Forecasting the Earth's Trapped Particle Distribution Using Hierarchical Bayesian Spatio Temporal Model

2014 International Conference on Science & Engineering in Mathematics, Chemistry and Physics

数学;化学;物理学

Suparta, Wayan^1 ; Gusrizal^1

Institute of Space Sciences (ANGKASA), Universiti Kebangsaan Malaysia, 43600 Bangi, Selangor Darul Ehsan, Malaysia^1

关键词: Hierarchical bayesian;    Mean absolute error;    Mean absolute percentage error;    Monte Carlo Markov chain;    Root mean square errors;    South atlantic anomalies;    Spatio-temporal models;    Statistical validation;   

Others  :  https://iopscience.iop.org/article/10.1088/1742-6596/495/1/012039/pdf
DOI  :  10.1088/1742-6596/495/1/012039

来源: IOP

【 摘 要 】

We employed the Hierarchical Bayesian spatio temporal (HBST) Gaussian Process (GP) model for forecasting the distribution of the Earth's trapped particle. The model was applied in the South Atlantic Anomaly (SAA) region. Data from 1-30 January 2000 of >30 keV electron flux acquired by National Oceanic and Atmospheric (NOAA) 15 satellite was carried to model. The purpose was to forecast the flux value on 31 January 2000. Gridding process of 10x10 lot-lan was performed after cleaning and log transforming data. The HBST GP model was undertaken by implementing the Monte Carlo Markov Chain (MCMC) method. The forecasting result was interpolated by using Kriging technique to draw the distribution map of particle flux. Statistical validation represented by mean square error, root mean square error, mean absolute error, mean absolute percentage error, bias, relative bias, and mean relative separation shows good indicators. The visual validation also figured a quite similarity with NOAA's map that the model capable to forecast the particle flux.