High-rate sampling data of Global Navigation Satellite Systems ionospheric scintillation acquired by a network of GPS Ionospheric Scintillation and TEC Monitor receivers located in the Svalbard Islands, in Norway and in Antarctica have been analyzed. The aim is to describe the “scintillation climatology” of the high-latitude ionosphere over both the poles under quiet conditions of the near-Earth environment. For climatology we mean to assess the general recurrent features of the ionospheric irregularities dynamics and temporal evolution on long data series, trying to catch eventual correspondences with scintillation occurrence. In spite of the fact that the sites are not geomagnetically conjugate, long series of data recorded by the same kind of receivers provide a rare opportunity to draw a picture of the ionospheric features characterizing the scintillation conditions over high latitudes. The method adopted is the Ground Based Scintillation Climatology, which produces maps of scintillation occurrence and of total electron content relative variation to investigate ionospheric scintillations scenario in terms of geomagnetic and geographic coordinates, interplanetary magnetic field conditions and seasonal variability. By means of such a novel and original description of the ionospheric irregularities, our work provides insights to speculate on the cause-effect mechanisms producing scintillations, suggesting the roles of the high-latitude ionospheric trough, of the auroral boundaries and of the polar cap ionosphere in hosting those irregularities causing scintillations over both the hemispheres at high latitude. The method can constitute a first step toward the development of new algorithms to forecast the scintillations during space weather events.
 The scintillation of radio wave signals is a consequence of the existence of refractive index fluctuations associated with electron density irregularities within the ionosphere and can be described as rapid, random variations of the amplitude and phase of radio waves passing through the ionosphere. Occurrence of scintillation is characterized by considerable spatial and temporal variability, which depends on the signal frequency, local time, season, solar and magnetic activity; it depends also on the satellite zenith angle and on the angle between the raypath and the Earth's magnetic field. Scintillation is most intense in the band around 20° on either side of the magnetic equator, and in the auroral and polar cap regions [Aarons, 1982, 1993; Basu and Basu, 1985; Kersley et al., 1988; MacDougall, 1990a, 1990b].
 Ionospheric scintillations may have a considerable effect on the performance of the satellite communication and navigation. In the case of Global Navigation Satellite Systems (GNSS), such as GPS, GLONASS and the forthcoming European GALILEO, scintillation may reduce the accuracy of the pseudorange and phase measurements, consequently increasing the positioning errors. During intense scintillation events, the signal power drops below the threshold limit, the receiver loses lock to the signal and the GPS positioning is not possible, as happened, for instance, during the Halloween storm in 2003 [Webb and Allen, 2004].
 As clearly described by Wernik et al. , two ways of modeling ionospheric scintillations can be followed: (1) modeling of the wave propagation in the irregular ionosphere and (2) modeling of the climatology of scintillation. This paper contributes with some new insights supporting the development of empirical models able to reproduce scintillation climatology. Our contribution, derived from a detailed study of an entire year of TEC (total electron content) and scintillation data acquired at high latitudes of both the hemispheres, provides a picture of ionospheric scintillations even during low solar activity conditions. Our characterization allows the investigation of the seasonal and IMF (interplanetary magnetic field) Bz dependence of the distribution of the plasma irregularities causing L band scintillations at high latitudes. After our first encouraging results, obtained analyzing data acquired in the period October–December 2003 and 2008 [Spogli et al., 2009, 2010], we have applied our original method, the Ground Based Scintillation Climatology (GBSC), on a larger and less perturbed data set considering the entire 2008, year of solar minimum. To our knowledge the work here described is the first attempt to characterize the scintillation climatology analyzing such a huge ground based data set in terms of temporal and spatial coverage. Recently, Li et al.  have analyzed almost one year of data between 2007 and 2008 from Ny-Ålesund and Laserman Hills reaching interesting results that are generally confirmed in our study. By means of the GBSC method we have investigated the scintillations occurrence and TEC variations sorting the experimental information according to Altitude Adjusted Corrected Geomagnetic Coordinates [Baker and Wing, 1989] and geographic reference and further characterizing the observations through the distinction between IMF Bz positive and negative conditions and separating the seasonal patterns. Moreover, this work presents the TEC climatology not only using the time rate of change of the differential carrier phase, termed ROT, but, to our knowledge for the first time, its root mean square (ROTrms), evaluated as described in section 3. This heavy computational exercise allows to: show the different nature of the ionospheric processes favoring amplitude and phase scintillations at high latitude, demonstrate the usefulness of 2-D TEC variation maps to draw the plasma structuring configuration, confirm the seasonality of the small scale ionospheric irregularities appearance and attempt an interpretation of the E and F regions role in the production of the irregularities causing the observed scintillations. Instead of being a drawback, the very different coverage of the southern and northern observational fields of view and the higher offset between the austral geographic and geomagnetic positions (dip pole is offset by about 24°) [see, e.g., Rodger and Smith, 1989] is here considered as a good opportunity to derive the ionospheric characterization mainly from the diversity of the experimental sites. The choice of the time interval, the entire 2008, a period characterized by very quiet helio-geophysical conditions, supports the aim of drawing a ionospheric scintillations portrait into which, on average, anomalies of the plasma distribution should not be included.
 The paper starts with the introduction of the experimental observations used in the data analysis (section 2), then it describes the GBSC method (section 3) and it closes with the discussion of the results and the concluding remarks (sections 4 and 5, respectively).
2. Measurements and Parameters Adopted
 The present investigation is based on the observations acquired by means of a chain of GSV4004 GPS Ionospheric Scintillation TEC Monitors (GISTM), that consist of NovAtel OEM4 dual-frequency receivers with special firmware specifically able to compute in near real time, from 50 Hz samplings, the amplitude and the phase scintillation from the GPS L1 (1575.42 MHz) frequency signal, and the ionospheric TEC (total electron content) from the GPS L1 and L2 (1227.6 MHz) carrier phase signals. The receiver provides the amplitude scintillation by computing the S4 index, which is the standard deviation of the received power normalized by its mean value. In this case it is derived from the detrended received signal intensity. A high-pass filter was used for detrending the raw amplitude measurements [Van Dierendonck et al., 1993, and references therein]. A fixed choice of a 0.1 Hz 3 dB cutoff frequency for both phase and amplitude filtering has been used here. For a more comprehensive discussion about the effects of filtering parameters on scintillation studies the reader is referred to Forte and Radicella . Phase scintillation computation was accomplished by monitoring the standard deviation σΦ of the detrended carrier phase. A high-pass sixth-order Butterworth filter was used for detrending raw phase measurements. Both S4 and σΦ are computed over 1, 3, 10, 30 and 60 s intervals. The receiver provides also TEC and relative TEC values computed over 15, 30, 45 and 60 s [Van Dierendonck et al., 1993]. In order to reduce the impact of nonscintillation related tracking errors (such as multipath), only indices computed from observations at elevation angles αelev, calculated from the receiver to the selected satellite, greater than 20° are considered. Scintillation indices can also be projected to the vertical, in order to account for varying geometrical effects on the measurements made at different elevation angles, as in the following formulae:
where σΦslant and S4slant are the indices directly provided by the receiver at a given elevation angle along the slant path. In the two above formulae, F(αelev) is the obliquity factor, that is defined as [Mannucci et al., 1993]:
where Re is the Earth radius and HIPP is the height of the Ionospheric Piercing Point. According to the formula (19) of Rino [1979a, 1979b], which describes the signal phase variance as a function of the zenith angle, and as described by Spogli et al. , the exponent a is assumed to be 0.5, while b depends on the spectral index of the phase scintillation spectrum p, and on the anisotropy of the irregularity. Currently, the value is reasonably chosen to be p = 2.6, corresponding to b = 0.9.
 The GISTMs network considered in this investigation consists of: two receivers located at Ny-Ålesund (NYA0 and NYA1: 78.9°N, 11.9°E; CGMLat 76.0°N), one at Longyearbyen (LYB0: 78.2°N, 16.0°E; CGMLat 74.7°N), one at Trondheim (NSF1: 63.4°N, 10.4°E; CGMLat 63.0°N) in the Northern Hemisphere, observing subauroral, auroral, cusp and cap latitudes; in Antarctica one receiver is located at Mario Zucchelli station (BTN0: 74.7°S, 164.1°E; CGMLat 77.1°S), looking at cusp and cap regions, and another one is at Concordia station (DMC0: 74.7°S, 164.1°E; CGMLat 77.1°S), looking at the polar cap (Table 1 and Figure 1).
Table 1. Receiver Identifier, Location, Geographic and Geomagnetic (AACGM Corrected at the IPP - 350 km) Coordinates of the GISTM Receiver Sitesa
Days of Data
The last two columns are the available days of data of each receiver and the corresponding percentage evaluated with respect to the 365 days of the year. ID, receiver identifier.
Mario Zucchelli Station
 The data availability and distribution for each station is sketched into the pie charts in Figure 2, where the data percentage is evaluated as: N(days of data in the period for the receiver)/N(days of data in the period for all receivers). The stations concur in similar portions to cover the Northern and Southern Hemisphere fields of view, except during the January/February period when the NYA0 station miss. As NYA1 station is practically colocated with NYA0 (they are about 1 km far apart), the NYA0 data gaps do not compromise the coverage of that particular field of view. Figure 3 illustrates the data coverage over Northern and Southern Hemispheres in terms of geographic (Figure 3, top) and corrected geomagnetic (Figure 3, bottom) longitude, highlighting the different sectors of coverage over the boreal and austral regions of investigation. In particular, the geographic frames of the two hemispheres are almost complementary (Figure 3, top), while looking at the corrected geomagnetic longitude few regions of overlapping can be found (Figure 3, bottom).
 The characterization of the IMF conditions has been done through the measurements made onboard the Advanced Composition Explorer (ACE) spacecraft, orbiting around the Lagrangian L1 libration point. The field component, Bx-By-Bz, measured by the ACE Magnetic Field Experiment (MAG) [Smith et al., 1998] at L1 point must be propagated to the magnetopause to consider the time taken by the solar wind to propagate to the Earth's vicinity from the point of measurements. By consequence the universal time of the IMF measurement done at the L1, tACE, propagated to the magnetopause is delayed of Δt, such that:
where tmagnetopause is the universal time at the magnetopause, RACE is the position of the ACE satellite and vSW is the solar wind velocity. The delay Δt ranges between 30 min and 100 min with an average of 59 min, in agreement with the delay adopted by Jayachandran et al. . The solar wind velocity vSW is measured by the ACE Solar Wind Electron, Proton, and Alpha Monitor (SWEPAM) [McComas et al., 1998], while RACE is measured by both MAG and SWEPAM. These quantities measured by ACE can be expressed in terms of different coordinates systems: for the remainder of the paper we refer to the Geocentric Solar Magnetospheric (GSM) system. Data elaborated by MAG and SWEPAM are available over different time intervals: as we are dealing with a statistical representation of the IMF orientation, we choose to merge the information over 1 h.
3. GBSC Method
 Data obtained merging the observation of the network, separating the northern and the southern contribution, have been analyzed using the GBSC technique. The original GBSC method has been recently used to analyze scintillation data acquired during the descending phase of the last solar maximum with very promising results [Spogli et al., 2009, 2010]. The core of the GBSC is the maps of phase and amplitude scintillation occurrence. Starting from these quantities, calculated over the 1 min interval (see scintillation indices definition in section 2), the method produces maps of the percentage occurrence of the available scintillation indices. All the satellites in view at each epoch are considered to produce the maps. The maps can be defined in a bidimensional coordinate system expressed in terms of couples of two of the followings: geographic coordinates (latitude and longitude), Altitude Adjusted Geomagnetic Coordinates, expressed at the Ionospheric Piercing Point (magnetic latitude and magnetic longitude) [Baker and Wing, 1989], universal time and magnetic local time.
 The binning is typically selected according to the available statistics and to a meaningful fragmentation of the map. The percentage occurrence O is evaluated in each bin of the map as:
where Nthr is the number of data points corresponding to the investigated scintillation index above a given threshold and Ntot is the total number of data points in the bin. Thresholds are chosen in order to distinguish between different scintillation scenarios: for moderate/strong scintillations, typical threshold values are 0.25 radians for σΦ and 0.25 for S4, while for weak conditions, they are 0.1 radians and 0.1, respectively. In this paper we used the higher threshold to identify the areas of the ionosphere affected by moderate/strong scintillation even under quiet conditions of the ionosphere.
 To remove the contribution of bins with poor statistics the selected accuracy, defined as [Taylor, 1997]:
must be set. In the above formula σ(Ntot) = (Ntot)1/2 is the standard deviation of the number of data points in each bin. Typically, a threshold value of Rthr = 2.5–5% is a good compromise between the necessity to include every meaningful bin in the map and to avoid possible overestimations of the accuracy due to scarce statistics. For the remainder of the paper we refer to Rthr = 5%.
 In order to help identifying the electron density gradients possibly leading to ionospheric scintillation, the GBSC is supported by the information on Rate of TEC changes (ROT). ROT is computed over 1 min intervals, calculating the difference between the relative (slant) TEC values provided by the receiver (section 2), resulting in a corresponding Nyquist period of 2 min. The scale length corresponding to the Nyquist period is given by the components of the ionospheric projection of the satellite motion and the irregularities in a direction perpendicular to the propagation path. From the statistical analysis of the line of sight velocity measured by SuperDARN HF radars, the plasma convection velocity at high latitudes is known to range between 100 m/s and 1 km/s [Ruohoniemi and Greenwald, 2005]. Consequently, the vector sum of the two velocities can vary between such magnitudes even during quiet periods, then the irregularity scale lengths sampled by ROT span form few to tens of kilometers [Basu et al., 1999].
 As we are dealing with several stations that provide not-calibrated TEC and to remove the unknown bias in TEC measurements [Coco et al., 1995], this work considers only the relative variation of TEC to identify the ionospheric plasma structures possibly causing scintillations. In particular, we have produced maps of ROT mean and root mean square (rms) values defined in the same coordinates system of the scintillation occurrence maps and made applying the same thresholds on the elevation angle and on the accuracy. To produce the ROT and ROTrms maps, the distribution of all the ROT values is evaluated in each bin. The corresponding bins of the mean and root mean square maps are then filled with the distribution mean value and rms, respectively. Larger absolute values of ROT (∣ROT∣) in a bin are associated with gradients of the electron density at the above mentioned spatial scale (few to tens of kilometers). ROTrms values are associated with the intrinsic variability of the ROT in the bin: a small value indicates that the associated ROT is weakly variable, confirming the sensitivity of that bin to host gradients at that scale, while larger ROTrms indicates that the associated gradient is in a wider scale range. According to the open literature about scintillation production mechanisms, the amplitude scintillation, differently from the phase scintillation is biased by irregularities probing size (on L band) of hundreds meters [Aarons, 1997, and references therein]. In fact, the amplitude and phase response of an electromagnetic wave to structure is determined by Fresnel filtering, which strongly suppresses the contribution amplitude for scales larger than the Fresnel scale. So that, irregularities smaller than about the radius of the first Fresnel zone (Df) produce amplitude scintillation [Hunsucker and Hargreaves, 2003], considering that Df = (λ · HIPP)1/2, where λ is the radio wavelength (∼19 cm for L1), and assuming HIPP = 350 km, Df is about 250 m.
 The choice to adopt TEC derived parameters (ROT, ROTrms) is driven by our hypothesis possibly ruling the ionospheric scenario giving (or not) scintillation effects. As already described above, ROT is not a direct measure of the electron density gradients, but it can vary by an order of magnitude due to the pierce point velocity variation with the zenith angle and its dependence on the spectral index of phase. High ∣ROT∣ is expected to respond to very large scales that have no associated amplitude variation, but if associated with high ROTrms indicate that other size scales are possibly present in the bin. What we guess from the ROT and ROTrms climatology is listed in Table 3, where all the permutations of ∣ROT∣ and ROTrms are associated with the possible irregularities scale sizes and with the corresponding scintillation index expected to occur accordingly. The verification of such hypothesis is discussed in section 4.
 All the output maps of the GBSC can be characterized by the geomagnetic conditions of the geospace through the Kp index and the IMF components. The Kp index is used to identify the quiet or disturbed behavior of a given day of the considered period, accordingly with what already described by Spogli et al. [2009, 2010]. As 2008 was a very quiet year, no distinction between different values of Kp has been done in this analysis.
 The characterization through the IMF allows to produce GBSC maps considering only the contribution of GISTM data acquired under the specified condition, that is for the ith component Bi ≥ 0 or < 0, with i = x, y or z. In this paper the GBSC maps are made separating the contribution for Bz ≥ 0 and Bz < 0, by means of the ACE measurements as described in section 2 (Table 2).
Table 2. Available Data in Both Hemispheres for the Different Conditions of the IMF and for the Considered Periods
North Bz ≥ 0 (%)
North Bz < 0 (%)
South Bz ≥ 0 (%)
South Bz < 0 (%)
4. Results and Discussion
 The logical thread driving the GBSC exercise is: (1) start with a simple geographic representation of scintillation occurrence and of the TEC variations; (2) sort the results according to a geomagnetic reference frame supported by the information of the auroral oval boundaries position given by the Feldstein, Holtzworth and Meng model [Feldstein, 1963; Holzworth and Meng, 1975]; (3) include, for the first time in the GBSC method, the distinction between IMF Bz conditions. These successive steps produce results here presented mainly in form of maps of the following parameters: S4 and σΦ > 0.25 occurrences, ROT and ROTrms.
 The low level of solar activity during 2008 can be seen as an advantage because the analysis is based on data acquired during a period in which the ionosphere should follow a quiet behavior, suitable to the climatological purpose of this study; on the other hand a low solar activity results in a low scintillation occurrence, particularly evident in the S4 occurrence maps here presented. Anyhow, as a first step toward a robust assessment of the ionospheric scintillation scenario, our discussions deals with the S4 occurrence as well, even if our considerations are referred mainly to the presence of S4 occurrence instead of discussing its rate in details as in the case of the other mapped parameters.
Figure 4 reports the occurrence of σΦ and S4 and the distribution of ROT and ROTrms as a function of geographic latitude and longitude over both the hemispheres for 2008. At first glance it is easily to recognize that over the northern regions the fields of view are approximately meridional distributed, while the southern coverage is almost zonal distributed. In this latter case the contribution coming from the DMC0 (left circle) and the one from the BTN0 station (right circle) are recognizable. Figures 5 and 6 describe the same quantities but separating the northern (Figure 5) and southern (Figure 6) seasonality, assuming as winter (summer) the period January/February and as summer (winter) the period July/August for the Northern (Southern) Hemisphere. From these maps some features are worth noting: the larger occurrence of phase with respect to amplitude scintillation is confirmed over both the hemispheres; the northern regions show a clear structure of high ROTrms around 63° of latitude in wintertime (Figure 5, left), narrower in latitude and longitude during summer (Figure 5, right). As ROTrms provides an information on the spreading of the ROT distribution, its enhancement observed over the Northern Hemisphere clearly identifies a narrow region where the irregularities at all scale sizes can exist (as expected from Table 3). In agreement with what described by radio tomography techniques by Pryse et al. [2005, 2006], we suppose that this region signs the position of the high-latitude trough: a persistent feature of the auroral ionosphere. As clearly stated by Pryse et al. , other concurrent mechanisms could play a role in that area, because the main trough forms at the interface between the midlatitude ionosphere and the auroral region as a result of complex interplay between different geophysical processes. Other concurrent mechanisms, such as high density solar EUV ionized plasma transport to and away from the polar cap [Foster et al., 2005], could play a role in the formation of that high ROTrms zone, but it is beyond the capability of our method to provide a more detailed explanation of what found. The southern data coverage does not allow observation of the subauroral sector in which the trough should reside.
Table 3. Expected Relationship Between ROT and ROTrms Permutations, Associated Irregularities Scale Sizes and Scintillation Phase and/or Amplitude Occurrence
predominant few kilometers scale
little few kilometers scale
 Analogously to Figures 4–6, in Figures 7–9 we have reported the same quantities versus the Corrected Geomagnetic Latitude (CGMLat) and the Magnetic Local Time (MLT), superimposing also the auroral oval boundaries given by the Feldstein, Holzworth and Meng model [Feldstein, 1963; Holzworth and Meng, 1975] for quiet and moderate magnetic activity levels (IQ = 0, IQ = 3). Sorting the climatology according to the geomagnetic reference some interesting features are shown up (Figure 7): an higher occurrence of the phase scintillation in the premidnight sector in the poleward edge of auroral boundary (IQ = 3), clearly visible over the north and envisaged also over the Southern Hemisphere in spite of the poor auroral coverage over Antarctica. Such results confirm our expectation (Table 3): regions with significant gradients (high ∣ROT∣) and low ROTrms are phase scintillations effective. The clear cusp signature in the southern ROT map between −74° and −78°CGMLat between 10:00 and 15:00 MLT with an associated low ROTrms produces, as expected (Table 3), a colocated phase scintillation increase (Figure 7, right). The northern ROT map (Figure 7, left) reveals the presence of TEC gradients at very high latitude around 82° in the afternoon and in the premidnight with ROTrms values varying around 0.5 TECU/min, likely indicating the presence of polar cap patches, and resulting in phase and amplitude scintillations can be possibly related to the cases 1 and 2 of Table 3. The northern ROT map of Figure 7 points out the high-latitude trough, visible in the postnoon northern sector between 66° and 68°CGMLat as a TEC depletion. The ROTrms maps in Figure 7 highlights the high variability of the ROT at the equatorward edge of the trough over the north (left) and a lower variability of ROT within the Northern and Southern Hemisphere cusp. Over the northern regions the amplitude scintillation occurrence (Figure 7) results to be very low and detectable around 60°, at noon and midnight where the combination of small TEC gradients and large ROTrms values occurred (verifying case 4, Table 3). At 70°, at the poleward edge of the trough, we observe high TEC gradients and low ROTrms values resulting in both (stronger) phase and (weaker) amplitude scintillations, confirming only partially our expectations (case 2, Table 3). The presence of the high-latitude trough below 68° is clearly highlighted in the winter map showing a change of the ROT sign at 12:00 MLT (Figure 8), confirming and supporting what expected in the works by Whalen  and Rodger et al.  and also recently verified by the radio tomography [Pryse et al., 2005]. Both the winter and summer ROTrms maps (Figure 8) indicate a high variability of the TEC gradients distribution on the equatorward edge of the trough around 62°CGMLat. The northern maps (Figure 8) show a slightly higher phase scintillation occurrence during wintertime, mainly confined in the auroral premidnight sector, confirming that maximum occurrence of scintillation activity should be recorded during months of little or no sunlight at F regions altitudes [see, e.g., Aarons, 1982]. The amplitude scintillation occurrence over northern regions is so low, especially during the northern wintertime to make very hard any physical interpretation of the maps (Figure 8), the only consideration is that to the northern S4 occurrence along the entire 2008 (Figure 7) contributes almost solely the summer rate, mainly located around 70°CGMLat, where the ROT and ROTrms, according to our expectation (case 2, Table 3), should favor mainly phase scintillations, and well inside the cap (above 80°CGMLat) in the afternoon and premidnight sectors (Figure 8). Figure 9 shows the seasonal behavior over the Southern Hemisphere: the cusp signature is recognizable at first glance in the summer plots of the phase scintillation occurrence and of the ROT map, likely mirroring the existence of the E layer and the higher F region ionization due to the 24 h persistence of solar irradiation. It is worth reminding that the southern cusp contribution comes from BTN0 whose geographic position is significantly shifted with respect to its magnetic location (Table 1), this should mean a different ionospheric characterization resulting from the interplay between solar irradiation and magnetosphere-ionosphere coupling respect to what observed over north. It is also worth noting the low values of ROTrms above −82°CGMLat (Figure 9) during wintertime, reflecting the quietness of the polar cap ionosphere over DMC0 during no sunlight months in 2008. The (poor) S4 occurrence observed over the southern latitudes results to be located at very high latitudes, inside the polar cap, possibly revealing the existence of small scale ionospheric irregularities, is not supported by our hypothesis (case 2, Table 3). Polar cap σΦ occurrence is also found, but its highest rate is observed during summer in the cusp region, underlining the important role of the plasma transportation and particles precipitation processes into the ionosphere as responsible of phase scintillation effects. Figures 7–9 confirm what we already found comparing the very disturbed period of October–December 2003 with the same months in 2008: within the polar cap phase scintillations are, on average, always present quite independently by the helio-geophysical conditions [Spogli et al., 2010].
 To investigate deeper on how the reconnection between the interplanetary magnetic field (IMF) and the geomagnetic field characterizes the scintillation activity we introduce the IMF Bz orientation, sorting the results according to northward and southward IMF conditions as described in section 2. The scintillation occurrence maps are reported in Figure 10, for the entire 2008, and in Figures 11 and 12 for the northern and southern seasonality respectively; the ROT and ROTrms maps are analogously reported in Figures 13–15. As expected the inclusion of the IMF orientation helps significantly the physical interpretation of the results. Figure 10 (left) shows the σΦ and S4 occurrence over the Northern Hemisphere sorted by Bz orientation: comparing the phase scintillation rate under IMF positive and negative conditions we observe that when IMF is southward the region of highest occurrence in the premidnight hours is wider in latitude and, even if with less extent, also in MLT, as expected [Aarons, 1997]; the northern and southern cusp region is affected by a significant phase scintillation activity only under Bz negative condition, when the ROT enhancement is larger and poleward extended. These results confirm the presence of electron density fluctuations already observed over Svalbard by Moen et al.  by means of EISCAT Svalbard Radar and attributed to the plasma inflow from the magnetosphere to the ionosphere at noon and to the outflow from the ionosphere to the magnetosphere just before midnight. The poleward spreading of the phase scintillations region during nighttime when Bz is negative can be explained with the presence of polar cap patches that are more likely to occur under such IMF orientation [McEwen and Harris, 1996]. Another signature of the presence of polar cap patches is the asymmetry around midnight that shows a highest phase scintillation rate in the hours before 24:00 MLT, as found by the statistical analysis of patches occurrence by Moen et al. . All these considerations are also valid for the phase scintillation occurrence over the Southern Hemisphere (Figure 10, right), even if the data coverage allows to appreciate only partially the midnight scintillations activity. For what concerns the amplitude scintillation occurrence (Figure 10) it is interesting to notice that even if poor, the S4 climatology suggests that the regions more sensitive to host small-scale irregularities result to be the subauroral and auroral areas (visible in the northern maps) and the polar cap (visible in the northern and southern maps). Moreover, the observations over both the hemispheres demonstrate that the amplitude scintillation within the cap is weakly dependent by the Bz conditions. Over the northern regions (Figure 10, left) the MLT sector sensitive to subauroral amplitude scintillation changes according to IMF orientation: when Bz is positive S4 enhances around noon, while when Bz is negative S4 peaks around noon and midnight. The cusp sector reveals to be very effective in producing phase scintillations, especially under southward IMF condition, and very ineffective in triggering amplitude scintillations, independently by the IMF orientation. Our results indicate that, statistically under solar minimum, the presence of small-scale irregularities (up to hundreds of meter) does not prevail in the cusp sector. The association between high ROTrms and the spreading of ROT distribution within the cusp could support such hypothesis: Figure 13 shows independently by the Bz sign and especially over the Southern Hemisphere, a low level of ROTrms around noon and an increase of ROT during the same hours. This could mean that at the cusp ionospheric irregularities at larger scales (ROT values correspond roughly to few/tens kilometers scales, see section 3), likely producing phase scintillations [Aarons, 1997], prevail to the ones at smaller scales (case 2, Table 3). The northern seasonality sorted by IMF orientation (Figure 11) shows a higher phase scintillation in the winter premidnight auroral sector, particularly when Bz is negative. According to the homologous ROT and ROTrms maps (Figure 14) in this region TEC gradients at few kilometers scales are almost absent with low ROTrms associated values. From our hypothesis (case 3, Table 3) we are not able to derive any explanation referred to the involved irregularities scale sizes, so we can try to attribute the phase scintillation enhancement to the high plasma dynamics typical of the southward IMF condition. Independently by the IMF orientation, we observe an absence of S4 occurrence during northern wintertime and a significant amplitude scintillation in summer, detectable especially at latitudes above 70° during nighttime and afternoon hours and at about 80° in the afternoon and pre midnight hours (Figure 11). It should indicate that in winter small-scale irregularities in the Northern Hemisphere are almost absent, while during summer such irregularities could be present where we observe amplitude scintillation. Around 60° we observe, on average, that when an area shows the presence of intense gradients (high ROT) in coincidence with a lowering of ROTrms values (Figure 14) that area, independently by the season, is ineffective in producing amplitude scintillations and effective in producing phase scintillation (case 2, Table 3). Only the MLT distribution of phase scintillation changes with the Bz sign (Figures 11 and 14). The southern seasonality (Figures 12 and 15) shows clearly a significant ROT and ROTrms enhancement in the cusp during summertime, that, under Bz negative condition is even more extended in MLT. As guessed in Table 3 (case 1), high ∣ROT∣ and ROTrms should activate irregularities at all scale sizes. From our results, only σΦ occurrence increases during summer (Figure 12), possibly confirming what found by Vickrey and Kelley  and more recently discussed by Kivanc and Heelis : during summer is present a conducting E region whose effect is to short out the electric fields followed by the increased ions diffusion across the magnetic field. Such conditions would produce a significantly bigger loss of irregularities intensities at smaller scales compared to those at larger scales resulting in steepening of the irregularity spectra. Inside the polar cap above −80° both ∣ROT∣ and ROTrms are low, and S4 occurrence shows up perhaps indicating the presence of irregularities at smaller scales; polar cap phase scintillation, as expected, results to be produced by a variety of irregularities sizes. Figure 13 shows how the ROTrms within the polar cap is IMF independent.
5. Summary and Conclusions
 Our work deals with the scintillation and TEC climatology, termed GBSC, derived from a high-latitude network of receivers acquiring 50 Hz amplitude and phase data from the GPS satellites constellation. The network consists of four receivers in the Northern Hemisphere and two in the southern one that does not allow a conjugated study of the scintillation occurrence, but that permits an investigation mainly based on the very different coverage. The northern array allows the observation of subauroral, auroral, cusp and cap sectors, while the southern array covers essentially the cusp and cap regions. The climatology has been derived using the measurements collected during 2008, a period of very quiet helio-geophysical conditions. Our climatology assesses the Rate Of TEC (ROT) scenario on irregularities scales of few to tens of kilometers and its related distribution by means of ROTrms mapping. As on L band the probing size is of the order of hundreds of meters, the ROT and ROTrms mapping should provide information on intensity and distribution of larger scale irregularities. In particular high ROTrms values would suggest the presence of other scale sizes, possibly including the smaller ones.
 The achievements of our study here presented have to be considered as preliminary but present promising potentialities for the future development of forecasting algorithms as well as for testing irregularities and scintillation models, such as: WBMOD [Secan et al., 1997, and references therein], or WAM [Wernik et al., 2007].
 The major results can be summarized as follows:
 1. At high latitudes, on average under low solar activity, scintillation is more likely to occur on the phase than on the amplitude of the received signal.
 2. Different combinations of ROT and ROTrms changes result in different scintillation scenarios: from our hypothesis, generally verified by the results, high level of ∣ROT∣ and ROTrms results in amplitude and phase scintillations; high level of ∣ROT∣ and low ROTrms values favor phase scintillations; low level of ∣ROT∣ and high ROTrms values results in amplitude and phase scintillations.
 3. Cusp sector results to be very effective in causing phase scintillations and almost ineffective in giving amplitude scintillations.
 4. Polar cap patches can cause phase and amplitude scintillations.
 5. The seasonal variability is observed mainly over the cusp, where the E layer appearance likely results in a ROT enhancement in the noon sector.
 6. The IMF orientation influences mainly the scintillations distribution in magnetic local time, highlighting the important role of the plasma inflow and outflow from and to the magnetosphere in the noon and midnight hours.
 Authors thank the Programma Nazionale di Ricerche in Antartide (PNRA), CNR (Consiglio Nazionale delle Ricerche), the Physical Sciences Research Council of the UK (EPSRC) and the Royal Society. The authors also thank the National Space Science Data Center (NSSDC) for the software calculating the auroral oval position, the model authors, Holzworth and Meng, and NASA for the ACE data.