turbulence Product (TFTP) from the Geostationary Operational Environmental Satellite (GOES)-R. The TFTD algorithm assumes that the tropopause folding is linked to the Clear Air Turbulence (CAT), and thereby the tropopause folding areas are detected from the rapid spatial gradients of the upper tropospheric specific humidity. The Layer Averaged Specific Humidity (LASH) is used to represent the upper tropospheric specific humidity calculated using COMS $6.7{\mu}m$ water vapor channel and ERA-interim reanalysis temperature at 300, 400, and 500 hPa. The comparison of LASH with the numerical model specific humidity shows a strong negative correlation of 80% or more. We apply the single threshold, which is determined from sensitivity analysis, for cloud-clearing to overcome strong gradient of LASH at the edge of clouds. The tropopause break lines are detected from the location of strong LASH-gradient using the Canny edge detection based on the image processing technique. The tropopause folding area is defined by expanding the break lines by 2-degree positive gradient direction. The validations of COMS TFTD is performed with Pilot Reports (PIREPs) filtered out Convective Induced Turbulence (CIT) from Dec 2013 to Nov 2014 over the South Korea. The score test shows 0.49 PODy (Probability of Detection 'Yes') and 0.64 PODn (Probability of Detection 'No'). Low POD results from various kinds of CAT reported from PIREPs and the characteristics of high sensitivity in edge detection algorithm.