Download presentation

Presentation is loading. Please wait.

Published byMohamed Hainsworth Modified over 2 years ago

1
Forecasting energetic electron flux at geostationary orbit P. Wintoft 1), H. Lundstedt 1), and L. Eliasson 2) 1) Swedish Institute of Space Physics - Lund Division 2) Swedish Institute of Space Physics - Kiruna Division

2
Abstract The energetic electron flux at geostationary orbit exhibits large variations on time scales from weeks, through days, and down to hours and below. During times of high flux levels the electron can cause internal charging on spacecraft. In this work we present results on how the electron flux level can be predicted by models driven by measured solar wind data. The electron flux data have been analysed through wavelet transforms obtaining filtered data capturing variations on the different time scales. We will discuss the model performance, prediction horizon, and the wavelet filtered electron data.

3
Data setCoverageReference OMNI ftp://nssdcftp.gsfc.nasa.gov/spacecraft_data/omni/ ACE GOES-8 2 MeV Data

4
Time scales The electron flux exhibits variations on different time scales –diurnal variation due to the orbit –extended high-flux periods (several days) –extended low-flux periods (several days) –sudden flux dropouts (hours)

5
Temporal averaging vs. wavelet decomposition Standard procedure to remove variations on time scales of 24 hours and less is to use daily average values. However, that destroys the dynamics of the data. Instead, we use a wavelet approach.

6
Wavelet approximations and details Level 1: Period = 1.4·2 1 = 2.8 hours Level 2: Period = 1.4·2 2 = 5.6 hours Level 3: Period = 1.4·2 3 = 11.8 hours Level 4: Period = 1.4·2 4 = 23.6 hours Daubechies 4 wavelet: Central period = 1.4

7
Data analysis

8
Captured variance Variance of log electron flux = 1.07 Variance of daily average log e-flux = 0.83 (78%) Variance of approximation at level 4 = 0.88 (83%)

9
Wavelet summary Using wavelet decomposition we may study the hourly average electron flux at varying degrees of detail. 83% of the variance is captured in A4. 96% of the variance is captured in A3=A4+D4.

10
Model Time-delay units => Detailed memory of past events. Internal feed-back units => Dynamics (diff. eq.) and averaging over time.

11
Solar wind input

12
Model output and observation

13
Superposed epoch analysis: Key event = flux increase over a 10 hour period

14
No signature in Bx or By. Negative Bz leads by 8 hours. Density increase leads by 20 hours. Density peak before rise in velocity => coronal holes.

15
Superposed epoch analysis: Key event = flux decrease over a 10 hour period

16
Signatures in solar wind plasma and magnetic field parameters does not exist or lags the flux deacrease. However, rotations in solar wind magnetic field (Bx changing sign) seems to be simultanous with the flux decrease.

17
Summary Wavelet analysis provides an efficient method of decomposing the flux variation on different time scales. Energetic flux increases are predictable up to 20 hours in advance. However, flux decreases may only be nowcasted. Model development is in progress –further tuning of weights. –inclusion of D4 will also capture diurnal variations. –analysis of the importance of the different solar wind parameters.

Similar presentations

© 2016 SlidePlayer.com Inc.

All rights reserved.

Ads by Google