GPS, the Ionosphere, and the Solar Maximimum
INNOVATION
GPS, the Ionosphere, and the Solar Maximum
Richard B. Langley University of New Brunswick
Oh, it was wild and weird and wan, and ever The solar maximum is upon us. It brings with called ions. Such an ionized gas is known as in camp o’ nights it higher and more variable ionospheric elec- a plasma. We would watch and watch the silver dance of tron densities that can try the souls of some The ionosphere is not bound by specific, GPS receivers. For any ranging system such fixed limits, although the altitude at which the mystic Northern Lights. as GPS, the propagation speed of the signals the ionosphere begins to be detectable is And soft they danced from the Polar sky and is of critical importance. This speed, when about 50 kilometers and stretches to heights swept in primrose haze; multiplied by the observed propagation time of 1,000 kilometers or more. The upper And swift they pranced with their silver feet, interval, provides a measure of the range. In boundary of the ionosphere is not well and pierced with a blinding blaze. the case where an electromagnetic signal defined as the electron distribution thins out propagates in a vacuum, the speed of propa- into the plasmasphere (or protonosphere, as gation is the vacuum speed of light — valid the dominant positive ions there are protons) So wrote Canadian poet Robert W. Service in for all frequencies. However, the signals and subsequently into the interplanetary the “Ballad of the Northern Lights.” The transmitted by the GPS satellites must pass plasma. northern lights, also known as the aurora through the Earth’s atmosphere on their way As altitude decreases, the ionosphere’s borealis, are a product of the complex to receivers on or near the Earth’s surface. absorption of EUV light increases as does the relationship between the Sun and the Earth. These signals interact with the constituent density of neutral atoms and molecules. The More frequent auroras at more southerly charged particles and neutral atoms and mol- net result is the formation of a maximum latitudes are evidence of the period of ecules in the atmosphere, resulting in a electron density layer. However, because dif- maximum solar activity now upon us. The changed propagation speed and direction — ferent atoms and molecules absorb EUV light solar maximum, which occurs approximately the signals are refracted. at varying rates, a series of distinct regions or every 11 years, also brings with it more active When describing atmospheric refraction layers of electron density exist, termed D, E, ionospheric conditions. The more frequent effects on radio waves, such as GPS signals, F1, and F2. The F2 layer is usually where the ejections of high-energy electromagnetic it is convenient to separate the effects of neu- maximum electron density occurs. The struc- radiation and particles from the Sun around tral atoms and molecules primarily contained ture of the ionosphere is not constant but is the time of the solar maximum results in in the troposphere (the lowest portion of the continually varying in response to changes in greater ionospheric electron densities and atmosphere) from the charged particles prin- solar radiation and the Earth’s magnetic field. more variable densities. And as the signals cipally contained in the ionosphere, the part This variability impacts the signals from from the GPS satellites must pass through this of the upper atmosphere that is significantly the GPS satellites as they pass through the more active ionosphere on their way to Earth- affected by solar activity. In recent “Innova- ionosphere on their way to users’ receivers. bound receivers, there are potential problems tion” columns, such as “Enhancing GPS: Before examining how the ionosphere for GPS users. In this month’s column, we will Tropospheric Delay Prediction at the Master affects GPS, let’s see how the Sun affects look at how solar activity affects the iono- Control Station” in the January 2000 issue of the ionosphere. sphere, how the ionosphere affects GPS, and GPS World, we have examined the tropos- how these effects can be ameliorated to reduce phere’s influence on the GPS signals. SOLAR ACTIVITY their impact. This month, we turn our attention to the In addition to ultraviolet light and x-rays, the ionosphere. Sun continuously emits a wide spectrum of electromagnetic radiation as well as particle THE IONOSPHERE radiation primarily in the form of electrons The ionosphere is the region of the Earth’s and protons, called the solar wind. The near- atmosphere where ionizing radiation primar- constant level of this radiation deviates by ily in the form of solar extreme ultraviolet only a small amount. But this variability is (EUV) and x-ray emissions causes electrons quite important, as it directly affects the par- to exist in sufficient enough quantities to ticles in the ionosphere. Heightened solar affect the propagation of radio waves. When activity is characterized by an increase in the the photons that make up the radiation number of flares, coronal holes, and coronal impinge on the atoms and molecules in the mass ejections. upper atmosphere, their energy breaks some A solar flare is a sudden energy release in of the bonds that hold electrons to their par- the solar atmosphere where electromagnetic ent atoms. The result is a large number of radiation, and sometimes energetic particles free, negatively charged electrons as well as and bulk plasma are emitted. A sudden positively charged atoms and molecules increase of x-ray emissions, resulting from a
44 GPS WORLD July 2000 www.gpsworld.com INNOVATION
300 200
200 150
100
SUNSPOT NUMBER 100 0 1750 1760 1770 1780 1790 1800 1810 1820 1830 1840 1850 DATE 50 300
0 200 1996 1998 2000 2002 2004 2006
100 Figure 2. The current sunspot cycle, number 23, is predicted to peak this year. Shown in this figure are the observed monthly SUNSPOT NUMBER 0 sunspot averages and the smoothed sunspot count predicted 1850 1860 1870 1880 1890 1900 1910 1920 1930 1940 1950 by scientists at NASA’s Marshall Space Flight Center. Also DATE shown by dotted lines is the statistical uncertainty of their
300 predictions.
cycle, shown in Fig- images of the Sun) arrive at the Earth, caus- 200 ure 1. Actually, rather ing perturbations in the Earth’s magnetic than merely counting field. The charged particles interact with the 100 individual sunspots, Earth’s neutral atmosphere, producing
SUNSPOT NUMBER solar astronomers excited ions and additional electrons. In fact, 0 compute a sunspot we can use measurements of the geomagnetic 1950 1960 1970 1980 1990 2000 2010 2020 2030 2040 2050 DATE number based on the activity as a proxy for ionospheric activity. number of sunspots The strong electric fields that are generated Figure 1. Daily observations of sunspots have been carried and sunspot groups. cause significant changes to the morphology out since 1749. Initially conducted at the Swiss Federal The day-to-day sun- of the ionosphere, changing the propagation Observatory in Zürich, the sunspot count is currently com- spot number can vary delay of GPS signals within time intervals as puted from observations by an international network of widely, so a running short as one minute. Such changes, primarily observers. Monthly averages of the International Sunspot average is typically in the polar and auroral ionospheres, can last Number show that the number of sunspots visible on the Sun computed to show for several hours. waxes and wanes with an approximate 11-year cycle. long-term trends. We As we noted in the introduction, solar are currently in cycle activity is also responsible for the colorful flare, causes a large increase in ionization in 23 and near its maximum, shown in Figure 2. auroras — the northern and southern lights. the lower regions of the ionosphere on the Another sign of increased solar activity The Earth is surrounded by a magnetic sunlit side of Earth. The x-ray radiation, trav- are coronal mass ejections (CMEs), huge cocoon (the magnetosphere) created by the eling at the speed of light, takes approxi- bubbles of gas threaded with magnetic field interaction of the Earth’s magnetic field mately eight minutes to arrive at the Earth, lines that are ejected from the Sun. (The with that of the Sun. When the solar wind is resulting in an effect soon felt after the flare corona is the outermost layer of the solar particularly strong, it appears to create tears occurs. High-energy particles can reach the atmosphere). Coronal mass ejections are in the cocoon, allowing charged particles Earth in a fraction of an hour. The lower- often associated with solar flares and promi- trapped in the magnetosphere to enter the energy particles flowing in the solar wind, on nence eruptions. The frequency of CMEs ionosphere and spiral down along the the other hand, take two to three days to reach varies with the sunspot cycle. At solar mini- Earth’s magnetic field lines. The particles, the Earth. Solar flares are usually associated mum, we observe approximately one CME primarily electrons, interact with oxygen with the strong magnetic fields that manifest per week. Near solar maximum, we observe atoms or nitrogen molecules in the atmos- themselves in sunspots — relatively cool an average of two to three CMEs per day. If a phere. Energy from an atom or molecule areas that appear as dark blemishes on the CME is directed toward the Earth, a world- excited by fast electrons is released as red Sun’s face. The sunspots are formed when wide disturbance of the Earth’s magnetic (oxygen or nitrogen) or green (just oxygen) magnetic field lines just below the Sun’s sur- field, called a geomagnetic storm, can occur. light. face become twisted and poke through the Space weather. Geomagnetic storms often Figures 3 through 6 illustrate a recent solar photosphere — the visible solar surface. result in ionospheric disturbances. Magnetic CME and its effect on the Earth’s magnetic Sunspots have been observed since at least storms and associated ionospheric storms field and atmosphere. The CME spawned the early 1600s with regular, daily sunspot occur when high-energy charged particles G4-level magnetic storms (storms or fluctua- counts beginning in 1749. The number of from solar flares, prominence eruptions, or tions in the magnetic field are rated on a scale sunspots waxes and wanes with an approxi- coronal holes (regions of exceptionally low- from 1, or minor, to 5, or extreme) as well as mately 11-year periodicity called the solar density that show up as dark areas in x-ray frequent auroras. www.gpsworld.com GPS WORLD July 2000 45 INNOVATION
REFRACTIVE INDEX So how does the ionosphere’s behavior, governed by solar activity, affect GPS? To measure the range between a GPS receiver and a satellite, we need to know precisely how fast the radio signals from a satellite propagate. We can characterize the propaga- tion speed of a radio wave (a pure carrier) in a medium such as the ionosphere by its refractive index, n, which equals the ratio of the speed of propagation in a vacuum (the vacuum speed of light), c, to the speed in the medium, v: c n= (1) Figure 6. When the shock front from v the April 4 CME reached the Earth, it Figure 3. On April 4, 2000, at 1541 The speed of propagation of a pure or triggered auroras that were seen in Coordinated Universal Time (UTC), a unmodulated wave is actually the speed at central Europe and as far south as magnetically complex sunspot region which a particular phase of the wave propa- Florida in North America. This image near the west limb of the Sun pro- gates, or the phase speed, and we refer to the shows an estimate of the auroral oval duced a large flare that was accompa- refractive index governing this speed as the on April 7, 2000 at 0242 UTC com- nied by a coronal mass ejection. This phase refractive index. A medium may be puted from observations made by the image, taken by the Extreme ultravio- dispersive, in which case the phase speed is a National Oceanic and Atmospheric let Imaging Telescope (EIT) on the function of the wave’s frequency. Administration (NOAA) Polar-orbiting Solar and Heliospheric Observatory A signal, or modulated carrier wave, can Operational Environmental Satellite (SOHO) spacecraft, shows the active be interpreted as the superposition of a group (POES), which continually monitors region near the right-side limb at of waves of different frequencies centered on the power flux carried by the precipi- about the two-o’clock position. the carrier frequency. If the medium is dis- tating protons and electrons that persive, each of the waves making up the produce the aurora.
Figure 4. The halo coronal mass ejection (CME) which occurred on April 4, 2000 was captured by the Large Angle and Spectrometric Coro- nagraph Experiment (LASCO) C2 coronagraph on the SOHO spacecraft. This running-difference image, the difference between two sequential images, shows the CME when its leading edge had already left the C2’s field of view. The solid-colored disk in the middle is an occulting disk that Figure 5. The fast-moving material in the CME launched from the Sun on April 4, blocks out the Sun’s intense light to 2000 arrived at the Earth two days later and generated a geomagnetic storm. Plots reveal the faint corona. The white of the geomagnetic field components at St. John’s, Newfoundland, show a sudden circle shows the size and position of impulse at about 1640 UTC on April 6. The field became quite disturbed shortly the Sun. thereafter.
46 GPS WORLD July 2000 www.gpsworld.com INNOVATION
group propagates at a different speed. The net nis about 0.9999838 at the GPS L1 frequency, nighttime side, in the absence of solar radia- result is that the modulation of the signal and ngis about 1.0000162 — values only tion, free electrons and ions tend to recom- propagates with a different speed from that of slightly different from unity, but nevertheless bine, thereby reducing the TEC as illustrated the carrier; this is called the group speed. significant. These refractive index values in Figure 7. The protonosphere, or upper- Corresponding to the phase refractive indicate that the phase speed of a wave in the most region of the ionosphere, may con- index, n, we can define a group refractive ionosphere is slightly greater than the vac- tribute up to 50 percent of the electron index, ng: uum speed of light, while the modulation or content during the nighttime hours. c group speed is slightly smaller. Typical nighttime values of vertical TEC ng = (2) The propagation speed of the GPS signals for mid-latitude sites are of the order of 10 17 vg varies all along their path through the iono- m-2 with corresponding daytime values of the It can be shown that sphere as the refractive index varies. The sig- order of 1018 m-2. However, such typical day- dn nals received by a GPS receiver on or near time values can be exceeded by a factor of ng =n+f (3) the Earth’s surface are affected by the cumu- two or more, especially in near-equatorial df lative or integrated effect of the ionosphere, regions. TEC also varies seasonally with where the derivative, dn/df, describes how so the time required for a signal to reach the higher values during equinoxes. the phase refractive index changes with fre- receiver is given by Spatially, at latitudes about 20 degrees quency. In general, a medium will not be n' either side of the geomagnetic equator, high homogeneous, in which case, n and n will be τ = dS (6) TEC values are produced by a so-called g c functions of position in the medium. S “fountain effect.” The effect is caused by Because of the varying refractive index, where S is the path of the signal and n´ is drifting electrons interacting with the Earth’s the signal’s path through a medium will be either the phase or group refractive index. magnetic field to produce large-scale move- bent. The path bending is a direct conse- We can convert this travel time to an equiva- ment of ionization. Polar and auroral regions quence of Fermat’s principle of least time lent distance traveled by multiplying it by the of the ionosphere also have anomalous TEC that states out of all possible paths that a vacuum speed of light. Inserting the expres- behavior but for a different reason. Globally, radio wave (and other electromagnetic waves sions for the phase and group refractive TEC is primarily determined by the Sun’s such as light) might take, it will take the path indices gives us EUV radiation, however some electrons that requires the shortest amount of time. ϭ c come from the magnetosphere and enter the The refractive index of the ionospheric ionosphere in the auroral regions as men- plasma is described by a rather complicated 40.3 Ne tioned earlier. expression developed by Edward Appleton =1– dS Although the Sun emits a broad spectrum f 2 and Douglas Hartree in the early 1930s. The (7) of radiation, as we’ve already discussed, the expression assumes an equal number of posi- S flux of the Sun’s radio emissions at a wave- tive ions and free electrons exist but that the ρ 40.3 length of 10.7 centimeters (2.8 GHz) is a use- = – Ne dS ions (being relatively massive in comparison) 2 ful indicator of solar activity. This particular f S have negligible effect on radio waves. In and wavelength was selected for monitoring pur- addition to several other constants including poses shortly after World War II, and daily the mass and charge of the electron, the ρ ρ 40.3 observations have been carried out at obser- p = + Ne dS (8) refractive index is determined by the electron f 2 vatories in Canada ever since. Recent density, the strength of the Earth’s magnetic S research has shown that while long-term field, and the frequency of the radio wave. where is the true satellite–receiver geomet- changes in the 10.7-centimeter solar flux For frequencies significantly higher than the ric range. (We have ignored the small effect seem to correspond with long-term solar plasma frequency (the “natural” frequency the bending of the signal’s path has on the cycle changes in TEC, short-term TEC at which ionospheric electrons oscillate, measured range.) Carrier-phase measure- changes are not well correlated with the day- ~ 15–30 MHz), and ignoring the effects of ments of the range are reduced by the pres- to-day changes. Only approximately 20 per- the magnetic field, the expression for the ence of the ionosphere (the phase is cent of the day-to-day fluctuations in TEC phase refractive index approximates to advanced), whereas pseudorange measure- can be attributed to changes in solar EUV. 40.3 N ments are increased (the signal is delayed) — The bulk of TEC changes result from varia- e ͐N dS n=1– 2 (4) by the same amount. s e is the integrated tions in temperature, composition, and f electron density along the signal path and is dynamics of the ionosphere. where Ne is the electronic density in recipro- called the total electron content (TEC). TEC cal cubic meters and f is the frequency in is nothing more than a count of the number of CORRECTIONS AND MODELS hertz. Using equation 3, the group refractive electrons in a column stretching through the So how does a GPS receiver cope with the index works out to ionosphere with a cross-sectional area of one ionosphere both when it is quiescent or well- square meter. behaved and when it is active? 40.3 Ne ng=1+ 2 (5) As we noted, both pseudorange and car- f TEC VARIABILITY rier-phase observations are biased by the What are typical values for n and ng? As we TEC is highly variable both temporally and presence of the ionosphere. This bias must be have mentioned, the density of electrons in spatially. The dominant variability is diurnal accounted for or position accuracy will suf- the ionosphere varies significantly in both following the variation in incident solar radi- fer. Dual-frequency receivers take advantage space and time. But taking a relatively high ation. Maximum ionization occurs at approx- of the ionosphere’s dispersive nature to cor- value of Ne of 1012 electrons per cubic meter, imately 1400 local time. On the ionosphere’s rect for its effect. A linear combination of the www.gpsworld.com GPS WORLD July 2000 47 INNOVATION
ing residual ionospheric range error results from the fact that the signals received at the two stations have passed through the iono- sphere at slightly different elevation angles. Therefore, the TEC along the two signal paths is slightly different, even if the vertical ionospheric profile is identical at the two stations. In differential positioning, the main result of this effect is a baseline shortening propor- tional to the TEC and proportional to the baseline length. Such errors can introduce significant scale and orientation biases in rel- ative coordinates. For example, at a typical mid-latitude site using an elevation cut-off angle of 20 degrees, a horizontal scale bias of –0.63 parts per million is incurred for each 1 ϫ 1017m-2 of TEC not accounted for. Never- theless, ignoring the ionospheric effect on very short baselines may be preferable to the use of the dual-frequency linear combination and the attendant higher observation noise. An empirical model also can be used to correct for ionospheric bias. The GPS broad- cast message, for example, includes the para- meters of a simple prediction model. While Figure 7. A “snapshot” of the global total electron content (TEC) at 1700 UTC on this model can sometimes account for April 6, 1995, clearly shows the ionosphere’s diurnal behavior and the equatorial approximately 70 to 90 percent of the day- anomaly spanning either side of the geomagnetic equator. Contours are in total time ionospheric delay, it was designed to 16 electron content units (1 TECU = 10 electrons per square meter). The map was remove only about 50 percent of the delay on computed using data from 74 stations of the International GPS Service network. a root-mean-square basis. More sophisticated ionospheric models, such as the Bent Iono- L1 and L2 pseudorange measurements may surement errors and other biases. spheric Model and the International Refer- be formed to estimate and subsequently In practice, N1 and N2 cannot be deter- ence Ionosphere, may not perform much remove almost all the ionospheric bias from mined, but as long as the phase measure- better than the broadcast model in part the L1 measurements: ments are continuous (no cycle slips) they because the top portion of the ionosphere is f 2 remain constant. Hence, the carrier-phase inaccurately represented. An estimate of 2 (9) measurements can be used to determine the solar activity drives these models. In the case dion,1 = 2 2 P1 –P2 +e f 2 –f1 variation in the ionospheric delay — the so- of the broadcast model, a running average of where f1 and f2 are the L1 and L2 carrier fre- called differential delay — but not the the observed solar flux values for the past quencies respectively, P1 and P2 are the L1 absolute delay at any one epoch. The estima- five days is used. But as we noted, the day-to- and L2 pseudorange measurements, and e tion of the differential delay in this way, hav- day variability of the ionosphere is not well represents random measurement errors and ing ignored third and higher order effects in correlated with variations in the flux. other biases such as satellite and receiver the expressions for the refractive indices, is Ionospheric delay corrections can be com- inter-frequency biases. (It is also possible to accurate to a few centimeters. For higher puted by a regional network of dual-fre- directly compute an ionosphere-free combi- accuracy we would need to account for the quency receivers and transmitted in real-time nation without the extra step of first comput- higher order terms neglected in the first order to single-frequency users by radio transmis- ing the ionospheric delay.) A similar approximation, including the geomagnetic sions or archived for data postprocessing. approach is used to correct carrier phase mea- field effect as well as ray path bending. This technique is used, for example, by Nat- surements with If measurements are made at only one car- ural Resources Canada for their GPS•C ser-
dion,1 = (10) rier frequency, then an alternative procedure vice and will be used for the Federal Aviation for correcting ionospheric bias must be used. Administration’s Wide Area Augmentation f 2 2 λ N – λ N – Φ – Φ + ε The simplest approach, of course, is to ignore System. 2 2 1 1 2 2 1 2 f 2 –f1 the effect. Surveyors sometimes follow this approach by employing relative positioning IONOSPHERIC SCINTILLATION where ⌽1 and ⌽2 are the L1 and L2 carrier- over short distances using single frequency If the number of electrons along a signal path phase measurements (in units of length) GPS receivers. Differencing between the from a satellite to a receiver changes rapidly, respectively, 1 and 2 are the L1 and L2 car- observations by simultaneously observing the resulting rapid change in the phase of the rier wavelengths respectively, N1 and N2 are receivers removes that part of the iono- carrier may present difficulties for the carrier the L1 and L2 integer cycle ambiguities spheric range error that is common to the tracking loop in a GPS receiver. For a GPS respectively, and e represents random mea- measurements at both stations. The remain- receiver tracking the L1 signal, a change of
48 GPS WORLD July 2000 www.gpsworld.com INNOVATION
only 1 radian of phase (corresponding to 0.19 intervals can hamper the determination of the CONCLUSION ϫ 1016 m-2 change in TEC, or only 0.2 per- correct number of integer cycles associated The solar maximum seems to be upon us. In cent of a typical 10 18 m-2 TEC) in a time with these phase discontinuities. If the varia- recent months, we have witnessed a signifi- interval equal to the inverse of the receiver tions of the ionospheric range bias exceed cant increase in solar activity with frequent bandwidth can be enough to cause problems one half of a carrier cycle, they may be solar outbursts and the attendant effects on for the receiver’s tracking loop. If the wrongly interpreted in the data processing as the Earth’s magnetic field and ionosphere. receiver bandwidth is only 1 Hz (which is a cycle slip. Although the auroras are the most visible just wide enough to accommodate the geo- Signal fading. Two regions exist where irreg- effects of this solar activity, we also witness metric Doppler shift), then when the second ularities in the Earth’s ionosphere often occur the impact on the various infrastructures of derivative of the phase exceeds 1 Hz per sec- causing short-term signal fading that can our modern society: spurious currents in ond, loss of lock will result. During such severely test the tracking capabilities of a electricity grids and pipelines, increased occurrences, the signal amplitude also gener- GPS receiver: the region extending approxi- atmospheric drag on low-orbiting spacecraft, ally fades. mately ± 20 degrees either side of the geo- and potential damage to the equipment on These short-term (less than 15 seconds) magnetic equator and the auroral and polar communications satellites. The signals from variations in the signal’s amplitude and phase regions. The fading can be so severe that the GPS satellites also can be significantly are known as ionospheric scintillations. Scin- signal level drops completely below the affected. At best, predictive ionospheric tillation effects are most pronounced in the receiver’s signal lock threshold. When this models lose their accuracy; at worst, a early evening hours and may affect GPS occurs, data are lost until the receiver reac- receiver may not even be able to track the receivers differently, depending on their quires the signal. The process of loss and signals. hardware and software differences. Some reacquisition of signals may go on for several But a silver lining for GPS does exist. The codeless dual-frequency receivers, for exam- hours. higher and more variable total electron con- ple, are particularly susceptible to loss of Such signal fading is also associated with tent values offer scientists an enhanced lock problems on the L2 signal when rapid geomagnetic storms. Occasionally, magnetic opportunity to study the ionosphere and its fluctuations exist in ionospheric electron storm effects extend to the mid-latitudes. complex relationship with solar activity. And content. Generally, receivers making carrier- During the storm that occurred in March with better understanding of the ionosphere phase measurements are more susceptible to 1989, during the previous solar maximum, and further improvements in hardware and scintillations than code-only receivers. range-rate changes produced by rapid varia- software, at the next solar maximum in about A temporary loss of lock results in a phase tions in TEC exceeded 1 Hz in one second. 11 years’ time, GPS receivers may be better discontinuity or cycle slip. A cycle slip As a result, GPS receivers with a narrow 1 Hz able to weather the space storms that it will must be repaired before the data following bandwidth continuously lost lock during the inevitably bring. the slip can be used. Large variations in worst part of the storm because of their ionospheric range bias during short time inability to follow the changes. ACKNOWLEDGMENTS The figures used to illustrate this article were kindly provided by the Solar Physics Group FURTHER READING Ⅲ “GPS and the Ionosphere,” by A.J. Mannucci, B.A. Iijima, U.L. Lindqwister, X. Pi, of Marshall Space Flight Center’s Space Sci- To learn more about solar activity and the ence Department (Figures 1 and 2); the ionosphere, see L. Sparks, and B.D. Wilson, Chapter 25 in Review of Radio Science: 1996-1999, edited Ⅲ Handbook of Geophysics and the Space SOHO EIT and LASCO consortia (Figures 3 by W.R. Stone for the International Union of Environment, edited by A.S. Jursa, the Air and 4); the Geological Survey of Canada Radio Science, Oxford University Press, 1999; Force Geophysics Laboratory, Hanscom AFB, (Figure 5); the Space Environment Center, pp. 625-665. Massachusetts, 1985. National Oceanic and Atmospheric Adminis- Ⅲ Introduction to the Space Environment, There is a wealth of current and interesting tration (Figure 6); Attila Komjathy, Univer- 2nd edition, by T. F. Tascione, Krieger material about space weather and the solar sity of Colorado (Figure 7). Thanks to Publishing Company, Malabar, Florida, 1994. maximum on the World Wide Web, for Ⅲ “Eye on the Ionosphere,” a column in Anthea Coster for helpful comments on a example ■ GPS Solutions, published quarterly by John draft of this article. Ⅲ Space Environment Center, National Wiley & Sons, Inc., New York. Oceanic and Atmospheric Administration (providing space weather alerts and warnings “Innovation” is a regu- For previous “Innovation” columns about and related educational material) lar column featuring the ionosphere, see