Open AccessTwo empirical models for short‐term forecast of Kp
Author(s) -
Luo B.,
Liu S.,
Gong J.
Publication year - 2017
Publication title -
space weather
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.254
H-Index - 56
ISSN - 1542-7390
DOI - 10.1002/2016sw001585
Subject(s) - term (time) , mean squared error , solar wind , magnetosphere , geomagnetic storm , meteorology , correlation coefficient , empirical modelling , earth's magnetic field , mathematics , forecast error , statistics , environmental science , econometrics , computer science , physics , simulation , plasma , quantum mechanics , magnetic field
In this paper, two empirical models are developed for short‐term forecast of the K p index, taking advantage of solar wind‐magnetosphere coupling functions proposed by the research community. Both models are based on the data for years 1995 to 2004. Model 1 mainly uses solar wind parameters as the inputs, while model 2 also utilizes the previous measured K p value. Finally, model 1 predicts K p with a linear correlation coefficient ( r ) of 0.91, a prediction efficiency (PE) of 0.81, and a root‐mean‐square (RMS) error of 0.59. Model 2 gives an r of 0.92, a PE of 0.84, and an RMS error of 0.57. The two models are validated through out‐of‐sample test for years 2005 to 2013, which also yields high forecast accuracy. Unlike in the other models reported in the literature, we are taking the response time of the magnetosphere to external solar wind at the Earth explicitly in the modeling. Statistically, the time delay in the models turns out to be about 30 min. By introducing this term, both the accuracy and lead time of the model forecast are improved. Through verification and validation, the models can be used in operational geomagnetic storm warnings with reliable performance.