Kalman filter improvement of the monthly smoothed Sunspot Number prediction
| Source |
SIDC (RWC-Belgium)
|
| Frequency |
Monthly
|
| Format |
Plain text
|
| Mail header |
Kalman filter improvement of the monthly smoothed Sunspot Number prediction
|
| SIDC code |
kalfil
|
Issued: 2012 May
#--------------------------------------------------------------------#
# MONTHLY REPORT OF THE INTERNATIONAL SUNSPOT NUMBER #
# FROM THE SOLAR INFLUENCES DATA ANALYSIS CENTER (RWC-BELGIUM) #
#--------------------------------------------------------------------#
Kalman filter improvement of the monthly smoothed Sunspot Number prediction
by SIDC classical method (SM) and by the Combined method (CM).
Predictions of SM and CM methods were taken from http://www.sidc.be/products/ri
The last provisional value, calculated for October 2011:59.9(+-5%)
KFSM KFCM
2011 Nov 63 2011 Nov 62
Dec 68 Dec 64
2012 Jan 66 2012 Jan 67
Feb 60.2 (5) Feb 61.9 (5)
Mar 61.6 (6) Mar 62.9 (6)
Apr 70.2 (6) Apr 70 (6)
May 73.4 (7) May 72.3 (8)
Jun 77.6 (9) Jun 75.5 (9)
Jul 82.1 (10) Jul 78.4 (10)
Aug 86.5 (11) Aug 81.1 (11)
Sep 90.5 (12) Sep 83.6 (12)
Oct 94.9 (13) Oct 85.9 (12)
Nov 99.5 (14) Nov 88.2 (13)
Dec 104 (15) Dec 89.2 (14)
2013 Jan 108 (16) 2013 Jan 89.7 (15)
Feb 111 (17) Feb 90.4 (15)
Mar 113 (18) Mar 90.6 (16)
Apr 114 (18) Apr 90.1 (16)
KFSM: Kalman filter prediction correction for SM. Standard deviation
of estimates errors are given in brackets.
KFCM: Kalman filter prediction correction for CM. Standard deviations
of estimates errors are given in brackets.
The improvement of the predictions is provided by applying an adaptive Kalman
filter to obtained medium-term predictions using the last six monthly mean
values of sunspot numbers, which cover the six months between the last available
value of the 13-month running mean (the starting point for the predictions)
and the current time. The proposed technique reduces stochastic component of
the last six monthly mean sunspot numbers that give significant information
about cycle evolution and provides effective estimate of sunspot activity
at the current time. This estimate becomes the new starting point for
the prediction updating that is shifted six month ahead in comparison with
the last observed 13-month running mean and provides an increase of prediction
accuracy for medium-term methods.
Correction technique was proposed by T. Podladchikova and R. Van der Linden and
improves medium term prediction methods as they are monthly updated using the last
available observations of smoothed sunspot numbers.
ref.: T. Podladchikova, R. Van der Linden, 2011: "A Kalman Filter Technique for Improving
Medium-Term Predictions of the Sunspot Number". Solar Physics. DOI: 10.1007/s11207-011-9899-y
#--------------------------------------------------------------------#
# Solar Influences Data analysis Center - RWC Belgium #
# Royal Observatory of Belgium #
# Fax : 32 (0) 2 373 0 224 #
# Tel.: 32 (0) 2 373 0 491 #
#--------------------------------------------------------------------#
# For more information, comments and suggestions write to #
# Ronald Van der Linden ronald@oma.be, #
# Tanya Podladchikova tatyana@oma.be #
#--------------------------------------------------------------------#
Kalman filter for the standard and combined methods
Issued: 2012 Apr
# MONTHLY REPORT OF THE INTERNATIONAL SUNSPOT NUMBER #
# FROM THE SOLAR INFLUENCES DATA ANALYSIS CENTER (RWC-BELGIUM) #
#--------------------------------------------------------------------#
Kalman filter improvement of the monthly smoothed Sunspot Number prediction
by McNish&Lincoln method (M&L). Predictions of M&L method were taken from
ftp://ftp.ngdc.noaa.gov/STP/SOLAR_DATA/SUNSPOT_NUMBERS/INTERNATIONAL/prediction/sunspot.predict
The last provisional value, calculated for September 2011:59.5(+-5%)
KFM&L
2011 Oct 61.7
Nov 63.9
Dec 65.6
2012 Jan 64.5 (5)
Feb 63.4 (6)
Mar 68.4 (6)
Apr 68.9 (7)
May 70.4 (8)
Jun 71.6 (9)
Jul 73 (10)
Aug 74.6 (11)
Sep 76.3 (12)
Oct 77.8 (13)
Nov 78.8 (13)
Dec 79.8 (14)
2013 Jan 80.8 (15)
Feb 81.8 (15)
Mar 82.7 (16)
KFM&L: Kalman filter prediction correction for McNish and Lincoln method.
Standard deviation of estimates errors are given in brackets.
The improvement of the predictions is provided by applying an adaptive Kalman
filter to obtained medium-term predictions using the last six monthly mean
values of sunspot numbers, which cover the six months between the last available
value of the 13-month running mean (the starting point for the predictions)
and the current time. The proposed technique reduces stochastic component of
the last six monthly mean sunspot numbers that give significant information
about cycle evolution and provides effective estimate of sunspot activity
at the current time. This estimate becomes the new starting point for
the prediction updating that is shifted six month ahead in comparison with
the last observed 13-month running mean and provides an increase of prediction
accuracy for medium-term methods.
Correction technique was proposed by T. Podladchikova and R. Van der Linden and
improves medium term prediction methods as they are monthly updated using the last
available observations of smoothed sunspot numbers.
ref.: T. Podladchikova, R. Van der Linden, 2011: "A Kalman Filter Technique for Improving
Medium-Term Predictions of the Sunspot Number". Solar Physics. DOI: 10.1007/s11207-011-9899-y
#--------------------------------------------------------------------#
# Solar Influences Data analysis Center - RWC Belgium #
# Royal Observatory of Belgium #
# Fax : 32 (0) 2 373 0 224 #
# Tel.: 32 (0) 2 373 0 491 #
#--------------------------------------------------------------------#
# For more information, comments and suggestions write to #
# Ronald Van der Linden ronald@oma.be, #
# Tanya Podladchikova tatyana@oma.be #
#--------------------------------------------------------------------#
Kalman filter for the McNish&Lincoln method
