ionospheric forecasting models
TRANSCRIPT
An Early Warning Ionospheric Alert System using GPS
Observations at Guntur, India
1&2 D.Venkata Ratnam, 1G. Sivavaraprasad,1 S .Lakshmi Narayana, 1M.Venu Gopala Rao and 1JRK Kumar
1Dept. of ECE, K L University, Vaddeswaram, Guntur Dist, Andhra Pradesh, India.2Dept. of Atmospheric Sciences, K L University, Vaddeswaram, Guntur Dist, Andhra Pradesh,
India.
E-mail: [email protected]
4th Feb ,2016
UIM 2016, NRSC, Hyderabad
Outline• Introduction
• Significance of Ionospheric Forecasting
• Ionospheric TEC Estimation
• Ionospheric Forecasting Models
• Results & Discussion
• Conclusion
Introduction
• GPS (Global positioning system), GLONASS and Galileo
• GAGAN (GPS Aided Geo Augmented Navigation)
• Ground Based Augmentation Systems
• Indian Regional Navigation Satellite Systems
• Pseudolite Based Navigation Systems
Predominate error source- Ionosphere- India (Low Latitude region)
1 TECU would cause about 0.3 m of position error
Importance of an Early Warning Ionospheric Alert System
Ionospheric , Solar and
Geomagnetic Sensors Data
Analysis, Modelling and Prediction of ionospheric
Data
Forecasting of IonosphericTime delays
App/Web based
IonosphericSpace weather
Monitoring system for GNSS users
S2/11/2016 4
The spatial and temporal variations of the ionosphere were analyzed using the TECvalues derived from three Indian low-latitude GPS stations separated by 12-180 innorthern latitudes and 78-820 in East longitudes for high solar activity year 2013.
Segment
source
Error source 1 error
(m), before
correction is
applied
SBAS[1]
(m -1)
GBAS/LAAS
(m - 1)
Space Satellite clock 3.0
Satellite
perturbations
1.0
Other (thermal
radiation, etc.)
0.5 0.64 0.21
Control Ephemeris error 4.2
Other (thruster
performance, etc.)
0.9
User Ionospheric delay 5.0
Tropospheric
delay
1.5 0.24
Receiver noise 1.5 1.5 1.5
Multipath 2.5 2.5 2.5
Other (inter
channel bias, etc.)
0.5 0.5 0.5
Total 8.0 3.04 2.96
Error budget of SBAS and GBAS/LAAS for category I landing (Pullen, 2000)
IPP locations, IPP measured delays
IPP locations, IPP measured delays
350Km
from
Ground
INRES
INRES
INMCCIPP
IGP
Ground
GPS satellite
GPS satelliteGEO
IGP delays,GIVE
User’s IPP
Estimates user’s IPP delay, UIVE
Broadcasts1. Differential corrections2. Ranging signal 3. Integrity information
Ionospheric thin shell
INLUS
GAGAN and Iono Model Concept
INMCC Indian Mission Control CenterINLUS Indian Land Uplink StationINRES Indian Reference stations
GPS satellite
0 5 10 15 20 250
1
2
3
4
5
6
Local time (Hrs)
Users
IP
P D
ela
y(m
)
Measured mean
UIPP delay
Estimated mean
UIPP delay
Mean UIPP error
Std UIPP delay error
July 06 2004
International Status• ADVANCED FORECAST FOR ENSURING
COMMUNICATIONS THROUGH SPACE (AFFECTS)
Development of the ionospheric forecasting system over Europe.
Participants:• Institute for Astrophysics, University of Göttingen (DE)
• Royal Observatory of Belgium (BE)
• Space Research Institute of NASU and NSAU (UA)
• Fraunhofer Institute for Physical Measurement Techniques (DE)
• Tromsoe Geophysical Observatory, University of Tromsoe
• German Aerospace Center, Neustrelitz (DE)
• EADS Astrium (EU)
• NOAA Space Weather Prediction Center (US)
Space weather alert and Apps are developed by AFFECTS research group
Official website: http://www.affects-fp7.eu/
KL University GNSS stations Koneru Lakshamaih University (Geographic 16.31N, 80.37E)
Vaddeswaram, that falls under the transition zone between theequatorial trough and the anomaly crest in Indian region.
Model : GPSstation 6, Novatel, Canada
KLU GNSS network consists of four permanent ionosphericmonitoring systems at Guntur, Machilipatnam,Bapatla andNarsapur with spacing of 100 kms along Bay of Bengal sea coastalbelt.
http://igscb.jpl.nasa.gov
72oE 75oE 78oE 81oE 84
oE 8oN
10oN
12oN
14oN
16oN
18oN
20oN
*KLU Guntur
*Hyderabad
*Bangalore
Longitude (Degrees)
Latitu
de (
Degre
es)
GNSS S TEC Data Used
S.No Station Name Geographical
Latitude in degrees N
Geographical Longitude
in degrees E
Year of
data
1 Guntur 16.37 80.37 2013
2 Hyderabad 17.41 78.55 2013
3 Bangalore 13.0212 77.57 2013
Comparison of GPS Observations with IRI Model Predicted Observations in 2013
11
Fig: Seasonal variation of GPSTEC over low latitude station, KL University,Vaddeswaram, India.
Local time (Hrs)
VT
EC
(TE
CU
)
During the high solar activity year 2013 (sun spot number 70) the maximum
values of TEC are observed from 55 to 75 TECU during the period from 1st Jan
to 30th June.
The TEC values are high in equinoctial months than the TEC values during
winter and summer seasons is due to the seasonal variation of TEC is directly
controlled by thermospheric neutral compositions.
Jan,2013 Feb,2013 Mar,2013
April,2013 May,2013June,2013
WinterEquinox
Equinox
Winter
SummerSummer
40TECU 45TECU 55TECU
60TECU 55TECU
45TECU
12
During the day time, the equator is hotter than the poles, therefore
meridional winds flows from equator towards pole.
The flow of meridional wind O/N2 ratio increases at equatorial and
low latitude stations. This increase is maximum in equinox.
At 350km altitude (F2 layer) N2 dissociation is the major process
which removes ambient electrons.
Hence the increase in O/N2 ratio (N2 decreases, i.e., loss decreases)
will result in higher electron density (TEC) and therefore in
equinoctial months TEC will be highest.
TEC diurnal variation is larger for winter months (January) than that of
the summer months (May) is due to the winter anomaly.
Winter anomaly effect is due to the seasonal changes in neutral gas
composition.
13
-13nT-92nT -98nT
June 29, 2013
Storm day
26.87TECU2.472TECU
24.9TECU
26-07-2013 27-07-2013 28-07-201329-07-2013
30-07-2013
Before the storm day at 12:00LT Dst index reaches to -13nT and TEC to 26.87TECU
at 24:00LT Dst to -92nT(Initial phase).
TEC Enhancement- Positive storm.
The prompt penetration of electric field is eastward
60TECU
70TECU
Geomagnetic storm 1 (March 17, 2013)
Geomagnetic storm 1 (March 17, 2013)
DST Index variations along with the original and comparison between forecasted VTEC valuesand predicted IRI – 07 & 2012 values
0 2 4 6 8 10 12 14 16 18-150
-100
-50
0
50
100
Number of Days
DST
Ind
ex V
alue
s (n
T)
DST Index
0 2 4 6 8 10 12 14 16 18
0
50
100
Number of Days
VT
EC
Val
ues
(TE
CU
)
Hourly Forecasting
Original Values ARMA model IRI-07 model IRI-12 model
Geomagnetic storm 1 (March 17, 2013)
0 0.5 1 1.5 2 2.5 3-20
-15
-10
-5
0
5
10
15
20
Number of Days (forecasted)
Obse
reve
d Err
or (T
ECU)
Error in Forecasting
ARMA model IRI-07 model IRI-12 model
Storm DayPrestorm Day Post storm Day
Forecasted error of ARMA compared with error of IRI – 07 & IRI -2012 model
16
March 17, 2013
Storm day-88nT
15nT
-106nT
59.81TECU
5.56TECU
14-03-2013 15-03-2013 16-03-2013 17-03-2013 18-03-2013
At the time of commencement of storm a pulse of increment in Dst
index to15nT at 06:00LT & TEC reaches to 47.01TECU, then starts
decreasing from 07:00LT.
The disturbance dynamo electric field as against the prompt penetration field is westward whereas daytime ionospheric dynamo electric field is in eastward direction.
This causes suppression of EIA and the consequent TEC depletion
TEC depletions-Negative storm
70TECU
62TECU
Geomagnetic storm 2 (June 29, 2013)
Geomagnetic storm 2 (June 29, 2013)
DST Index variations along with the original and comparison between forecasted VTEC valuesand predicted IRI – 07 & 2012 values
0 5 10 15 20 25 30-150
-100
-50
0
50
100
Number of Days
DST
Ind
ex V
alue
s (n
T)
DST Index
0 5 10 15 20 25 300
50
100
Number of Days
VT
EC
Val
ues
(TE
CU
)
Hourly Forecasting
Original Values ARMA model IRI-07 model IRI-12 model
Geomagnetic storm 2 (June 29, 2013)
0 0.5 1 1.5 2 2.5 3
-20
-15
-10
-5
0
5
10
15
Number of Days (forecasted)
Obs
erev
ed E
rror
(TEC
U)
Error in Forecasting
ARMA model IRI-07 model IRI-12 model
Storm Day Post storm DayPrestorm Day
Forecasted error of ARMA compared with error of IRI – 07 & IRI -2012 model
Analysis of Ionospheric Time Delays
Forecasting Methods
Analysis of Ionospheric Time Delays
Forecasting Methods(cntd.,)
1 50 100 150 200 250 300 350 365-5
0
5
10
No.of Days
Actu
al
Fo
reca
st E
rro
r o
f M
eth
od
s (m
)
ARIMA Method
HW-Additive Method
HW-Multiplicative Method
Performance Evaluation of Early Warning
Ionospheric Forecasting Models
1 2 3 4 5 6 7 8 9 10 11 120
20
40
60
No.of Months (Jan-Dec,2013)
MA
PE
(%
)
Multiple Forecast Models MAPE error measurements over 2013 year
Additive
Multiplicative
ARIMA
1 2 3 4 5 6 7 8 9 10 11 120
0.5
1
No.of Months (Jan-Dec,2013)
MA
E(m
)
Multiple Forecast Models MAE error measurements over 2013 year
1 2 3 4 5 6 7 8 9 10 11 120
1
2
No.of Months (Jan-Dec,2013)
MS
D(m
)
Multiple Forecast Models MSD error measurements over 2013 year
GPS-TEC and SSN for the first EOF associated coefficient
2/11/2016 22
Coherence wavelet spectrum between three stations observation data
and the Dst index
2/11/2016 23
Proposed Ionospheric Warning System
Early
warning
• Detection and estimation of solar transient event
• 1-2 days before arrival at Earth
L1-warning
• Measurement of solar transient event near the lagrangian point L1
• Approx. 30 minutes before arrival at Earth
Forecast-warning
• Forecast of ionospheric Perturbations/TEC
• Local Forecast
• A few hours before ionospheric perturbations
Ionospheric-Alert
• Measurement of ionospheric perturbations (Scintillations, TECgradients, flares,…. )
• Near real time alert
Ref: http://elib.dlr.de/89916/1/085.Borries.pdf
Conclusion
• A novel method of approach was introduced for early warning ionosphericalert systems using GPS observations through regression and statisticalmethods.
• The performance of early warning methods to alert ionospheric delays (1h-ahead) over low latitude Guntur station, India has evaluated.
• It was seen that the forecasting results were in good agreement withdiurnal variations of annually observed ionospheric GPS observations.
• The forecasting accuracy of analysed methods was in the range of 53-85%and Holt-Winter multiplicative method is 2% more accurate than othermodels.
• In future, multiple GPS stations data from different geographical locationswill be considered for testing the forecasting capability of ARMA along withARIMA and Holt-Winter models under nominal and sever geomagneticstorm conditions.
• The outcome of these results would be helpful in the process of shieldingthe communication navigation systems from adverse space weather events.
• Ionospheric forecassting models will be implemented with ionospheric TEC,gradients and other solar, geomagnetic indices as input parameters.
Contact: Dr.D .Venkata [email protected],[email protected].
TEC (Total Electron Content)
Total number of free electrons present in a unit cross section area.
Time delay directly propositional to the TEC and inversely proportional to square of the frequency.
Ionospheric Effects on the CNS Applications