Spread F is a widely studied subject, and the occurrence of spread F is affected by many factors. One of these factors is acoustic gravity waves (AGWs) which are very important in seeding spread F. Since most of the AGWs in the ionosphere originate from the lower atmosphere, there should be some regional features of spread F due to the different meteorological or ground conditions immediately beneath the local ionosphere. In this paper, a data set with a time coverage of one solar cycle from two Chinese stations located at exactly the same latitude and a 38 degrees separation in longitude, and having sharp contrasts of ground meteorological conditions, are used to make comparisons of spread F occurrence rates. The results showed that the total number of occurrence or occurrence percentage at Changchun station (very near the coast) is always much higher than that at Urumqi station (in the very center of the Europe-Asia continent). The annual maxima of spread F occurrence are in summer and winter. Other features of spread F occurrence at these two stations are in agreement with known properties of spread F. However, the great difference of occurrence frequency between the two stations is striking and worth further studying.
 Since Booker and Wells  first reported the equatorial spread F phenomenon, much progress has been made in this field. Observations revealed the main morphological features of spread F occurrence, including its dependences on the solar cycle, season and time of the day, longitudinal and latitudinal locations and ionospheric background [Niranjan et al., 2003; Subba Rao and Krishna Murthy, 1994; Bowman, 1990; Rastogi, 1980]. Since 1970s, the midlatitude spread F has drawn the attentions of scientists. Bowman  analyzed midlatitude observational data and concluded that the spread F occurrence levels were lower for high solar-flux values and higher for low solar-flux values. He believed that these phenomena could be expected if the UA-NPD (Upper Atmosphere Neutral Particle Density) changes have an influence on the wave amplitudes of the medium-scale traveling ionospheric disturbances (TIDs). Both computer simulation [Ossakow et al., 1979] and data analysis [Niranjan et al., 2003; Bowman, 1984] showed that the spread F occurrence is closely correlated with the peak height of the F region. What is particularly interesting is that in different regions of the world the spread F often shows different occurrence levels in relation to solar activity and seasons [Earle et al., 2006; Chandra et al., 2003; Abdu et al., 1998; Saksena, 1996; Huang et al., 1987; Bhaneja et al., 2009].
 Among the above researches, one of the striking phenomena is the regional feature of spread F occurrence. It is known that there is an obvious longitudinal effect of spread F occurrence both in the equatorial regions and midlatitudes [Earle et al., 2006; Mwene et al., 2004; Abdu et al., 1992; Batista et al., 1986]. Similar study on the Asian sector was also conducted by Igarashi and Kato . It is important to study the causes of longitudinal effects of spread F occurrence in order to better understand the factors affecting the spread F occurrence.
 As a case study, this paper is aiming at presenting some new observational findings of longitudinal effects of spread F using ionosonde data from two Chinese stations. These stations have almost exactly the same geographic latitudes but with a significant difference of longitude. In particular, there is a sharp contrast of the ground conditions of these two stations. One of them (Changchun Station) is along China's coast, near the vast ocean, while the other (Urumqi Station) is in the very center of Euro-Asian continent. There have been discussions on the seeding role of AGWs in triggering spread F (for earlier results, see Booker , and more recently Nicolls and Kelley  and Xiao et al. ) and as is known, most of the AGWs in the ionosphere originate from the lower atmosphere. We hope that this case study will provide some hints about whether or not the ground meteorological conditions influence the overlying ionosphere. In section 2 the source of the data used and the method of analysis of spread F statistics are described briefly, and section 3 presents the results of observational data analysis on the spread F occurrence at the two stations. Section 4 contains discussion on the results and conclusions.
2. Database and Method of Analysis
 The data set contains hourly values of F region critical frequency (foF2) from the two stations with a time coverage of 1992–2001. This is almost one solar cycle from the descending phase to ascending phase. The type of ionosonde as well as the ionogram scaling procedure is identical for the two stations. The foF2 in a tabulated form based on the traditional ionograms are suffixed with different letters to indicate observational conditions. Suffixed letter “F” or “Q” represents spread F occurred at that hour. In our statistics, this is counted as one time of spread F occurrence and it enters into our database for further analysis as outlined here. Each individual occurrence was summed up for 1 whole day to obtain the total number of occurrences (or daily number of occurrence) to make daily, monthly and yearly statistics. For the convenience of comparison, the term “occurrence percentage” is used in some figures and this is a measure of the frequency with which the spread F has occurred at a particular hour. In other words, it is a ratio of the times of occurrence at a particular hour to the total possible counts of that hour during a given period (1 month or years). Table 1 shows the geographic and geomagnetic parameters for the two stations.
Table 1. Geographic and Geomagnetic Parameters of the Two Stations
3. Observations and Results of Analysis
Figure 1 is the daily number of spread F occurrences at the two stations and some corresponding geophysical parameters for the 10 years. It is seen obviously that on the years when solar 10.7 fluxes are low (roughly from 1994 to early 1999), the occurrences at both stations are high, thus showing a negative correlation between solar 10.7 fluxes and spread F occurrence. Figure 2 is the occurrence percentage versus local time for the 10 years, both stations have the same trend of occurrence with local time dependence of occurrence but showed a striking feature in that the occurrence number at Changchun is much larger than that of Urumqi for every local hour during the night. Figures 3 and 4 reveal the general trend of occurrence versus local time and seasons. It seems that the frequently occurring time maximizes at 0200–0400 local time (LT) and occurrence frequency tends to be higher in summer and winter seasons than in equinox. Figure 5 shows the seasonal variation of occurrence at the two stations.
 From Figures 1 to 5, the total number of spread F occurrences at the two stations has some very clear common features such as: (1) during the low solar activity years (1994–1997), spread F occurrences at both stations show apparent high value; (2) for the 10 year period the total number of occurrences at both stations show the same seasonal variations. That is, high in winter and summer, and low in equinox months; (3) local time distribution of spread F occurrence at both stations peaks around postmidnight (0200–0400 LT). These features are in good agreement with the results given by many authors in the past. See Niranjan et al. , Igarashi and Kato , Bowman , and Isao et al. .
 The most remarkable finding shown in these statistics is the large difference of the occurrence rates at the two stations. No matter what the statistics is based on, either 10 year total, or annual, seasonal, monthly or local time average, occurrence rates of spread F are higher over Changchun station than over Urumqi station. This difference indicates a very strong longitudinal effect for the two stations which have almost the same latitudes. Some details will next be discussed.
 As is shown in Figure 1, the relationship between occurrence percentage of spread F and the F10.7 is apparently anticorrelated for both stations. During high F10.7 periods (fourth panel), the percentage is under 10% for Changchun and 5% for Urumqi, while for low F10.7 years (first panel), the percentage is more than twice as high for both stations. From Figure 5, it is seen again that on average, the occurrences at Changchun are higher than those at Urumqi station at any season.
4. Conclusions and Discussion
 Longitudinal effects of midlatitude spread F have been discussed by many authors. In this case study, two stations at exactly the same latitude with longitudinal difference of about 38 degrees were chosen to further investigate spread F. These two midlatitude stations feature a very sharp contrast of the ground conditions. One of them (Changchun Station) is along China's coast while the other (Urumqi Station) is in the very center of Euro-Asian continent. Our aim is to find whether and to what extent the lower atmosphere conditions influence the ionospheric spread F. The data covered a period of almost one solar cycle from 1992 to 2001.
 The most striking fact in this statistics is that both the total number of occurrences and occurrence percentage at Changchun station are much larger than that at Urumqi station under almost all conditions, especially during the low solar activity years.
 In each of the two stations, the main features of spread F occurrence from our statistics are in good agreement with the midlatitude observations in the other sectors of the world reported by other authors as quoted above. For example, the peak hours of spread F occurrence are during the postmidnight period and there are anticorrelations between the occurrence frequency and the solar 10.7 fluxes. One of the results which should be stressed here is that the annual maxima of spread F occurrence are in summer and winter, while the occurrence is fairly low during the equinox months. This seasonal variation is the same as that pointed out by Igarashi and Kato  that remarkable occurrence peaks appear from June to July in summer and from December to January in winter, noticing that their results also came from Far East stations. Causes for such seasonal variations remain unclear.
 There are a lot of factors that influence the midlatitude spread F occurrence, such as electron density gradient, neutral wind, electric field and local geomagnetic field. At these two stations, geomagnetic latitudes are similar and no large differences between the two stations in electron density gradients are expected at the same local time although we have not verified it from the observational data. It seems that among the influencing factors, the neutral wind including the AGW probably plays an important role. Although the role of AGW in triggering spread F is still not proven yet, lots of discussions are being conducted both theoretically and observationally. For example, Abdu et al.  theoretically discussed the role of gravity waves in the equatorial region and pointed out that the zonal polarization electric field in an instability development can be significantly enhanced under the action of perturbation winds from gravity waves. Ogawa et al.  discussed the possible processes to seed plasma bubbles by comparison between scintillation activity and Earth's brightness temperature variation by meteorological satellite. It seems that among the influencing factors, the neutral wind including the AGW probably plays an important role. Observationally, the role of AGW as a seed to trigger spread F has been discussed by many authors [e.g., Nicolls and Kelley, 2005; Xiao and Zhang, 2001]. Xiao et al.  presented observational evidence of strong meteorological events such as typhoon which can cause wavelike structures (TIDs), followed by nighttime spread F. Alam Kherani et al.  in their paper concluded that the results demonstrated that the gravity wave seeding can excite the collisional interchange instability and give rise to plasma bubbles, depending on ambient ionospheric conditions.
 Now this paper provides observational facts on the systematic large difference of spread F occurrence rate at two stations with the same latitude. It remains a challenging question to explain the difference. Considering the seeding role of AGWs in the excitation of ionospheric spread F, it could be reasonably assumed that levels of AGWs activities are different in the ionospheric F regions over the two stations. It is well known that most of AGWs in the ionosphere originate from the lower atmosphere and ground conditions are one of the determining factors in generating AGWs. For the two stations, although they have the same geographic latitude, there is large disparity in their terrain topography. Changchun lies very near the coast facing a large area of sea. On the contrary, Urumqi is located in the very center of Euro-Asian continent, the largest continent of the world. This may be very important because they give different source conditions of AGW's generation, which, in turn, gives different influence on the ionospheric F region [Wan et al., 1998; Ogawa et al., 2006].
Wu et al.  reported a study on foF2 of the same two stations and showed that in general foF2 at Urumqi was greater than that at Changchun, especially during the daytime. They attributed this to the difference of neutral wind over these two stations, because neutral winds also depend on ion drag which is influenced by local geomagnetic configuration. Certainly their suggestion cannot apply to our study here because the factors which control foF2 are not exactly the same as for triggering spread F. The variation of foF2 is connected with the heights of F2 region so may also affect spread F occurrence rate but it is not the key factor in determining the difference of spread F rate here. A threshold of foF2 may be necessary for the occurrence of spread F (need further study), but is not sufficient. Statistically, indeed, foF2, hmF2 (or h'F2) are all correlated with spread F because they provide necessary background conditions for the occurring of spread F. Meanwhile, even under favorable background conditions, certain triggering factors like AGWs, reversal of electric field and so on are often needed for the onset of spread F. Considering the data we used do not cover the time period in the work of Wu et al. , we also checked the foF2 data we used in this paper. The preliminary result showed that in the period involved in this paper, there were differences of foF2 between these two stations during nighttime but these differences were not systematic, namely, no regular patterns. For example, during the periods of January–February 1993, November–December 1995, and April–May 1997, spread F occurrence rates at Changchun were all higher than those of Urumqi (Figure 1), while foF2 of Changchun for the same periods were greater than, lower than, and equal to those of Urumqi, respectively.
 Of course, this assumption should be further studied. Since direct measurements of the AGWs near the ground surface are difficult to make, because their amplitude is very small in the lower atmosphere, our observational facts in this sense indicate that ground conditions could influence the ionosphere through AGWs and thus put forth issues worth further discussion and investigation.
 The data set we used has a time coverage of one solar cycle. During this long period, a few months' data are not available. In this statistics there is lack of Urumqi data for October 1994, June–September 1995, and January, February, October, and November 1997, which is indicated in Figure 1. Although this could affect some details of the analysis at Urumqi station, on the whole the result is convincing because the amount of the lost data is small compared to the 10 years.
 From middle May to December 1996 there was an exception (and this is the only exception) that the occurrence at Changchun station is abnormally low compared with occurrence in the other low solar activity (low F10.7) years. What caused this anomaly is unclear and also needs further study.
 This work was supported by the NSFC (grants 40636032, 40774082, and 40974091) and Project Supported by the Specialized Research Fund for State Key Laboratories. We thank China Research Institute of Radio Wave Propagation for providing the ionosonde data.