HERMANUS MAGNETIC OBSERVATORY - Public Website · 2012. 7. 1. · GSV4004B GISTM and conventional Ashtech dual frequency GPS receiver installed on Gough Island. Observations were

Hermanus Magnetic

Observatory

Page 1 of 71

Classification: CONFIDENTIAL

Classification: CONFIDENTIAL DOC: 6021-0003-709-A1

HERMANUS MAGNETIC OBSERVATORY

Correlation between energetic charged particle precipitation

over the South Atlantic Magnetic Anomaly

and L-band Ionospheric

Scintillation over South Africa: Investigation in support of SA’s

SKA bid.

Doc No: 6021-0003-709-A1

Prepared by: Drs Ben Opperman and Pierre Cilliers

Prepared for: Dr Adrian Tiplady

Date: 22/11/2010

Hermanus Magnetic Observatory P O Box 32 HERMANUS 7200

Hermanus Magnetic

Observatory

Page 3 of 71



Table of Contents

Table of Contents ............................................................................................................. 3 List of Figures ................................................................................................................... 5 List of Tables .................................................................................................................... 8 1 Introduction .............................................................................................................. 9 2 Methodology and data .............................................................................................. 9 3 Background ............................................................................................................... 9 4 Ionosphere and Total Electron Content .................................................................. 13 5 Ionospheric Scintillation ......................................................................................... 14 6 Proxies for S4 amplitude scintillation ..................................................................... 15 7 Validating the S4-proxy algorithm .......................................................................... 18 8 Correlation between geomagnetic disturbance and EPP ........................................ 20

8.1 Disturbance storm time index (Dst) ............................................................... 21 9 Data challenges ....................................................................................................... 23

9.1 Non-availability of data. ................................................................................. 23 9.2 False positives associated with cycle slips. .................................................... 24

10 DMSP SSJ/4 EPP sensor ........................................................................................ 26 11 Data processing....................................................................................................... 27

11.1 DMSP ............................................................................................................. 27 11.2 GPS ................................................................................................................. 29 11.3 Integrating SSJ/4 and S4p data ........................................................................ 30

11.3.1 Visual comparison .................................................................................. 30 11.3.2 Statistical correlation of median values .................................................. 31 11.3.3 Statistical correlation of integrated values ............................................. 31

12 Results .................................................................................................................... 31 13 Discussion ............................................................................................................... 35 14 Conclusions ............................................................................................................ 38 15 Key Project Participants ......................................................................................... 38

15.1.1 Dr BDL Opperman. Co-ordinator ......................................................... 38 15.1.2 Dr PJ Cilliers. ......................................................................................... 38 15.1.3 Dr LA McKinnell. .................................................................................. 38

16 Appendix A: Precipitation – Scintillation correlation results ................................. 39 16.1 Gough Island 2000 ......................................................................................... 40 16.2 Gough Island 2001 ......................................................................................... 42 16.3 Gough Island 2003 ......................................................................................... 44 16.4 Gough Island 2004 ......................................................................................... 46 16.5 Gough Island 2008 ......................................................................................... 48 16.6 Cape Town 2000 ............................................................................................. 50 16.7 Cape Town 2001 ............................................................................................. 52 16.8 Cape Town 2003 ............................................................................................. 54 16.9 Cape Town 2004 ............................................................................................. 56 16.10 Cape Town 2008 ......................................................................................... 58 16.11 Perth 2000 ................................................................................................... 60 16.12 Perth 2001 ................................................................................................... 62 16.13 Perth 2003 ................................................................................................... 65 16.14 Perth 2004 ................................................................................................... 67

Hermanus Magnetic

Observatory

Page 4 of 71



16.15 Perth 2008 ................................................................................................... 69 17 References .............................................................................................................. 71 18 Web pages .............................................................................................................. 71

Hermanus Magnetic

Observatory

Page 5 of 71



List of Figures

Figure 1. Contours indicating geomagnetic field strength illustrate the weak

magnetic field over the South Atlantic Magnetic Anomaly region. Superimposed on the figure are the magnetic equator, equatorial anomaly (± 15° from geomagnetic equator) and aurora oval edge. ......... 10

Figure 2. World map of the AP-8 MAX integral proton flux >10 MeV at 500 km altitude. (www.spenvis.oma.be/help/background/traprad/traprad.html) . 11

Figure 3. World map of the AE-8 MAX integral electron flux >1 MeV at 500 km altitude (bottom). (www.spenvis.oma.be/help/background/traprad/traprad.html) ................ 11

Figure 4. Electron particle precipitation as observed with the DMSP SSJ4-instrument. Superimposed on the image are the geomagnetic equator (magenta), equatorial anomaly regions (cyan) and locations of Gough Island, Cape Town and Perth (yellow triangles). ..................................... 12

Figure 5. Proton particle precipitation as observed with the DMSP SSJ4-instrument. Superimposed on the image are the geomagnetic equator (magenta), equatorial anomaly regions (cyan) and locations of Gough Island, Cape Town and Perth (yellow triangles). ..................................... 12

Figure 6. Auroral oval footprint shifting with Kp index. During Geomagnetically disturbed conditions, aurora-associated particle precipitation is observed in the auroral oval. With increased Kp values, associated with high disturbance, auroral observations extend northwards (in Southern hemisphere) as indicated by the Kp-lines. www.swpc.noaa.gov/Aurora/globeSE.html. Note that Southern Africa and the South Atlantic Magnetic Anomaly are not affected by auroral particle precipitation during disturbed conditions. ................................... 13

Figure 7. Elevation weighting coefficient (β) of Du et al. used to relate ROTI to the S4-proxy S4p ......................................................................................... 17

Figure 8. ROTI as S4p-proxy calculated from Gough Island GPS data on 28 October 2003 (day 301). The increased ROTI values around 11:00 UT are indicative of scintillation events observed by five satellites. Severe geomagnetic storm (Kp = 9, Dst = -475 nT) was experienced on this day. ................................................................................................ 17

Figure 9. Geographic location of ROTI- S4p-proxy occurrences (red circles) superimposed on GPS satellite IPP ground trace (green dots), as observed from Gough Island on 28 Oct 2003. Large blue circles represent different satellite elevation angles. The relative high elevation occurrence of ROTI illustrates that observations are not associated with (low elevation) multipath effects. These ROTI occurrences were observed around the same time around 11:00 UT by five different satellites. ............................................................................. 18

Figure 10. Comparison of S4 and ROTI and other related parameters from GPS observations at Ascension Island, (JASTM 61 (1999) pp 1219-1226 ) ... 19

Figure 11. Comparison of GISTM-observed S4 scintillation with elevation weighted S4 proxy (S4p,) as observed respectively by a Novatel

Hermanus Magnetic

Observatory

Page 6 of 71



GSV4004B GISTM and conventional Ashtech dual frequency GPS receiver installed on Gough Island. Observations were done along the ray path from satellite PRN10. The S4p were determined at 5-minute intervals from GPS data sampled at 30s intervals, the ROTI values were determined at 30 second intervals while the S4 values were recorded at 1 minute intervals. The top panel in the diagram shows the elevation (Elv) of satellite PRN10 on the same time scale to indicate that the observed event occurred at an elevation of about 50 ⁰, well above the horizon, and thus not due to multipath effects. Panel 2 shows the normalised power on the L1-signal (CNo) recorded at a sampling rate of 50 samples per second with the GISTM to illustrate the power dip that is associated with the S4-event. The TEC shown in panel 5 depicts the small but rapid increase in the TEC which is associated with the dip in the power, which seems to indicate that the S4-event was caused by the ray path traversing a region of increased electron density, which increased the absorption of the signal. The slope of the TEC curve is shown as the Rate of Change of TEC (ROT) in panel 6. Panel 7 shows the Rate of Change of TEC Index, which is standard deviation of the ROT, with averaging done over a 5-minute period. ....................................................................................................... 19

Figure 12. Comparison of GISTM-observed S4 scintillation with elevation weighted S4 proxy (S4p,) as observed respectively by a Novatel GSV4004B GISTM and conventional Ashtech dual frequency GPS receiver installed on Gough Island. Observations were done along the ray path from satellite PRN18. ................................................................. 20

Figure 13. Year 2000 Proton flux values at various energy levels (top) compared to Dst values (bottom) .............................................................................. 22

Figure 14 Year 2004 Proton flux values at various energy levels (top) compared to Dst values (bottom) .................................................................................. 22

Figure 15. Example of false positive ROTI caused by cycle slips not associated with sustained perturbed ionospheric phase observation. For satellite PRN13, observed from Perth on 12 June 2008, cycle slips are observed in L1 and L2 phase as phase jumps (top), abrupt jumps in slant TEC with ~2 TECU (middle) and associated ROTI (bottom). Note the relative high ROTI values associated with these cycle slips. Such ROTI calculations typically are associated with false positive scintillation events. ................................................................................... 25

Figure 16. Example of false positive ROTI scintillation events associated with cycle slips observed on all GPS receiver channels for 12 June 2008 at Perth. Scrutiny revealed this day’s raw data was too suspicious to use and was subsequently ignored. ................................................................. 26

Figure 17. Geographic windows of 10˚ centred on the key locations ......................... 28 Figure 18 SSJ4 Electron and ion flux and energy levels observed at South Atlantic

Anomaly region around 2001-03-16-06:41 UTC . The red line indicates the geographic latitude of the Gough Island GPS receiver used for the ionospheric scintillation calculations. Note that particle precipitation values are an order of magnitude larger than for Cape Town. ........................................................................................................ 28

Hermanus Magnetic

Observatory

Page 7 of 71



Figure 19. SSJ4 Electron and ion flux and energy levels observed at South Africa around 2001-03-16-03:17UTC. The red line indicates the geographic latitude of the Cape Town GPS receiver used for the ionospheric scintillation calculations. ....................................................... 29

Figure 20. SSJ4 Electron and ion flux and energy levels observed over Australia around 2001-03-16-08:2UTC. The red line indicates the geographic latitude of the Perth GPS receiver used for the ionospheric scintillation calculations. .............................................................................................. 29

Figure 21. Year 2003 Gough Island electron (Ne) and ion (Ni) flux (panes 1-2), 5-minute median S4p (ROTI), pane 3) and Dst (bottom). Note the large S4p values around day 319 which appears to coincide with a large Dst value and apparent increased electron and ion flux. ................................. 32

Figure 22. Year 2003 statistical correlation of Gough Island electron (Ne) and ion (Ni) flux with 5-minute median S4p (ROTI) ............................................. 33

Figure 23. Year 2003 Statistical correlation between, respectively, daily integrated electron flux (top) and ion flux (bottom) and S4p ..................................... 33

Figure 24. Year 2001 days 88-94 illustrate Cape Town and Perth observations of proton flux (black crosses) compared with S4p scintillation (blue dots). No GPS data was available for Gough Island for this period, but proton flux is illustrated for completeness. .............................................. 36

Figure 25. Year 2003 days 303-307 illustrate Cape Town, Perth and Gough island observations of proton flux (black crosses) compared with S4p scintillation (blue dots). Minimum Dst on day 303 was -383 nT ............. 37

Hermanus Magnetic

Observatory

Page 8 of 71



List of Tables

Table 1. Geomagnetic field parameters at selected locations on June 2010. ............ 10 Table 2. Archived GPS data availability, in days, with percentage days available

for a specific year given in brackets ()..................................................... 23 Table 3 Geographic and Geomagnetic Coordinates of key locations used in the

study together with the GISTM data availability. .................................... 27 Table 4. Correlation coefficients between S4p and EPP. (∫S4p represents integrated

S4p) ............................................................................................................ 34

Hermanus Magnetic

Observatory

Page 9 of 71



1 Introduction The objective of the study is to investigate the possible effect(s) of energetic particle precipitation (EPP), observed over the South Atlantic Magnetic Anomaly (SAMA) region, on ionospheric scintillation over Southern Africa and Australia. This study was conducted in support of the Square Kilometre Array (SKA) bid for identifying a suitable host location based on the stability of the ionosphere so as to minimize modulation of trans-ionospheric radio astronomy signals. Theoretical studies (Figure 2 and 3) and satellite X-ray and energetic proton and electron measurements (Figure 4 and 5) have indicated that the eastern limb of the SAMA extends to the southern tip of Africa. This phenomenon begs the question as to what extent SAMA EPP influences the South African ionosphere. Although studies by Gledhill (1976), Haggard (2004) and Abdu et al. (2005) identified EPP as a possible source of increased ionospheric ionization observed over the SAMA, a study by Sibanda (2006) concluded that it was not possible to establish a direct connection between EPP events and ionospheric disturbances over the South African region. By extending the data sets used by Sibanda (2006) and extending the observation period and regional coverage, this study set out to specifically investigate the relation between SAMA EPP and ionospheric scintillation over South Africa and Australia.

2 Methodology and data EPP was expressed in terms of energetic electron and proton flux as well as particle energy measurements by the Defence Meteorological Satellite Programme (DMSP). DMSP measurements over SAMA were compared to ionospheric scintillation indices calculated from GPS measurements observed at Gough Island (South Atlantic), Cape Town (South Africa) and Perth (Australia) and statistics were derived to quantify EPP-scintillation correlations. To cover a representative range of geomagnetic and solar conditions, this study was conducted for geomagnetic quiet and disturbed conditions using observations over an almost complete solar cycle for the years 2000, 2001, 2003, 2004 and 2008. Solar cycle 23 reached its peak in 2001 and significant solar disturbances, with accompanying geomagnetic/ionospheric disturbances, were observed around solar maximum with major events also occurring in 2003 and 2004.

3 Background The South Atlantic Magnetic Anomaly (SAMA, See Figure 1) is a region located between Southeast Brazil and South Africa, where the magnetic field strength is particularly low due to the eccentric nature of the Earth's magnetic dipole. The minimum value of the total geomagnetic field of roughly 22575 nT is found within the SAMA at about 26° South and 54° West. Energetic Particle Precipitation (EPP) is primarily observed in the aurora regions and over the SAMA. EPP involves the deposition of energetic particles from the Van Allen radiation belts

Hermanus Magnetic

Observatory

Page 10 of 71



into the ionosphere, which would result in an increase in ionization and conductivity of the upper atmosphere. The precipitation mechanism proceeds by energetic electrons being scattered into the loss cone and subsequently interacting with atmospheric neutrals. The altitude at which the electrons are deposited depends on their energy and pitch angle as well as the strength of the magnetic field. Lower magnetic field strengths, such as those found within the SAMA, are thus conducive to EPP.

Figure 1. Contours indicating geomagnetic field strength illustrate the weak magnetic field over the South Atlantic Magnetic Anomaly region. Superimposed on the figure are the magnetic equator, equatorial anomaly (± 15° from geomagnetic equator) and aurora oval edge.

Table 1. Geomagnetic field parameters at selected locations on June 2010.

Location Magnetic Latitude

Inclination Total Field strength (nT)

Approximate weakest field position over SAMA (26⁰S, 54⁰W)

-17.8 -32° 04’ 22 575

Gough Island (40°21'S, 9°52'W)

-42.5 -63° 44’ 24 658

Cape Town (34°11'S, 18°26'E)

-36.7 -65° 59’ 25792

Perth (31°48'S, 115°53'E) -25.22 -66°15’ 58 284 Note: Total Field Strength and Inclination calculated by means of the IGRF11 model (www.ngdc.noaa.gov/geomag/magfield.shtml). Magnetic Latitude calculated by the on-line CGM model (omniweb.gsfc.nasa.gov/vitmo/cgm_vitmo.html)

Hermanus Magnetic

Observatory

Page 11 of 71



Figure 2. World map of the AP-8 MAX integral proton flux >10 MeV at 500 km altitude. (www.spenvis.oma.be/help/background/traprad/traprad.html)

Figure 3. World map of the AE-8 MAX integral electron flux >1 MeV at 500 km altitude (bottom). (www.spenvis.oma.be/help/background/traprad/traprad.html)

Hermanus Magnetic

Observatory

Page 12 of 71



Figure 4. Electron particle precipitation as observed with the DMSP SSJ4-instrument. Superimposed on the image are the geomagnetic equator (magenta), equatorial anomaly regions (cyan) and locations of Gough Island, Cape Town and Perth (yellow triangles).

Figure 5. Proton particle precipitation as observed with the DMSP SSJ4-instrument. Superimposed on the image are the geomagnetic equator (magenta), equatorial anomaly regions (cyan) and locations of Gough Island, Cape Town and Perth (yellow triangles).

Electron particle flux from DSP ssj4 sensor day 001 of 2000

-180 -120 -60 0 60 120 180-90

-60

-30

0

30

60

90

0.5

1

1.5

2

2.5

3

x 108

Ion particle flux for day 001 of 2000

-180 -120 -60 0 60 120 180-90

-60

-30

0

30

60

90

0.5

1

1.5

2

x 107

Hermanus Magnetic

Observatory

Page 13 of 71



Figure 6. Auroral oval footprint shifting with Kp index. During Geomagnetically disturbed conditions, aurora-associated particle precipitation is observed in the auroral oval. With increased Kp values, associated with high disturbance, auroral observations extend northwards (in Southern hemisphere) as indicated by the Kp-lines. www.swpc.noaa.gov/Aurora/globeSE.html. Note that Southern Africa and the South Atlantic Magnetic Anomaly are not affected by auroral particle precipitation during disturbed conditions.

4 Ionosphere and Total Electron Content The ionosphere is the portion of the atmosphere that lies between about 50 km and about 2000 km in which the density of free electrons formed by photo-ionization is large enough to have an appreciable effect on the propagation of radio waves. The ionosphere is divided into D, E and F-layers. The D-layer is the innermost layer ranging from 60 km to 90 km; the E-layer ranges from 90 km to 120 km and the F-layer, in which the peak electron density occurs, extends from about 200 km to more than 500 km above the surface of Earth. The Total Electron Content (TEC) is defined as the integral of electron density along a cylindrical column centred on a ray path between a radio receiver and a transmitting satellite through the atmosphere. Faraday rotation, a left-handed elliptical polarisation rotation of EM rays propagating through electrically-conducting plasma in the presence of a magnetic field, is directly proportional to TEC (which includes contributions from the interstellar medium) and has relevance in radio astronomy signals. In the context of GPS satellites orbiting at 20 200 km, TEC is derived from a linear combination of the phase observations on the L1 (1.5754 GHz) and L2 (1.2276 GHz) carrier waves and given in TEC Units (TECU) with 1 TECU = 1016 electrons.m-2 (Schaer, 1999):

1

2 2 1 12 21 2

1 1 [ ]TEC TECUf f

where

α : 40.28x1017 m.s-2.TECU-1 φ1, φ2 : Phase observed on L1, L2 carrier waves f1, f2 : L1, L2 frequencies (Hz) λ1, λ2: L1, L2 wavelengths (m)

Hermanus Magnetic

Observatory

Page 14 of 71



5 Ionospheric Scintillation Ionospheric scintillations are rapid variations (1-10 Hz) in the electron density of the ionosphere, particularly near the F2 layer, that cause rapid fading on HF Radio (3-30 MHz) communications, deterioration of the accuracy of GPS navigation, and fluctuations in the polarization, amplitude and phase of radio-astronomy signals from cosmic sources. As an EM waves pass through the ionosphere, the phase velocity is advanced and the group velocity delayed by an amount proportional to the TEC. Ionospheric scintillation manifests as rapid amplitude fading and phase shifts which occur when the ray path of a trans-ionospheric signal traverses ionospheric irregularities in the form of small scale F-region field aligned patches of increased electron density or bubbles of decreased electron density Field aligned patches are more common at auroral latitudes while bubbles more frequently occur near the equator. At mid-latitudes, both can occasionally occur. The severity of amplitude scintillation is expressed in terms of the S4 index, defined as the normalised second central moment of the signal intensity i.e.

22

4 ,I I

SI

where I is the intensity of the signal measured at the receiver, expressed as the square of the amplitude. Values of 4S below 0.3 are called weak scintillations, which have minimal effect, while values above 0.3 are strong scintillations. For weak scintillation, 4S is proportional to the variance of the electron density variations 2N . Phase scintillation is defined in terms of the phase scintillation index, or the standard deviation of phase fluctuations defined as

22 , where is the received phase. With dedicated L-band ionospheric scintillation monitors such as the GSV4004B GPS Ionospheric Scintillation and TEC Monitor (GISTM) installed on Gough Island, the amplitude and phase is sampled at 50 Hz, and averaged over 1 minute. The scale of the irregularities determines its effect. If the scale of the irregularities is much larger than the Fresnel radius ,fz r amplitude variation is minimal. Here is the wavelength and r is the distance from the irregularities to the receiver. At or below fz amplitude variations on signals traversing the irregularities are significant. At typical ionospheric heights (~400 km for the F2-layer peak) and assuming vertical propagation, fz is of the order of 276 m at the GPS L1 frequency. Ionospheric scintillation depends on solar activity, geomagnetic activity, season, time of day, geographical location and frequency. Scintillations occur predominantly in the equatorial band that extends from about 20⁰S to 20⁰N of the magnetic equator, and in the auroral and polar cap regions. The processes that produce scintillations in these two regions are quite different, leading to significant differences in the characteristics of the resulting scintillations. In the equatorial regions, ionospheric scintillations predominantly occur during the period between dusk and local midnight, when the upward E B drift creates low density structures

Hermanus Magnetic

Observatory

Page 15 of 71



moving upward through the F-layer of the ionosphere, an effect known as the Equatorial Fountain. Equatorial scintillations increase during the equinoxes and during solar maximum. In the high latitudes, ionospheric scintillation is association with field aligned patches of increased electron density and are peaked in the period centred on midnight magnetic local time (MLT). In the mid-latitudes, scintillations may result from equatorial irregularities and high latitude irregularities moving towards the mid-latitudes, as well as from the precipitation of energetic particles into the ionosphere. Other mechanisms for high and mid-latitude scintillations show a strong dependence of the direction of the component of the interplanetary magnetic field (IMF) which is aligned with the Earth’s axis (Bz). During northward-directed Bz the cause of scintillations is bursty bulk flow irregularities of unknown origin, while during southward-directed Bz, the scintillations are caused by the electric field associated with field-line mapped movement of plasma in the magnetosphere. Ionospheric scintillation has been noted to occur in regions of large temporal and spatial gradients of the total electron content (TEC). Scintillations can also occur during daylight hours and at mid-latitudes when Sporadic-E is present in the E-layer. Sporadic-E layers are thin layers of highly dense plasma at heights of about 100 km in which large density gradients can exist. However, scintillations produced by Sporadic-E are much less common and less predictable than those produced by the F-layer processes described above. An index derived from the rate of change of TEC, the so-called Rate of change of TEC index (ROTI), is often used as a proxy for 4S . In this analysis of the effects of precipitation on ionospheric scintillation, ROTI is primarily used to express the intensity of scintillation, since high sampling rate ionospheric scintillation monitors were not available at the locations of interest during the previous solar minimum and solar maximum.

For weak scintillations, 4 2.51S

f where f is the carrier frequency. This means that

scintillation is stronger at lower frequencies. At the geomagnetic equator 4S does not depend on geomagnetic activity, whereas high-latitude scintillation is frequently observed during disturbed magnetic conditions and is thought to be associated with an influx of high energy electrons that gain entry into the Earth’s polar cap regions when the solar wind IMF couples with the Earth’s magnetic field. A similar situation occurs in the South Atlantic Magnetic Anomaly (SAMA) due to the increased precipitation of high energy electrons in the atmosphere over the SAMA. Existing models to predict ionospheric scintillation are best developed for equatorial regions, but are inadequate for mid-latitude regions.

6 Proxies for S4 amplitude scintillation In the absence of measured L-band 4S scintillation values at the GPS stations of which the data was used in this study, a number of proxies for the amplitude scintillation index

4S can be used. The rate of change of TEC index (ROTI), defined as the standard deviation,

22ROTI ROT ROT

Hermanus Magnetic

Observatory

Page 16 of 71



of the rate-of-change of TEC (ROT) with the averaging done over a 5-minute period from 30 second sampled GPS data has been proposed (Pi et al.,1997), Studies by (Pi et al.,1997); (Basu et al.,1999) and (Beach and Kintner, 1999) concluded that ROTI could be used as a proxy for assessing the presence of ionospheric scintillation. Beach and Kintner (Beach and Kintner, 1999) concluded that ROTI roughly proportional to measures of TEC fluctuation for weak scintillation with ROTI (2 5)S4 rendering good expressions for their measurements. Similarly, Basu et al.found ROTI (2 )S4 using their dataset. The quantitative relationship between ROTI and S4 varies considerably due to variations of the ionospheric projection of the satellite velocity and the ionospheric irregularity drift. The study by (Basu et al, 1999) indicated that ROTI is selective of Fresnel scale structures of 400 m at GPS frequencies. In this study the approach of (Du et al 2000) was followed in quantifying a proxy scintillation index (S4p) related to ROTI by an elevation-weighted coefficient . The coefficient is calculated from a function of satellite’s motion relative to the ionosphere, which is assumed to be concentrated on an imaginary thin shell 400 km above earth (the phase screen):

4 pS ROTI where

2

36.2135 10 vs

2 2 2 22 cos cosp I p Iv v v v v i s

pe s

v r hv

R r

cos( )cos

r hs

1 cossin rr h

Parameters and relevant values used in the calculation:

Re: Earth radius (6378 km) h: Assumed ionospheric height (400 km) θ: Satellite elevation angle i: GPS orbital plane inclination (55o) vI: Ion drift velocity (120 m.s-1) vp: Velocity of the ionospheric pierce point moving at h vs: Satellite orbital speed (3874 m.s-1) rs: Satellite orbital altitude (20 000 km)

The elevation-weighted coefficient (β) shown in Figure 7 selectively suppresses low elevation ROTI compared to high elevation ROTI. An example illustrating S4p proxy, calculated from ROTI for Gough Island, is illustrated in Figure 8 for 28 October 2003 during which a severe geomagnetic storm occurred.

Hermanus Magnetic

Observatory

Page 17 of 71



Figure 9 shows the geographic location of the S4p occurrences (red circles) superimposed on GPS satellite IPP ground trace (green dots), as observed from Gough Island on 28 Oct 2003.

Figure 7. Elevation weighting coefficient (β) of Du et al. used to relate ROTI to the S4-proxy S4p

Figure 8. ROTI as S4p-proxy calculated from Gough Island GPS data on 28 October 2003 (day 301). The increased ROTI values around 11:00 UT are indicative of scintillation events observed by five satellites. Severe geomagnetic storm (Kp = 9, Dst = -475 nT) was experienced on this day.

0 3 6 9 12 15 18 21 240

5

10

15

20

25

30

Time [UTC hours]

S 4p

Gough Island S4p scintillation proxy calculated from ROTI. 28 October 2003 (day 301)

Hermanus Magnetic

Observatory

Page 18 of 71



Figure 9. Geographic location of ROTI- S4p-proxy occurrences (red circles) superimposed on GPS satellite IPP ground trace (green dots), as observed from Gough Island on 28 Oct 2003. Large blue circles represent different satellite elevation angles. The relative high elevation occurrence of ROTI illustrates that observations are not associated with (low elevation) multipath effects. These ROTI occurrences were observed around the same time around 11:00 UT by five different satellites.

7 Validating the S4-proxy algorithm Several studies (Pi et al. 1997, Beach & Kintner 199, Du et al. 2000) have shown correlation between S4 amplitude scintillation and ROTI and confirmed the theory for the proxy ROTI-S4 use. To establish confidence in the S4-proxy algorithm the GPS-derived elevation weighted S4-proxy, S4p and the ROTI from which it is derived were compared with actual GISTM S4- scintillation results obtained from the GISTM installed on Gough Island in September 2008(Table 3). Because of the prolonged solar minimum period towards the end of solar cycle 23, very few significantly geomagnetic disturbed periods were observed. A moderate geomagnetic storm observed on 3 August 2010 (Kp = 5) rendered significant S4 counts for Gough Island which were however not matched by any concomitant scintillations observed in Hermanus. Figure 10 shows a comparison of S4 with ROTI for data from the equatorial GPS station at Ascension. Figure 11 and Figure 12 show some examples using data from Gough Island obtained during the geomagnetic storm on 3 August 2010. The parameters Co/N (signal-to-noise-ratio) and S4 were obtained from the Novatel GISTM receiver, while the remaining parameters were from the Ashtech standard GPS receiver. These comparisons, combined with exhaustive hand-checking of results, supplied the necessary confidence in the algorithms and software.

344 346 348 350 352 354 356-45

-44

-43

-42

-41

-40

-39

-38

-37

-36

Longitude [Degrees]

Latit

ude

Gough Island ROTI scintillation on 28 Oct 2003

80o

60o

40o

Hermanus Magnetic

Observatory

Page 19 of 71



Figure 10. Comparison of S4 and ROTI and other related parameters from GPS observations at Ascension Island, (JASTM 61 (1999) pp 1219-1226 )

Figure 11. Comparison of GISTM-observed S4 scintillation with elevation weighted S4 proxy (S4p,) as observed respectively by a Novatel GSV4004B GISTM and conventional Ashtech dual frequency GPS receiver installed on Gough Island. Observations were done along the ray path from satellite PRN10. The S4p were determined at 5-minute intervals from GPS data sampled at 30s intervals, the ROTI values were determined at 30 second intervals while the S4 values were recorded at 1 minute intervals. The top panel

Hermanus Magnetic

Observatory

Page 20 of 71



in the diagram shows the elevation (Elv) of satellite PRN10 on the same time scale to indicate that the observed event occurred at an elevation of about 50 ⁰, well above the horizon, and thus not due to multipath effects. Panel 2 shows the normalised power on the L1-signal (CNo) recorded at a sampling rate of 50 samples per second with the GISTM to illustrate the power dip that is associated with the S4-event. The TEC shown in panel 5 depicts the small but rapid increase in the TEC which is associated with the dip in the power, which seems to indicate that the S4-event was caused by the ray path traversing a region of increased electron density, which increased the absorption of the signal. The slope of the TEC curve is shown as the Rate of Change of TEC (ROT) in panel 6. Panel 7 shows the Rate of Change of TEC Index, which is standard deviation of the ROT, with averaging done over a 5-minute period.

Figure 12. Comparison of GISTM-observed S4 scintillation with elevation weighted S4 proxy (S4p,) as observed respectively by a Novatel GSV4004B GISTM and conventional Ashtech dual frequency GPS receiver installed on Gough Island. Observations were done along the ray path from satellite PRN18.

8 Correlation between geomagnetic disturbance and EPP

The most extreme conditions for observing particle precipitation occurs during geomagnetic disturbed conditions associated with increased solar activity as manifested in solar flares, X-ray flares or coronal mass ejections (CMEs). During such events the Earth’s geomagnetic field is severely disturbed because of the Interplanetary Magnetic

Hermanus Magnetic

Observatory

Page 21 of 71



Field (IMF)-magnetosphere coupling and enhanced equatorial ring currents. Apart from increased particle precipitation, the ionosphere is significantly disturbed on a global scale, making it difficult to distinguish between enhanced ionization due “normal” geomagnetic response or to particle precipitation. To facilitate the identification of particle precipitation occurrences during and outside storm-time days, disturbed days were identified by the Disturbance storm time index (Dst).

8.1 Disturbance storm time index (Dst) The Dst is an index of magnetic activity and is derived from a network of near-equatorial geomagnetic INTERMAGNET observatories that measure the intensity of the globally symmetrical equatorial ring current. These observatories are located at San Juan (Puerto Rico), Honolulu (Hawaii), Hermanus (South Africa) and Kakioka (Japan). Severe geomagnetic storms result in large negative Dst values. The Oct 2003 and Nov 2004 geomagnetic storms (Kp = 9) e.g. resulted in Dst values of up to -475 nT. Moderate storms (Kp = 5) have Dst values of about -50 nT. To illustrate the correlation between geomagnetic activity and EPP, hourly global proton flux measurements in the > 1, 2, 4, 10, 30 and 60 MeV range from the OMNI2 data sets were compared to disturbance storm time (Dst) values and are presented for the years 2000 (Figure 13) and 2004 (The OMNI2 proton flux values were obtained from the IMP-7 and IMP-8 satellites recorded during the Charged Particle Measurement Experiment (CPME) & Energetic Particle Experiment (EPE) from 1973-2005. [sd-www.jhuapl.edu/IMP/imp_index.html]. The correlation between increased EPP with associated large negative Dst values is clear from the figures. From these results it’s evident that increased EPP is observed during geomagnetically disturbed periods.

Hermanus Magnetic

Observatory

Page 22 of 71



Figure 13. Year 2000 Proton flux values at various energy levels (top) compared to Dst values (bottom)

Figure 14 Year 2004 Proton flux values at various energy levels (top) compared to Dst values (bottom)

30 60 90 120 150 180 210 240 270 300 330 3600

0.5

1

1.5

2

2.5

3x 104 2000 Proton flux at various energy levels (OMNI data)

Day of year

Pro

ton

flux

[1/(c

m2

sec

ster

)]

> 1 MeV> 2 MeV> 4 MeV> 10 MeV> 30 MeV> 60 MeV

0 30 60 90 120 150 180 210 240 270 300 330 360-400

-300

-200

-100

0

Dst: 2000

Dst

[nT]

Local time [day-of-year]

30 60 90 120 150 180 210 240 270 300 330 3600

0.5

1

1.5

2

2.5

3x 10

4 2004 Proton flux at various energy levels (OMNI data)

Day of year

Pro

ton

flux

[1/(c

m2

sec

ster

)]

> 1 MeV> 2 MeV> 4 MeV> 10 MeV> 30 MeV> 60 MeV

0 30 60 90 120 150 180 210 240 270 300 330 360-400

-300

-200

-100

0

Dst: 2004

Dst

[nT]


Hermanus Magnetic

Observatory

Page 23 of 71



9 Data challenges

9.1 Non-availability of data. The availability of relevant GPS data used in the study is summarized in Table 2

For some years GPS data was very sparse. Gough Island data availability varied between 67-83% over the period. No

supplementary data was available due to the isolation of the island lack of receiver redundancy on the island.

Cape Town (South Africa) data availability varied between 21-100% with data in 2003-2004 being particularly sparse. Where possible gaps in the Cape Town data were supplemented from the Simonstown and Hermanus GPS receivers, both within 100 km from Cape Town. Prior to 2005 Hermanus data was only recorded from (06:00 – 20:00 LT) daily due to CDSM surveying policy at the time. No night-time ROTI measurements were subsequently possible when Hermanus data prior to 2005 was used.

Perth data availability was mostly above 85%, except for one year (63%). Perth data was supplemented by Dongara Data for the 130 day period during 2001when Perth data was not available.

Table 2. Archived GPS data availability, in days, with percentage days available for a specific year given in brackets (). Year Receiver

2000

2001 2003 2004 2008

Gough Island (GOUG) 09:52:51.3W, 40.:20:55.2S

286 (78) 255 (70) 245 (67) 261 (72) 305 (83)

Cape Town (CTWN) 18:28:06.1E, 33:57:04.8S

336 (91) 304 (83) 126 (34) 76 (21) 365 (100)

Hermanus (HNUS) 19:13:22.3E, 34: 25:28.2S

- - 239 325 -

Simonstown (SIMO) 18:26:22.4E, 34: 11:16.4S

- - 266 57 -

Combined (CTWN, SIMO, HNUS)

336 (91) 304 (83) 334 (92) 340 (93) 365 (100)

Perth (PERT) 115:53:06.9E, 31:48:07.1S

314(86) 229 (63) 325 (89) 354 (97) 349 (95)

Dongara (YAR1) 115:20:49.1E, 29:02:47.6S

- 130 - - -

Combined (PERT,YAR1)

314(86) 359 (98) 325 (89) 354 (97) 349 (95)

Hermanus Magnetic

Observatory

Page 24 of 71



9.2 False positives associated with cycle slips. The biggest challenge in processing the GPS data was eliminating false positive scintillation events associated with cycle slips observed in the GPS L1-L2 phase data. A cycle slip is attributed to a momentarily signal loss-of-lock by the GPS receiver with an associate unknown integer ambiguity offset added to the measured L1 and L2 phase. The difference between consecutive TEC values calculated across a cycle slip may subsequently vary from a few to several thousand TECUs, leading to very large rate-of change of TEC (ROT) values across a cycle slip and subsequent large ROTI (scintillation S4 proxy) values. Cycle slips (or phase jumps) are often associated with disturbed ionospheric conditions and are indications of a scintillation event, but in this study it was found the majority cycle slips occurred randomly during non-disturbed conditions. It was also evident that, in the event of an ionospheric disturbance, the magnitude of the phase jump is not necessarily related to the severity of the event/disturbance. Large ROTI values associated with large cycle slips subsequently do not necessarily represent large ionospheric scintillations and might be misleading when interpreting results. In the initial ROTI analysis false positives constituted about 40% of the results and a general methodology had to be developed to correct for cycle slips and eliminate false positives without affecting true scintillation observations. The method was developed over several iterations as cycle slips presented it in various forms. To find a general methodology proved difficult because of the various ways cycle slips represented themselves. At times relative small cycle slips (~0.25 - 2 TECU) could occur during disturbed conditions, but such slip-values might be smaller than valid epoch-to-epoch TEC differences or disturbance amplitude (Figure 15). In such cases cycle slip correction may smooth out valid scintillation observations. At other times, very steep, but valid consecutive TEC differences (> 10 TECU) were observed during disturbed conditions, but no cycle slips occurred and care had to be taken not to eliminate valid such scintillation events. To eliminate false scintillation observations, cycle slips were identified by checking for time-gaps and large TEC differences over continuous data arcs. On identifying a cycle slip in a continuous data arc, it was corrected by adding a relevant offset. Cycle slip correction reduced the number of false positives to

Hermanus Magnetic

Observatory

Page 25 of 71



Figure 15. Example of false positive ROTI caused by cycle slips not associated with sustained perturbed ionospheric phase observation. For satellite PRN13, observed from Perth on 12 June 2008, cycle slips are observed in L1 and L2 phase as phase jumps (top), abrupt jumps in slant TEC with ~2 TECU (middle) and associated ROTI (bottom). Note the relative high ROTI values associated with these cycle slips. Such ROTI calculations typically are associated with false positive scintillation events.

14 15 16 17 180.85

0.9

0.95

1

1.05

1.1

1.15

1.2x 108

Pha

se

Phase observed for PRN 13 on 12 June 2008 (day 164) from PERTH IGS GPS Receiver

L1L2

14 15 16 17 1828

29

30

31

32

33Phase-derived Slant TEC observed for PRN 13 on 12 June 2008 (day 164) from PERTH IGS GPS Receiver

TEC

[TE

CU

]

14 14.5 15 15.5 16 16.5 17 17.5 1828

30

32

TEC

[TE

CU

]

14 15 16 17 180

5

10

RO

TI

Time [UTC hours]

ROTI and Phase-derived Slant TEC observed for PRN 13 on 12 June 2008 (day 164) from PERTH IGS GPS Receiver

Hermanus Magnetic

Observatory

Page 26 of 71



Figure 16. Example of false positive ROTI scintillation events associated with cycle slips observed on all GPS receiver channels for 12 June 2008 at Perth. Scrutiny revealed this day’s raw data was too suspicious to use and was subsequently ignored.

10 DMSP SSJ/4 EPP sensor The SSJ/4 sensor, a precipitating Electron and Ion Spectrometer, was built by the United States Air Force (USAF) Research Lab and Space Vehicles Directorate. It was designed to measure the flux of charged particles as they enter the Earth’s upper atmosphere from the near-Earth space environment. The SSJ/4 sensor has flown on board the Defence Meteorological Satellite Programme (DMSP) satellites. DMSP satellites are in a sun-synchronous, low altitude polar orbit. The orbital period is 101 minutes and the nominal altitude is 830 km. The DMSP satellites are three axes stabilised with particle detectors configured to point toward local zenith (Hardy et al., 1985). The DMSP SSJ/4 particle precipitation sensor data provide a complete energy spectrum of the low energy particles that cause the aurora and other high latitude phenomena. The data set consists of integrated electron and ion particle fluxes between 30 eV and 30 KeV and satellite ephemeris and magnetic coordinates where the particles are likely to be absorbed by the atmosphere. The differential number flux is the number of particles crossing a unit area from a unit solid angle per second at each energy level. The parameters relevant for this project are the number flux and the energy flux and are stored in 15 second intervals along with relevant satellite coordinates at observation epochs. The number flux (Nf) is derived by integrating the differential number flux across all energies. It is a measure of intensity and is independent of the energy, i.e. the number of particles crossing a unit area from a unit solid angle per second regardless of the energy. Similarly, the differential energy flux is the amount of energy crossing

0 3 6 9 12 15 18 21 240

5

10

15

20

25

30

35

40ROTI associated with cycle slips obsereved on 12 June 2008 (day 164) from PERTH IGS GPS Receiver

Time [UTC hours]

RO

TI

Hermanus Magnetic

Observatory

Page 27 of 71



the same unit area from a unit solid angle per second at each energy level, which when integrated across all energy levels, gives the energy flux (Ef), i.e. Ef is the total energy crossing a unit area from a unit solid angle per second. Dividing the energy flux by the number flux gives the average energy of the particles i.e. < E > = Ef/Nf The detectors also record high energy ions that penetrate both the satellite and the instrument. This is most noticeable in the South Atlantic Anomaly and at the "horns" of the radiation belts.

Table 3 Geographic and Geomagnetic Coordinates of key locations used in the study together with the GISTM data availability.

Location Geographic Coordinates of antenna

Geomagnetic coordinates CGM

Date since which 1 minute scintillation parameters logged

Date since which 50 Hz raw data logged

Hermanus Magnetic Observatory (HMO)

34°25'24.87"S, 19°13'24.36"E

42.27°S 83.22°E L 1.86

2010/06/01 2010/06/01

Gough Island 40°20'58.11"S, 9°52'49.17"W

42.29°S 51.09°E L 1.86

2008/09/21 2008/09/21

Note: Corrected Geomagnetic Coordinates were calculated using the on-line calculator at omniweb.gsfc.nasa.gov/cgi/vitmo/vitmo_model.cgi

11 Data processing

11.1 DMSP SSJ4 data for DMSP satellite F13 were obtained from NOAA for years 2000, 2001, 2003, 2004 and 2008. Relevant Matlab ® scripts were developed to read the ASCII orbit files to extract energetic particle flux/energy measurements and corresponding geographic (and magnetic) coordinates. Relevant close overpasses of the F13 satellite over the three locations (Gough Island, Cape Town, and Perth) were identified by a geographic window of 10˚x 10˚ (Figure 17) in longitude and latitude centred on each location. Using this window, electron and ion flux and energy values were extracted to represent particle precipitation measurements over the respective locations. All EPP values were checked for outliers and the cleaned results stored separately for each of the three locations for later comparison and correlation with GPS scintillation measurements. Examples of EPP measurements along a DMSP F13 satellite overpass are illustrated in Figure 18 to Figure 20 It was assumed that the 15-second sampled EPP data observed during the transit of each 10˚ geographic window represented EPP flux and energy over the whole 10˚ window. It was also

Hermanus Magnetic

Observatory

Page 28 of 71



assumed that EPP continues over the geographic window at similar flux and energy levels for some time (Δτ) after the satellite has passed through the window and can no longer take measurements of that geographic region. This period, Δτ, is assumed to be sufficiently long to permit EPP flux and energy levels observed over each window to be considered concurrent with the closest (5-minute interval) S4p calculation.

Figure 17. Geographic windows of 10˚ centred on the key locations

Figure 18 SSJ4 Electron and ion flux and energy levels observed at South Atlantic Anomaly region around 2001-03-16-06:41 UTC . The red line indicates the geographic latitude of the Gough Island GPS receiver used for the ionospheric scintillation calculations. Note that particle precipitation values are an order of magnitude larger than for Cape Town.

-150 -100 -50 0 50 100 150

-80

-60

-40

-20

0

20

40

60

80

Longitude [Deg]

Latit

ude

[Deg

]

Geographic windows around key locations

Gough Island

Cape Town

Perth

-60 -40 -20 00

2

4

6x 107 SAA Ne flux

-60 -40 -20 00

2

4

6

8x 108SAA Electron energy flux

-60 -40 -20 010

15

20

25SAA Electron EAVE

-60 -40 -20 00

1

2

3x 107 SAA Ni flux

Latitude-60 -40 -20 00

1

2

3x 108 SAA ION energy flux

Latitude-60 -40 -20 05

10

15SAA ION EAVE

Latitude

Hermanus Magnetic

Observatory

Page 29 of 71



Figure 19. SSJ4 Electron and ion flux and energy levels observed at South Africa around 2001-03-16-03:17UTC. The red line indicates the geographic latitude of the Cape Town GPS receiver used for the ionospheric scintillation calculations.

Figure 20. SSJ4 Electron and ion flux and energy levels observed over Australia around 2001-03-16-08:2UTC. The red line indicates the geographic latitude of the Perth GPS receiver used for the ionospheric scintillation calculations.

11.2 GPS Dual frequency GPS data in the form of Receiver Independent Exchange (RINEX) observation files were obtained for Gough Island, Simonstown, Perth and Dongara from the International GNSS Service [IGS] for the years under investigation. RINEX data for Cape Town and Hermanus GPS receivers were sourced from the South African Chief Directorate Surveys and Mapping [CDSM]. At the time none of the South African receivers contributed data towards IGS and were subsequently not available in the IGS network. All RINEX data were pre-

-50 -40 -30 -20 -10 00

1

2

3x 106 RSA Ne flux

-50 -40 -30 -20 -10 00

2

4

6x 107 RSA Electron energy flux

-50 -40 -30 -20 -10 05

10

15

20

25RSA Electron EAVE

-50 -40 -30 -20 -10 00

5

10

15x 105 RSA Ni flux

Latitude-50 -40 -30 -20 -10 00

5

10

15x 106 RSA ION energy flux

Latitude-50 -40 -30 -20 -10 00

5

10

15RSA ION EAVE

Latitude

-60 -40 -20 01

2

3

4

5x 105 AUS Ne flux

-60 -40 -20 00

5

10x 106AUS Electron energy flux

-50 -40 -30 -20 -10 010

15

20

25AUS Electron EAVE

-60 -40 -20 00

1

2

3x 107 AUS Ni flux

Latitude-60 -40 -20 01

2

3

4

5x 106 AUS ION energy flux

Latitude-50 -40 -30 -20 -10 05

10

15

20AUS ION EAVE

Latitude

Hermanus Magnetic

Observatory

Page 30 of 71



processed to remove any GLONASS data and only maintain GPS-observed L1, L2 phase and C1 and P2 code (pseudo range) data. Erroneous or corrupted data files were deleted. Orbital (sp3) position files for GPS satellites were obtained from [IGS]. RINEX and sp3 files were ordered in relevant directory structures for algorithmic batch data processing. Batch processing involved:

Reading RINEX and sp3 files Calculating phase slant TEC along signal paths (up to 32 GPS satellites are visible

daily) Calculating satellite azimuth and elevation angles as observed from the receiver

position Limiting slant TEC values to elevations > 40o to eliminate multipath effects Calculating ROTI values from standard deviation of 5-minute windowed ROT values Calculating proxy-S4 values from ROTI Storing the results in Matlab binary files

HMO-developed Matlab® scripts were used to conduct the batch processing. All GPS receivers used in this study are permanently installed, dual frequency, geodetic grade receivers operating with pillar-mounted choke ring antennas to minimize the effect of multipath signal scattering. The major applications of these receivers are space geodesy, global ionospheric monitoring and surveying. Typical receivers used to register data used in his study include the Turbo Rogue SNR-8100, Ashtech Z-XII3, Ashtech UZ-12 (micro Z-12), Ashtech Z-FX, Trimble 4000SSI and Trimble NetR5. Depending on the GPS constellation geometry, at any given time 3-12 satellites are visible above the local horizon and each of these receivers can simultaneously record observables of up to twelve GPS satellites. It is subsequently possible to have up to 12 S4p proxy values per receiver at each 5-minute epoch, supplying sufficient redundancy in the observation data.

11.3 Integrating SSJ/4 and S4p data The DMSP F13 satellite orbits transits the selected geographic windows twice daily, in the morning and evening. Because of the satellite’s relative high ground velocity, typically fewer than 20 data points are registered during each window transit. To meaningfully compare and correlate the sparse 15-second sampled EPP observations with up to 12 S4p (ROTI) calculations for each 5 minute epoch, three approaches were followed: visual comparison, statistical correlation of median values and statistical correlation of daily integrated values.

11.3.1 Visual comparison For each 5-minute S4p (ROTI) data epoch, all available S4p (ROTI) values were binned into the median S4p value for that 5 minutes. This gave a representative scintillation value for each epoch and served as a non-linear filter to eliminate spurious outliers. For each year, the SSJ/4 Ne and Ni flux, S4p (ROTI) and Dst were then plotted on a four-pane graph (e.g. Figure 21). At this point all available EPP observations were included in the comparison. This visual comparison permitted an immediate overview of possible correlations and proved valuable in

Hermanus Magnetic

Observatory

Page 31 of 71



identifying outliers and interesting events and for investigating and remedying suspicious S4p (ROTI) values.

11.3.2 Statistical correlation of median values To quantitatively correlate EPP and scintillation values for each relevant geographic key point, it was required to first identify the median S4p epochs (at 5-minte intervals) which closely matched EPP observation epochs during each window transit. F13 satellite total window transit times were typically less than two minutes with data sampled at 15-second intervals during this period. It was subsequently necessary to derive a single representative EPP value for the transit period which would then be correlated with the closest S4p 5-minute spaced value. For each year, this single EPP value was obtained by taking the median value for each of the electron and ion flux and energy values for each relevant key point window transit period in that year. Corresponding times in the median S4p values for that year were then identified. A Linear regression analysis was conducted using the median EPP and median S4p data sets for each year. Using the regression coefficients, a regression line was fitted to the data and a correlation coefficient calculated. An example result is illustrated in Figure 22

11.3.3 Statistical correlation of integrated values Because of the limited temporal daily coverage of F13 over each key position, the daily integrated EPP flux was also statistically correlated with the daily integrated 5-minute median S4p (∫S4p) values for each key point, as a measure to provide for possible miss-correlation between S4p and EPP. Regression analysis was conducted and correlation coefficient calculate to quantify the correlation between, respectively, the daily integrated median S4p and the daily integrated electron flux (Ne), and the daily integrated median S4p and daily integrated EPP ion flux (Ni). An example result of correlations of daily integrated values is illustrated in Figure 21

12 Results All the figures illustrating the visual and statistical comparison and correlation between EPP and S4p scintillation proxy are given in Appendix A. The correlation coefficients calculated between EPP and S4p for all three locations for the years under consideration, are given in Table 4

Hermanus Magnetic

Observatory

Page 32 of 71



Figure 21. Year 2003 Gough Island electron (Ne) and ion (Ni) flux (panes 1-2), 5-minute median S4p (ROTI), pane 3) and Dst (bottom). Note the large S4p values around day 319 which appears to coincide with a large Dst value and apparent increased electron and ion flux.

0 30 60 90 120 150 180 210 240 270 300 330 3600

1

2x 107 Gough annual Ne flux: 2003

Nf [

cm-2

.s-1

.sr-1

.eV

-1]

0 30 60 90 120 150 180 210 240 270 300 330 3600

1

2x 107 Gough annual Ni flux: 2003

Nf [

cm-2

.s-1

.sr-1

.eV

-1]

30 60 90 120 150 180 210 240 270 300 330 3600

10

20

30Gough annual ROTI binned in median of 5-minute intervals : 2003

RO

TI

0 30 60 90 120 150 180 210 240 270 300 330 360-400

-200

0

Dst: 2003

Dst

[nT]


Hermanus Magnetic

Observatory

Page 33 of 71



Figure 22. Year 2003 statistical correlation of Gough Island electron (Ne) and ion (Ni) flux with 5-minute median S4p (ROTI)

Figure 23. Year 2003 Statistical correlation between, respectively, daily integrated electron flux (top) and ion flux (bottom) and S4p

0 1 2 3 40

2

4

6

8

10x 10

6 Gough Ne flux - ROTI correlation. 2003

median ROTI

med

ian

Ne

flux

y = 147411.6353x + 1593755.4189Corr. Coeff = 0.026011

0 1 2 3 40

5

10

15x 10

7

Gough Ne Energy - ROTI correlation. 2003

median ROTI

med

ian

Ne

Ene

rgy

y = 3479779.1814x + 10172061.302

Corr. Coeff = 0.049723

0 1 2 3 40

2

4

6

8

10

12x 10

6 Gough Ni flux - ROTI correlation. 2003

median ROTI

med

ian

Ni f

lux

y = 2967304.0588x + -55086.4871


0 1 2 3 40

2

4

6

8

10x 10

7 Gough Ni Energy - ROTI correlation. 2003

median ROTI

med

ian

Ni E

nerg

y

y = 10213721.9427x + 2754569.6791


0 20 40 60 80 100 120 140 160 180 2000

1

2

3

4

5

6x 107

Daily integrated median ROTI

Dai

ly in

tegr

ated

Ne

flux

Gough Daily integrated Ne flux - ROTI correlation. 2003

y = 3680.8355x + 17060364.9888

Corr. Coeff = 0.0087891

0 20 40 60 80 100 120 140 160 180 2000

2

4

6

8x 107


Dai

ly in

tegr

ated

Ni f

lux

Gough Daily integrated Ni flux - ROTI correlation. 2003

y = 65153.1734x + 3967880.0369


Hermanus Magnetic

Observatory

Page 34 of 71



Table 4. Correlation coefficients between S4p and EPP. (∫S4p represents integrated S4p) S4p

2000 ∫S4p 2000

S4p 2001

∫S4p 2001

S4p 2003

∫S4p 2003

S4p 2004

∫S4p 2004

S4p 2008

∫S4p 2008

GOUG Ne Flux

-0.06 -0.14 0.01 -0.04 .02 .01 .04 .06 -0.04 0.06

GOUG Ne Energy

-0.07 x 0.08 x .05 x .00 x -0.05 x

GOUG Ni Flux

-0.07 -0.13 0.21 0.23 .21 .13 .05 .10 -0.03 0.04

GOUG Ni Energy

-0.08 x 0.11 x .13 x .03 x -0.03 x

CTWN Ne Flux

-0.04 -0.11 -0.14 0.01 -0.11 -0.17 0.11 -0.04 0.07 0.17

CTWN Ne Energy

-0.05 x -0.13 x -0.04 x 0.03 x 0.08 x

CTWN Ni Flux

-0.06 -0.13 -0.07 0.05 0.29 0.19 -0.07 -0.07 0.04 0.16

CTWN Ni Energy

-0.07 x -0.13 x 0.03 x 0.00 x 0.06 x

PERT Ne Flux

0.20 0.27 0.01 0.07 -0.05 0.11 0.08 0.00 0.08 0.08

PERT Ne Energy

0.07 x -0.05 x 0.00 x -0.01 x 0.08 x

PERT Ni Flux

0.22 0.13 -0.10 -0.09 0.07 0.44 0.10 -0.06 0.06 0.11

PERT Ni Energy

0.01 x 0.02 x 0.29 x 0.09 x 0.05 x

Hermanus Magnetic

Observatory

Page 35 of 71



13 Discussion The results in Table 4 suggest that statistical correlation coefficients between the observed EPP flux and energy measurements and scintillations in this study are generally very low. The highest correlation coefficient is 0.44 (Perth 2003, integrated ion flux-scintillation) and all other ion-scintillation correlations below 0.3. The largest ion-scintillation correlation coefficients (> 0.2) were observed close to solar maximum (Gough 2001, 2003; Cape Town 2003; Perth 2000, 2003). The largest electron – scintillation correlation value was 0.27 (Perth (2000, integrated electron flux).

A number of days with interesting scintillation and precipitation events were identified in the visual comparison. Results for two events in respectively 2001 (Figure 24) and 2003 (Figure 25) illustrate that scintillation and precipitation observations were not correlated with scintillation typically occurring before precipitation: On day 90 of 2001 a severe geomagnetic disturbance was observed (Dst = -387 nT). Visual analysis of EPP and S4p for days 88-94 (Figure 24) revealed that though significant scintillations were observed at Cape Town for this day, no increased EPP was observed. Conversely, when increased EPP was observed on day 94, no scintillations of significance were observed. No EPP data was available for Perth on day 90 when scintillations were observed, but on days 93 and 94, when increased EPP was observed, as for Cape Town, no associated scintillations were observed. Additional days with similar results (included in the Appendix) are also mentioned. This list is ordered by year, but is not exhaustive. Cape Town 2000 Day 329: Increased S4p, but no noticeable EPP increase. Relatively small Dst decrease (-50 nT). Dst only decreased on the following day. Cape Town 2001 Day 116, S4p increased, but EPP only increased on day 117. Day 257 seems to show simultaneous S4p and PP, however day 263 shows increased EPP, but no increased S4p. Day 302 has increased S4p, but no increase in EPP. Gough 2001 Day 255: Increased EPP, but no increased S4p. Day 348-365, increased EPP, but no increased S4p Perth 2001 Day 249. Increased EPP, but no increased S4p Day 255. Increased EPP, but no increased S4p.

Hermanus Magnetic

Observatory

Page 36 of 71



Figure 24. Year 2001 days 88-94 illustrate Cape Town and Perth observations of proton flux (black crosses) compared with S4p scintillation (blue dots). No GPS data was available for Gough Island for this period, but proton flux is illustrated for completeness.

Cape Town 2003 Day 94: Increase in EPP, no S4p, increase Day 301: S4p, increase, but EPP only increases on day 307 (Figure 25) Cape Town 2004 Day 313: Both EPP and S4p increases (S4p only nominally). Dst -286 nT. Day 319: EPP increase rapidly, no S4p increase

88 89 90 91 92 93 94 950

0.5

1

1.5

2

2.5

3Cape Town Ni flux and S4p: 2001

5-m

inut

e m

edia

n S

4p

88 89 90 91 92 93 94 950

0.5

1

1.5

2

2.5

3x 107

Pro

ton

flux

[1/(c

m2

sec

ster

)]

88 89 90 91 92 93 94 950

5

10Perth Ni flux and S4p: 2001

5-m

inut

e m

edia

n S

4p

88 89 90 91 92 93 94 950

1

2x 107

Pro

ton

flux

[1/(c

m2

sec

ster

)]

88 89 90 91 92 93 94 950

0.5

1Gough Island Ni flux and S4p: 2001

Day of year [Local time]88 89 90 91 92 93 94 95

0

5

10x 107

Pro

ton

flux

[1/(c

m2

sec

ster

)]

Hermanus Magnetic

Observatory

Page 37 of 71



Figure 25. Year 2003 days 303-307 illustrate Cape Town, Perth and Gough island observations of proton flux (black crosses) compared with S4p scintillation (blue dots). Minimum Dst on day 303 was -383 nT

Gough Island 2004 Day 209: Dst decreased to min of -197 nT, and S4p noticeably increased to a maximum of 1.2. EPP, however, remained nominal on this day and only started increasing from day 212-214, during recovery phase, i.e. 3 days after the Dst minimum. Day 313: Dst decreased to -370 nT and Ni-EPP increased as well. Unfortunately no GPS data was available for this day. However, from day 316-319, 3 days after Dst minimum, there was a rapid increase in PP, but no significant S4p increase. Cape Town 2008 A few S4p events, but no associated EPP increase Gough 2008 No significant S4p or EPP events Perth 2008 Day 60. Increase in S4p, no EPP increase Day 87: Increase in S4p, no EPP increase

303 303.5 304 304.5 305 305.5 306 306.5 3070

1

2

3

4

5Cape Town Ni flux and S4p: 2003

5-m

inut

e m

edia

n S

4p

303 303.5 304 304.5 305 305.5 306 306.5 3070

1

2

3

4x 106

Pro

ton

flux

[1/(c

m2

sec

ster

)]

303 303.5 304 304.5 305 305.5 306 306.5 3070

1

2

3

4

5Perth Ni flux and S4p: 2003

5-m

inut

e m

edia

n S

4p

303 303.5 304 304.5 305 305.5 306 306.5 3070

0.5

1

1.5

2

2.5

3x 108

Pro

ton

flux

[1/(c

m2

sec

ster

)]

303 303.5 304 304.5 305 305.5 306 306.50

1

2

3

4

5Gough Ni flux and S4p: 2003

5-m

inut

e m

edia

n S

4p

Day of year [Local time]303 303.5 304 304.5 305 305.5 306 306.5

0

5

10x 106

Pro

ton

flux

[1/(c

m2

sec

ster

)]

Hermanus Magnetic

Observatory

Page 38 of 71



14 Conclusions During the period 2000-2008 several events with increased EPP were identified. There were also a number of occasions with L-band scintillation, but the scintillation events were not in general correlated with the increased EPP. The correlation between the 5-minute median S4p and EPP as well as between the daily integrated scintillation and EPP was investigated by means of time series plotted on the same time scale and scatter plots of concurrent events. The magnitude of the correlation coefficients derived from the scatter plots were less than 0.1 for most of the locations and years for both electron and proton precipitation. The low correlation derived from the scatter plots, together with the lack of synchronicity of EPP and precipitation in the time-series analysis, is taken to indicate very little or no correlation between EPP over the SAMA and L-band scintillation observed at each of the three locations: Gough Island, Cape Town, and Perth.

15 Key Project Participants The following three key participants were responsible for executing the project..

15.1.1 Dr BDL Opperman. Co-ordinator Ionospheric Research Physicist. Hermanus Magnetic Observatory.

PhD in Ionospheric Physics (Rhodes University). Competencies. Computational Physics, Ionospheric Physics, GPS, Orbital mechanics, Digital signal processing.

15.1.2 Dr PJ Cilliers. Co-author Ionospheric Research Physicist. Hermanus Magnetic Observatory : Competencies. Computational Electromagnetics, Ionospheric Physics, GPS, Digital signal processing.

15.1.3 Dr LA McKinnell. Managing Director, Ionospheric Research Physicist. Hermanus Magnetic Observatory.

Research Associate. Department of Physics and Electronics, Rhodes University.

Key competencies. Ionospheric physics, neural networks, Ionosondes, computational physics.

Hermanus Magnetic

Observatory

Page 39 of 71



16 Appendix A: Precipitation – Scintillation correlation results

Appendix A presents all the results obtained in this study in the following formats for each of the three locations, Cape Town, Gough Island and Perth: 1. Time series of EPP flux plotted on the same time scale as the L-band scintillation index (S4p) and the geomagnetic storm index Dst. 2. Scatter plots of median EPP against median S4p. 3. Scatter plots of daily time-integrated EPP against daily time-integrated median S4p. Note: In all the figures in appendix A, the values labelled ROTI are actually the elevation-weighted S4-proxy, S4p, derived from ROTI. See Figure 11 and Figure 12 and the preceding text for the relation between S4p and ROTI.

Hermanus Magnetic

Observatory

Page 40 of 71



16.1 Gough Island 2000

0 30 60 90 120 150 180 210 240 270 300 330 3600

2

4

6x 10

6 Gough annual Ne flux: 2000N

f [cm

-2.s

-1.s

r-1.e

V-1

]

0 30 60 90 120 150 180 210 240 270 300 330 3600

1

2

3x 10

6 Gough annual Ni flux: 2000

Nf [

cm-2

.s-1

.sr-1

.eV

-1]

30 60 90 120 150 180 210 240 270 300 330 3600

2


RO

TI

0 30 60 90 120 150 180 210 240 270 300 330 360-400

-200

0

Dst: 2000

Dst

[nT]


Hermanus Magnetic

Observatory

Page 41 of 71



0 0.2 0.4 0.6 0.8 10

1

2

3

4x 106Gough Ne flux - ROTI correlation. 2000

median ROTI

med

ian

Ne

flux

y = -511154.299x + 919320.4745Corr. Coeff = -0.060973

0 0.2 0.4 0.6 0.8 10

1

2

3

4

5x 107Gough Ne Energy - ROTI correlation. 2000

median ROTI

med

ian

Ne

Ene

rgy

y = -8670742.1719x + 14049386.1293

Corr. Coeff = -0.071507

0 0.2 0.4 0.6 0.8 10

0.5

1

1.5

2

2.5x 106Gough Ni flux - ROTI correlation. 2000

median ROTI

med

ian

Ni f

lux

y = -357842.4184x + 703616.0548

Corr. Coeff = -0.074687

0 0.2 0.4 0.6 0.8 10

0.5

1

1.5

2

2.5x 107Gough Ni Energy - ROTI correlation. 2000

median ROTI

med

ian

Ni E

nerg

y

y = -3331196.3771x + 6634053.5432

Corr. Coeff = -0.076414

30 40 50 60 70 80 90 100 1100

0.5

1

1.5

2

2.5x 107


Dai

ly in

tegr

ated

Ne

flux


y = -83370.5258x + 14137079.4383

Corr. Coeff = -0.14124

30 40 50 60 70 80 90 100 1100

0.5

1

1.5

2x 107


Dai

ly in

tegr

ated

Ni f

lux


y = -48809.0907x + 9748510.6989

Corr. Coeff = -0.12841

Hermanus Magnetic

Observatory

Page 42 of 71




0 30 60 90 120 150 180 210 240 270 300 330 3600

2

4


Nf [

cm-2

.s-1

.sr-1

.eV

-1]

0 30 60 90 120 150 180 210 240 270 300 330 3600

1


Nf [

cm-2

.s-1

.sr-1

.eV

-1]

30 60 90 120 150 180 210 240 270 300 330 3600

10


RO

TI

0 30 60 90 120 150 180 210 240 270 300 330 360-400

-200

0

Dst: 2001

Dst

[nT]


Hermanus Magnetic

Observatory

Page 43 of 71



0 0.5 1 1.5 2 2.50

1

2

3

4

5

6x 106 Gough Ne flux - ROTI correlation. 2001

median ROTI

med

ian

Ne

flux

y = 29558.6754x + 2049026.7314

Corr. Coeff = 0.0050904

0 0.5 1 1.5 2 2.50

1

2

3

4

5

6x 107Gough Ne Energy - ROTI correlation. 2001

median ROTI

med

ian

Ne

Ene

rgy

y = 4633617.8032x + 9370337.5821


0 0.5 1 1.5 2 2.50

0.5

1

1.5

2x 107

Gough Ni flux - ROTI correlation. 2001

median ROTI

med

ian

Ni f

lux y = 3638630.9801x + 554585.3299


0 0.5 1 1.5 2 2.50

0.5

1

1.5

2

2.5x 107

Gough Ni Energy - ROTI correlation. 2001

median ROTI

med

ian

Ni E

nerg

y

y = 2422707.5205x + 5057671.7948


0 50 100 1500

1

2

3

4

5x 107


Dai

ly in

tegr

ated

Ne

flux


y = -16435.4629x + 23536793.375Corr. Coeff = -0.041848

0 50 100 1500

2

4

6

8

10x 107


Dai

ly in

tegr

ated

Ni f

lux


y = 142239.4246x + 3051745.4063Corr. Coeff = 0.2294

Hermanus Magnetic

Observatory

Page 44 of 71




0 30 60 90 120 150 180 210 240 270 300 330 3600

1


Nf [

cm-2

.s-1

.sr-1

.eV

-1]

0 30 60 90 120 150 180 210 240 270 300 330 3600

1


Nf [

cm-2

.s-1

.sr-1

.eV

-1]

30 60 90 120 150 180 210 240 270 300 330 3600

10

20


RO

TI

0 30 60 90 120 150 180 210 240 270 300 330 360-400

-200

0

Dst: 2003

Dst

[nT]


Hermanus Magnetic

Observatory

Page 45 of 71



0 1 2 3 40

2

4

6

8

10x 106 Gough Ne flux - ROTI correlation. 2003

median ROTI

med

ian

Ne

flux

y = 147411.6353x + 1593755.4189Corr. Coeff = 0.026011

0 1 2 3 40

5

10

15x 107


median ROTI

med

ian

Ne

Ene

rgy

y = 3479779.1814x + 10172061.302


0 1 2 3 40

2

4

6

8

10

12x 106 Gough Ni flux - ROTI correlation. 2003

median ROTI

med

ian

Ni f

lux

y = 2967304.0588x + -55086.4871


0 1 2 3 40

2

4

6

8

10x 107 Gough Ni Energy - ROTI correlation. 2003

median ROTI

med

ian

Ni E

nerg

y

y = 10213721.9427x + 2754569.6791


0 20 40 60 80 100 120 140 160 180 2000

1

2

3

4

5

6x 107


Dai

ly in

tegr

ated

Ne

flux


y = 3680.8355x + 17060364.9888

Corr. Coeff = 0.0087891

0 20 40 60 80 100 120 140 160 180 2000

2

4

6

8x 107


Dai

ly in

tegr

ated

Ni f

lux


y = 65153.1734x + 3967880.0369


Hermanus Magnetic

Observatory

Page 46 of 71




0 30 60 90 120 150 180 210 240 270 300 330 3600

5


Nf [

cm-2

.s-1

.sr-1

.eV

-1]

0 30 60 90 120 150 180 210 240 270 300 330 3600

5

10


Nf [

cm-2

.s-1

.sr-1

.eV

-1]

30 60 90 120 150 180 210 240 270 300 330 3600

5


RO

TI

0 30 60 90 120 150 180 210 240 270 300 330 360-400

-200

0

Dst: 2004

Dst

[nT]


Hermanus Magnetic

Observatory

Page 47 of 71



0 0.2 0.4 0.6 0.8 1 1.20

2

4

6

8x 106

Gough Ne flux - ROTI correlation. 2004

median ROTI

med

ian

Ne

flux

y = 399770.2873x + 638473.422Corr. Coeff = 0.042179

0 0.2 0.4 0.6 0.8 1 1.20

2

4

6

8

10x 107


median ROTI

med

ian

Ne

Ene

rgy

y = 644995.131x + 8295946.0217Corr. Coeff = 0.0048725

0 0.2 0.4 0.6 0.8 1 1.20

2

4

6

8x 106

Gough Ni flux - ROTI correlation. 2004

median ROTI

med

ian

Ni f

lux

y = 439088.1419x + 450877.7436Corr. Coeff = 0.047116

0 0.2 0.4 0.6 0.8 1 1.20

2

4

6

8x 107

Gough Ni Energy - ROTI correlation. 2004

median ROTI

med

ian

Ni E

nerg

y

y = 1943211.1991x + 4742072.0354Corr. Coeff = 0.022909

0 20 40 60 80 100 120 140 1600

2

4

6

8x 107


Dai

ly in

tegr

ated

Ne

flux


y = 16065.9x + 6566501.6309


0 20 40 60 80 100 120 140 1600

2

4

6

8x 107


Dai

ly in

tegr

ated

Ni f

lux


y = 26731.0789x + 3395421.204


Hermanus Magnetic

Observatory

Page 48 of 71




0 30 60 90 120 150 180 210 240 270 300 330 3600

1

2


Nf [

cm-2

.s-1

.sr-1

.eV

-1]

0 30 60 90 120 150 180 210 240 270 300 330 3600

5


Nf [

cm-2

.s-1

.sr-1

.eV

-1]

30 60 90 120 150 180 210 240 270 300 330 3600

1

2Gough annual ROTI binned in median

Documents

HERMANUS MAGNETIC OBSERVATORY - Public Website · 2012. 7. 1. · GSV4004B GISTM and conventional Ashtech dual frequency GPS receiver installed on Gough Island. Observations were