This paper studies the ionospheric response to an X5.7/3B solar flare that occurred at 10:03 UT on 14 July 2000. With Global Positioning System (GPS) observations, temporal evolution of the ionospheric total electron content (TEC) values was obtained within the latitude range of 30°N ∼ 45°N and the longitude range of 15°E ∼ 45°E. It was found that dayside TEC values were enhanced during the flare event, which could be as large as 5 TECU (1 TECU = 1016/m2) in regions with small solar zenith angles. The enhancement tended to depend on latitude, longitude and the solar zenith angle of the subionospheric point. However, the TEC enhancement derived from the latitude belt between 30°N and 45°N was not symmetrical about either the longitude or the local hour; it was smaller in the local morning than in the afternoon. The TEC enhancement in the Southern Hemisphere seems to be larger than that in the Northern Hemisphere for the same solar zenith angle. This implies that the background levels of the ionosphere and thermosphere had some influence on the TEC enhancement. The temporal variation of TEC shows minor correlative disturbances from 10:15 UT to 10:27 UT when the solar flare was in the maximum phase. It is likely that the minor disturbances resulted from the evolution of flare emission in the EUV domain.
 Sudden enhancements in solar X-ray and EUV radiation during solar flares can produce an immediate increase in ionospheric ionization in various degrees at different heights; altogether, they are called sudden ionospheric disturbances (SIDs). SIDs are generally recorded as sudden enhancements of total electron content (SITEC) [Mendillo and Evans, 1974; Davies, 1980], the short wave fadeout (SWF) [Stonehocker, 1970], sudden frequency deviations (SFD) [Donnelly, 1967], sudden phase anomalies (SPA) [Jones, 1971; Ohshio, 1971], and the geomagnetic solar flare effect (GSFE), etc. [Liu et al., 1996]. Two detailed papers reviewing SIDs were published by Mitra  and, recently, by Davies . The usual techniques for studying ionospheric responses to large flares include mainly methods using radars and satellite beacons. With these methods, the response of the ionosphere to flare radiation has been studied thoroughly. Two problems remain to be studied further, and they are the enhancement of electron density at different heights in the ionosphere due to the extra radiation from the flare and the spatial response of the ionosphere at different solar zenith angles and local times. Through backscattering experiments, Taylor and Watkins  measured the electron density of the lower ionosphere during one flare and found that the largest relative increase in electron density occurred at 90 km. From the time history and height variation in the enhancement of electron density, they concluded that X-rays are the main cause of flare effects at h < 150 km [Taylor and Watkins, 1970].
 The total electron content along the line of sight from a satellite to a receiver can be derived from the observation of a satellite beacon. Garriott et al. [1967, 1969] observed and discussed a sudden increase in TEC recorded by monitoring the VHF signals received from the ATS-1 satellite. Subsequently, many publications focused on TEC enhancements and the contribution from each region of the ionosphere, based on data from satellite beacons [Mendillo and Evans, 1974; Thome and Wagner, 1971]. Excellent time resolution and extensive spatial coverage make these measurements more suitable for observation of global ionospheric disturbances. TEC data provide the chief means of routinely monitoring F region flare effects. Using 20 separate TEC measurements collected from observatories in the Northern Hemisphere, Mendillo et al.  presented a synoptic picture of the ionospheric response to one large solar flare as observed in total electron content data recorded throughout the sunlit hemisphere. The results showed the advantage of this technique for revealing the global ionospheric response to a flare through this special case study. Currently, it is well known that most of SITEC is produced in the F1 and F2 regions of the ionosphere during a large solar flare. This increase was considered mainly due to the ionization caused by the increase of the solar EUV continuum emission. In addition to the increase in electron density of different ionospheric regions, TEC enhancements of the ionosphere in response to flares of different classes have also been studied. Because sudden enhancements in X-ray and EUV emission during flare events are closely correlated with solar centimeter radio bursts observed from the ground [Arnoldy et al., 1968], Matsoukas et al.  studied the correlation of solar radio bursts and the occurrence of SITEC in the ionosphere.
 Two-frequency observations of GPS signals provide a relative ionospheric delay of these electromagnetic waves traveling through a dispersive medium. The TEC along the line of sight can be derived from this ionospheric delay [Lanyi and Roth, 1988]. This method has been widely used to study ionospheric disturbances in recent years. The world-wide distribution of GPS receivers in the International GPS Services for the Geodynamics (IGS) network make it possible to draw global TEC maps [Moore, 2001]. With this network, Ho et al. [1996, 1998] studied the global distribution of TEC variations during two geomagnetic disturbances. Using a high spatial resolution GPS network over Japan, Saito et al.  observed one TID that traveled southeast from the high latitude. Afraimovich  suggested the concept of a new technique for global detection of atmospheric disturbances of natural and technological origin, and using this method, two large flares' effects on the ionosphere were studied. Using the TEC derived from GPS data recorded by the IGS network, Zhang and Xiao [2000a, 2000b] studied the global ionospheric response to two large solar flares. These consistent results showed that the method of using GPS data to study ionospheric disturbances is reliable, and its spatial coverage is more suitable than other satellite beacon methods to study global ionospheric disturbances.
 The flare that occurred on 14 July 2000 caused an impressive response of the ionosphere. The accompanying coronal mass ejection (CME) and resultant energetic proton event caused a severe ionospheric storm and one large-scale traveling ionospheric disturbance (LSTID). Using data from the Japanese high spatial resolution GPS network, we analyzed this TID [Zhang et al., 2002]. The purpose of this paper is to analyze the response of the global ionosphere to this flare event using GPS data from IGS.
 The large solar flare on 14 July 2000 was classified as an X5.7/3B by the NOAA Space Environment Center. The start, maximum, and end times of the flare were 10:03 UT, 10:24 UT, and 10:48 UT, respectively. The location of the flare on the solar disc was at (22°N 07°W). Accompanying the flare, a large CME occurred, which caused a severe geomagnetic storm in the near-Earth space on 15–16 July 2000. According to J. Allen (personal communication, 2001), the storm reached a peak 24-hour intensity of Ap* = 192 (currently the 28th largest storm since the Kp/ap indices have been established). The X-ray and EUV radiation from this flare strongly affected the sunlit ionosphere. Figure 1 shows the HF Doppler sounding record during the flare observed in Beijing (39.99°N, 116.33°E). The standard frequency signal that the HF Doppler sounding received was transmitted from China National Time Service Center located at Xi'an Observatory (34.37°N, 109.22°E), and its frequency is 10 MHz. In Figure 1, it can be seen that the SFD effect on radio waves is obvious. To analyze the response of the global ionosphere to the extra radiation emitted by this flare, we selected some special GPS sites located on the dayside during the flare. These sites are located between the latitudes of 30°N and 45°N, and between the longitudes of 15°E and 45°E, for which the local time was about noon during the flare. The TEC obtained from the GPS sites between 30°N and 45°N shows the ionospheric flare response at different longitudes or local times. The TEC obtained from the GPS sites between 15°E and 45°E shows the behavior of the ionosphere at different latitudes over the Northern and Southern Hemispheres. Table 1 is a list of these GPS sites and their geographic locations. At least four TEC values at different subionospheric points can be derived from one GPS site at every 30-second interval.
Table 1. GPS Site Names and Their Geographical Location
 Observations by a dual-frequency GPS receiver include the pseudorange and the carrier phase for each frequency, from which very accurate TEC can be derived. In order to compare the TEC values obtained from different satellite-receiver pairs, the TEC along each line of sight should be converted to a vertical TEC at every subionospheric point, which is the point of intersection of the line of sight and the ionospheric shell. Since the timescale of a sudden increase of TEC during a flare is usually short, the TEC enhancement caused by the extra radiation of the flare can be considered to be at the same subionospheric point. During the TEC calculation, the satellites' instrumental bias is removed, but the receivers' instrumental bias is not removed. Since this study is focused on ionospheric disturbances caused by flare radiation, only relative TEC changes or TEC enhancements during the flare are useful and it is reasonable to ignore the receiver's biases. For a detailed description of the method of calculating TEC, refer to Mannucci et al.  or Zhang and Xiao [2000c].
3. Results and Discussion
3.1. TEC Curves
Figure 2 shows the temporal TEC variations derived from GPS data observed at several GPS sites during this flare. The name of these GPS sites is indicated in each panel. The various curves in each panel show the temporal TEC variations derived from different GPS satellite data, and the three dashed reference lines in each panel indicate the flare's start, maximum, and end times, respectively. These curves show that TEC began to show the increase at about 10:15 UT, and increasing continuously until about 10:27 UT. After 10:27 UT, temporal TEC variations began to decrease and the decay of the enhancement lasted longer than one hour. Comparing with HF Doppler shift records, the beginning time of the ionospheric TEC response to the flare lagged behind the HF Doppler shift records. It is well known that the HF Doppler shift only shows the temporal differential effect of the electron density along the ray path in the lower ionosphere and the TEC shows the integral effect of the electron density along the signal path through the ionosphere. Therefore, the HF Doppler record during solar flare responds to flare more sensitively than the ionospheric TEC. It was also found that the HF Doppler record reveals obvious changes during small class flare but distinguishable change could not be seen in the temporal TEC curves for the flare with same class.
 From the temporal TEC variations obtained from different GPS sites, it can be seen that the values of TEC enhancement between 10:15 UT and 10:27 UT vary largely at different sites, some TEC enhancements are as large as 5 TECU, and others are smaller. We also obtained some of TEC curves derived from GPS observations in the nightside, and the ionospheric TEC curves did not reveal the recognizable enhancement caused by the extra flare radiation. This phenomenon can be easily understood by considering the photoionization process and the solar zenith angle at every subionospheric point during the flare. Compared with the TEC enhancements during the flare on 6 November 1997, of which the X-ray class was as high as X9.4 and the position of the flare on the solar disc was (18°S, 63°W), the values of the TEC enhancements during this flare are much larger. The largest TEC enhancements in the latitude of 40°N during the flare of 6 November 1997 were about 2.8 TECU, and the TEC rising time was only about 6 minutes [Zhang and Xiao, 2000a]. As mentioned by Matsoukas et al. , the flare radiation is not the only factor in determining the ionospheric response to the flare. Some other factors, such as the flare position on the solar disc, are also important. Compared with the flare occurred on 6 November 1997, this flare originated nearly at the Sun's central meridian line. Hence, the value of TEC enhancements is larger. Figure 3 shows the temporal TEC curves derived from GPS data of one site during the flare of 6 November 1997.
3.2. Global Distribution of TEC Enhancement
 One of the advantages for the GPS network over other satellite beacon methods in studying ionospheric responses to flares is its dense global coverage. A complete picture of the ionospheric behavior during a flare can be obtained by the data from GPS receivers located on the dayside of the Earth. Using the data from 28 GPS sites, some of which are aligned along a latitude belt between 30°N and 45°N, and others are aligned along a longitude belt between 15°E and 45°E, a large number of temporal TEC curves can be obtained. Usually, the apparent moving speed of a subionospheric point is smaller than 200 m/s. The distance through which the subionospheric point moves during a TEC increase due to the extra radiation of a flare is about 100 km. Therefore it is reasonable to assume that the response of the ionosphere to the radiation of a flare is about the same over this distance.
 In order to determine the value of the TEC enhancement caused only by the X-ray and EUV radiation of a flare, the TEC change caused by background radiation during this period should be removed according to the curve's trend before the flare occurs. Sometimes, other kinds of ionospheric disturbances such as TID exist. These types of ionospheric disturbances can also be seen in the temporal TEC variation curves and will affect the accuracy of the TEC enhancement determination. To ensure accuracy, temporal TEC curves that have these kinds of ionospheric disturbances during a flare are not used.
 Considering the factors mentioned above, the TEC enhancements caused by the radiation of the 14 July flare were derived from temporal TEC curves, as shown in Figure 4. The positions of the dots in the figure indicate the location of subionospheric points during the sudden TEC increase caused by the extra radiation of the flare. The gray-level of the dot indicates the different scale values of the TEC enhancement. The figure clearly shows the global ionospheric response to the radiation of the flare. On the whole, larger TEC enhancements appear near the regions of local noon and at lower latitudes. To make a more in-depth analysis for the relationship of TEC enhancements with longitude and latitude of subionospheric point, the TEC enhancements with respect to the longitude and the latitude are plotted in Figures 5 and 6, respectively.
Figure 5 shows the relationship of the TEC enhancement with the longitude, i.e., local time. The latitude of the subionospheric point selected for Figure 5 is between 28°N and 45°N. The dashed line indicates the solar zenith angle at different local times along the latitude line of 40°N at 10:20 UT (divided by 20, same in Figure 6). The figure shows an obvious trend: As the zenith angle decreases, the TEC enhancement increases. In addition, it can be seen that the TEC enhancement values are not symmetrical about the solar zenith angle. It seems that at the same solar zenith angle, the TEC enhancements at local morning are smaller than that at the local afternoon. For example, the value of the TEC enhancement at local time 08:00 am is about 2.5 TECU, but the value of the TEC enhancement at local time 18:00 pm is about 3.5 TECU. This tendency agrees with Matsoukas et al.'s  results but does not agree with Zhang and Xiao's [2000b]. By calculating the TEC enhancement using GPS data observed at the GPS sites located in the same latitude belt during the flare on 6 November 1997, we obtained the opposite result. That is, the value of the TEC enhancement at the local morning time was larger than that at the local afternoon time. Matsoukas et al. mentioned that besides the solar zenith angle, other factors such as the absolute value of TEC, the neutral and electron density profiles, etc. play an important role in the diurnal dependence of the SITEC occurrence. It is reasonable to expect that these factors also play an important role in the diurnal dependence of the TEC enhancement during a flare. The most obvious difference between these two flares is that one of them occurred in summer and the other in winter relative to the Northern Hemisphere. To understand the difference in the diurnal dependence of the TEC enhancement, further theoretical analyses and case studies are necessary.
Figure 6 shows the dependence of the TEC enhancement on the latitude. The longitude of the subionospheric point selected for this figure is between 15°E and 45°E, where the local time was near noon during the flare. The dashed line in the figure indicates the solar zenith angle at local noon. This figure shows that the value of the TEC enhancement during the flare depends largely on the solar zenith angle but also is not symmetrical about it. Due to the limited GPS sites in the Southern Hemisphere, the dots for the TEC enhancement there are very sparse. Still, it can be seen that the TEC enhancement in the Southern Hemisphere is larger than that in the Northern Hemisphere when the solar zenith angle is the same. According to photoionization process of the ionosphere, the value of the TEC enhancement during flare is determined by both the flux of extra flare radiation and the component densities of compositions of the upper atmosphere, such as the density of the oxygen atom. The parameters of the flare radiation should be similar if the solar zenith angle is the same. Therefore, the difference of the TEC response to flare between Southern and Northern Hemispheres at the same zenith angle should be caused by the difference of the compositions of the atmosphere. To verify this, more case studies and analyses are needed.
 In the paper by Mendillo et al.  on the ionospheric response to one great solar flare in 1972, no obvious relationship was found between the magnitudes of the increase in the total electron content and solar zenith angles at the various subionospheric points. Figure 7 shows the relationship between TEC enhancement and solar zenith angles at various subionospheric points during the flare on 14 July 2000. The dots in the figure represent the data plotted in Figure 6, the triangles in the figure represent the data plotted in Figure 5. From this figure, the obvious relationship of the values of the TEC enhancement and solar zenith angles can be found: the larger the solar zenith angle, the smaller the TEC enhancement. It can also be seen that the different values at a similar zenith angle is about 1TECU, but, at angles of about 30 degree from solar zenith, the values of TEC enhancements scatter about 2.5 TECU. By analyzing the TEC enhancement data at 30 degree of the zenith, it is found that the subionospheric points of these data are located mainly in the Southern and Northern Hemispheres near local noon, and the TEC enhancement in the Southern Hemisphere is larger than that in the Northern Hemisphere if the solar zenith angle is the same. Actually, this trend can also be seen in Figure 6.
3.3. Minor Disturbances of TEC
 The time history of the evolution of different flares differs greatly from one another. It is natural to expect that the time history of the ionospheric response to a particular flare will be controlled greatly by that flare's evolution. The timescale for a large flare's effect on the ionosphere usually lasts about one hour or more. Because of the complicated photochemical and physical processes in the ionosphere during a flare, it is impossible to determine exactly when the ionospheric response ends. If the time of flare evolution up to its peak is too long, it is difficult to determine the real value of the TEC enhancements caused by the flare's extra radiation, even if the flare's intensity is very great.
 As mentioned above, although the flare began at 10:03 UT, the recognizable response of the ionospheric TEC to the flare started at about 10:15 UT. The temporal TEC variations increased quickly thereafter, until three minutes later after the X-ray flux reached its peak. Removing the change of TEC due to background solar radiation and recombination processes occurring in the ionosphere, the largest value of the TEC enhancement due to the extra radiation of the flare is as high as 5.6 TECU. Figure 2 suggested that, in the TEC curves between 10:15 UT and 10:27 UT, some small minor disturbances exist. In order to verify whether these small disturbances did occurred in the ionosphere, the time differential results for temporal TEC variations derived from different satellite-receiver pairs are plotted in Figure 8, in which the correlation of these disturbances is obvious. Because the spatial coverage for different curves in Figure 8 is about 180 degrees of longitude, it is reasonable to assume that these minor correlative disturbances reveal the evolution of the flux of X-ray or EUV radiation emitted by this flare. Mendillo et al.  also noticed a similar phenomenon in the study of the ionospheric behavior during the great solar flare on 7 August 1972. By comparing the time rate of change of total electron content with the radio burst at 35000 MHz, they found that the agreement between them is truly remarkable. Mendillo et al. also explained this result from continuity equation considerations. In Figure 8, the flux evolution of soft X-rays (0.5 ∼ 4Å, 1 ∼ 8Å) observed from GOES satellites during the flare on 14 July 2000 is plotted too, and no obvious correlation of the rate of change of the TEC and the X-ray flux is found. The magnitude of the minor TEC disturbance in the Figure 8 is about 1015 m−2. The ionization source producing these minor TEC disturbances probably includes soft X-ray, hard X-ray and EUV. Because the soft X ray observed in GOES has no such kind of minor disturbances, and the flux of the hard X-ray can't cause the ionization as high as 1015 m−2. So we can conclude that the EUV flux must have such kind of disturbance, which is responsible for the minor TEC disturbances. Considering the soft X-rays are responsible for the ionospheric TEC enhancement below 150 km, we can also deduce that these minor disturbances occur in the higher ionospheric regions.
4. Concluding Remarks
 Using the data from 28 GPS sites located in the sunlit hemisphere between the latitudes of 30°N and 45°N and between the longitudes of 15°E and 45°E during the large flare starting at 10:03UT on 14 July 2000, many temporal TEC variation curves were obtained. The morphology of ionospheric effects associated with this flare has been presented using TEC enhancements derived from these TEC curves. It has been found that the distinguishable influence of the flare on the ionosphere in temporal TEC curves begins at 10:15 UT, then the TEC increases quickly until 10:27 UT. The largest value of the TEC enhancement due to extra flare radiation is 5.6 TECU.
 The location of the subionospheric point affects the value of the TEC enhancement caused by the flare radiation. TEC enhancements in regions having smaller solar zenith angles were larger than at in the regions having larger solar zenith angles. From the TEC enhancements at subionospheric points between 28°N and 45°N, it can be seen that the diurnal variation of TEC enhancements is obvious; the TEC enhancements at local times in the morning and afternoon are smaller than that around local noon. But the values of TEC enhancement are not symmetrical about the local noon meridian. It seems likely that the TEC enhancements in the afternoon are larger than that in the morning hours, although this is not in agreement with the result obtained from a study of the flare of 6 November 1997.
 From TEC enhancements obtained between the longitudes of 15°E and 45°E, the effect of latitude is also obvious. The higher the latitudes are, the smaller the value of the TEC enhancement. From the comparison of TEC enhancements in the Southern and Northern Hemispheres, it appears that the value of the TEC enhancement in the Southern Hemisphere is larger than the result in the Northern Hemisphere when the solar zenith angle is about the same. This result shows that the background conditions of the ionosphere and thermosphere at the time of the event are important for controlling the amplitude of the sudden increase of TEC during this solar flare.
 The obvious relationship of the values of the TEC enhancement and solar zenith angles can be found from the result that the larger the solar zenith angle, the smaller the TEC enhancement.
 Some minor correlative disturbances can be seen in temporal TEC variation curves between 10:15 UT and 10:27 UT; these disturbances occur over the large sunlit region and should have a direct relationship to the flare radiation. By comparing the time rate of change of the TEC curves with the X-ray flux obtained by GOES satellites, no correlation has been found. Hence we conclude that these disturbances are caused by the ionization of extra EUV radiation emitted during the flare, and these minor disturbances happened at higher levels of the ionosphere.
 We thank IGS for providing highly accurate GPS data. We would like to thank Joe H. Allen for constructive discussions and suggestions. The National Natural Science Foundation of China (grants 40134020 and 49990455) and the Ministry of Posts and Telecommunications of Japan jointly supported this work.