Ionospheric delay gradient monitoring for GBAS by GPS stations near Suvarnabhumi airport, Thailand

PUBLICATIONS RESEARCH ARTICLE Key Points: Ionospheric delay gradient observed in Thailand during plasma bubble occurrences Data analysis procedure for ionospheric delay gradient estimation Correspondence to: S. Rungraengwajiake, sarawootersky@gmail.com Citation: Rungraengwajiake, S., P. Supnithi, S. Saito, N. Siansawasdi, and A. Saekow (2015), Ionospheric delay gradient monitoring for GBAS by GPS stations near Suvarnabhumi airport, Thailand, Radio Sci., 50, 1076 1085, doi:10.1002/ 2015RS005738. Received 4 MAY 2015 Accepted 21 SEP 2015 Accepted article online 25 SEP 2015 Published online 27 OCT 2015 Ionospheric delay gradient monitoring for GBAS by GPS stations near Suvarnabhumi airport, Thailand Sarawoot Rungraengwajiake 1, Pornchai Supnithi 1, Susumu Saito 2, Nattapong Siansawasdi 3, and Apitep Saekow 4 1 Faculty of Engineering, King Mongkut s Institute of Technology Ladkrabang, Bangkok, Thailand, 2 Communication, Navigation and Surveillance Department, Electronic Navigation Research Institute, Tokyo, Japan, 3 Air Navigation Radio Aids Department, Aeronautical Radio of Thailand, Bangkok, Thailand, 4 Faculty of Science and Technology, Stamford International University, Bangkok, Thailand Abstract Ground-based augmentation system (GBAS) is an important augmentation system that provides the differential corrections and integrity information from the reference stations to the aircrafts for precision approach and landing. It is known that the nonuniform ionospheric characteristics called ionospheric delay gradient can cause the errors in differential corrections degrading the accuracy and safety level if they are undetected by the reference stations. Since the characteristics of the ionosphere are different for each region, the ionospheric delay gradient observations in equatorial and low-latitude regions are necessary for developing the suitable ionospheric threat models. The purpose of this work is to analyze the ionospheric delay gradients observed by three GPS stations near Suvarnabhumi airport in Bangkok, Thailand, which is located in the low-latitude region. The ionospheric irregularities in this region are mainly caused by the plasma bubble, which usually occurs after sunset. The GPS data with plasma bubble occurrence during the September equinox 2011 and 2012 are therefore analyzed. In addition, the data analysis procedure utilizing the rate of total electron content change index for this region is proposed. The results show that the ionospheric delay gradients observed in the west-east direction appear higher than the south-north direction, varying from 28 to 178 mm/km during plasma bubble occurrences. 2015. American Geophysical Union. All Rights Reserved. 1. Introduction The Global Navigation Satellite Systems (GNSS) now plays a major role in the aviation navigation. However, the atmospheric propagation delays, especially the ionospheric delay, are the remaining error sources which degrade the accuracy of aircraft positioning. To improve the accuracy of the system, the differential techniques have been developed to mitigate the errors based on the assumption that the ionosphere is uniform in the nearby area. In order to meet the safety requirements of aviation, several augmentation systems have been developed and standardized by the ICAO (International Civil Aviation Organization). For example, the SBAS (satellite-based augmentation system) and GBAS (ground-based augmentation system) [International Civil Aviation Organization, 2010] provide the differential corrections and additional information for integrity assessment to support the aircraft navigation for all operational levels. For the GBAS which provides higher accuracy of differential corrections, the GBAS ground facility is designed to augment as well as alert the aircrafts when the integrity cannot be assured during the approach and landing. An importance benefit of using GBAS instead of the legacy aircraft navigation such as the ILS (Instrument Landing System) is the ability to support multiple runways and multiple approaches by a single GBAS ground facility. However, it is well known that the nonuniform ionospheric structure called the ionospheric delay gradient can cause the errors of differential corrections broadcast to the aircraft. Datta-Barua et al. [2002], first investigated the errors of differential corrections observed during the ionospheric storm on 6 April 2000 over CONUS (Conterminous United States) region, of which the results show that it possibly reached hundreds of mm/km. From this discovery, Luo et al. [2002] simulated the effects of spatial ionospheric gradient on the accuracy of aircraft positions. In the worst case scenario, the ionospheric spatial gradient anomalies could cause the vertical positioning errors of the aircrafts to exceed the position bounds computed in the aircraft. Recently, the impacts and mitigation techniques of ionospheric anomalies to GBAS have been extensively studied and proposed [see, for example, Luo et al., 2003, 2005; Walter et al., 2004; Ene et al., 2005; Konno et al., 2006; Lee et al., 2006; Konno, 2007;Ramakrishnan et al., 2008; RUNGRAENGWAJIAKE ET AL. IONOSPHERIC DELAY GRADIENT IN THAILAND 1076

Pullen et al., 2009; Lee et al., 2011a; Khanafseh et al., 2012]. In particular, the very strong ionospheric storm on 20 November 2003 in the local afternoon caused by the coronal mass ejection from the Sun, which led to the storm-enhanced plasma density over the American sector. The observation and validation checks also confirmed that this extreme ionospheric anomaly can cause the ionospheric delay gradients as high as 413 mm/km [Lee et al., 2011b; Pullen et al., 2009]. Consequently, the ionospheric threat model was developed and proposed based on the GPS data obtained from CORS (Continuously Operating Reference Station) during the extreme ionospheric events in 2000 2004 [Datta-Barua et al., 2010]. This model is already certified on the system design approval by the Federal Aviation Administration (FAA) and currently used in the Honeywell International SLS-4000 for CAT-I precision approach service in CONUS. However, the extensive previous studies and the current ionospheric threat model covered mostly the midlatitudes regions which focus on the ionospheric irregularities over this area. For the equatorial and low-latitude regions, however, it is well known that the plasma bubble is a common phenomenon which can also cause the ionospheric delay gradient. The plasma bubble is an area of low electron density originated from the bottom side of the ionosphere, which generally occurs in the equatorial and low-latitude regions after sunset. The ionospheric delay gradient observations are studied in several regions [see, for example, Yoshihara et al., 2007; Dautermann and Mayer, 2010; Srinivas et al., 2014]. The effects of plasma bubble to GBAS are studied using three-dimensional plasma bubble model [Saito et al., 2009]. The background electron density distribution generated by the NeQuick model combined with the moving rectangular depletion region of electron density is used to simulate the plasma bubble drift. The simulation results show that a single plasma bubble can cause the vertical positioning error of more than 10 m, which is near the safe limit for aircraft automatic landings in category III weather conditions if undetected. In fact, the plasma bubble has a complex shape and can have multiple occurrences at the same time as multiple bubbles. The ionospheric delay gradient observations during plasma bubble occurrences are therefore important for validation and improvement of the ionospheric threat model. In this work, we analyze the ionospheric delay gradient results observed by three GPS monitoring stations near Suvarnabhumi airport, Bangkok, Thailand. Base on data availability, the GPS data during plasma bubble occurrences in 2011 and 2012 are analyzed. In addition, the data analysis procedure incorporating the rate of total electron change index (ROTI) for equatorial and low-latitude ionospheric delay gradient monitoring stations is proposed. The key differences are the ionospheric anomaly detection and the receiver bias calibration concept, which are more simple and suitable than the existing procedure. 2. Ionospheric Delay Gradient Analysis Procedure 2.1. Ionospheric Delay Estimation In order to estimate the ionospheric delay gradient, the ionospheric delays selected from station pairs are first estimated. Since, the slant ionospheric delay (I) is proportional to the amount of electrons in terms of slant total electron content (STEC), which can be expressed as I ¼ 40:3 f 2 STEC (1) where f is the frequency of radio signal that passes through the ionosphere and STEC is the amount of electrons aligning in the line of sight between a satellite and a ground station, generally expressed in TEC unit or TECU (1 TECU = 10 16 electrons/m 2 ). Note that the current protected signal for civil aviation is based on L-1 signal only so the ionospheric delay gradient estimation is computed from the ionospheric delay on L-1 frequency. To estimate the STEC, the linear combinations of pseudorange and carrier phase measurements from dual-frequency GPS receiver are widely used [Garner et al., 2008]. The STECs can be derived from pseudorange (STEC P ) and carrier phase (STEC L ) measurements i.e., STEC P ¼ KP ð 2 P 1 Þ; (2) and STEC L ¼ KL ð 1 L 2 Þ (3) where P 1, P 2 and L 1, L 2 are the pseudorange and carrier phase measurements expressed in range (meters) at the L-1 (1575.42 MHz) and L-2 (1227.60 MHz) frequency, respectively, and the constant K = 9.5196 TECU/m for STEC is expressed in TECU. The different combination is due to the ionospheric divergence effect [Misra and Enge, 2011]. Although the slant TEC derived the carrier phase measurements (STEC L ) is less noisy than that from the RUNGRAENGWAJIAKE ET AL. IONOSPHERIC DELAY GRADIENT IN THAILAND 1077

dstec (TECU) dstec (TECU) 10 5 0-5 STFD-KMIT -10 0 5 10 15 20 10 5 0-5 AERO-KMIT -10 0 5 10 15 20 Time (UTC) Figure 1. Example uncertainties offset of differential STEC between nearby stations on 17 August 2011. (top) STFD-KMIT station pair; (bottom) AERO-KMIT station pair. The different colors indicate the different GPS satellites. pseudorange measurements (STEC P ), it still contains the integer ambiguity which can cause the STEC L to be negative. In order to retain the precise level of STEC L and also remove the integer ambiguity effect, the STEC L value is adjusted to the mean level of STEC P for each continuous arc (assuming that the integer ambiguities are constant). However, the adjusted STEC still has the inherent bias called interfrequency bias (IFB) that needs to be accounted for. The IFB can be caused by both GPS satellite and receiver because the differential extra time delay due to the internal electronic circuits, antenna cable length, receiver front end, and also environment temperature [Rideout and Coster, 2006]. So the adjusted STEC can be expressed as STEC adj ¼ STEC þ B S þ B R (4) where B S and B R are the satellite and receiver IFB (in TECU), respectively. To determine the absolute STEC, these biases need to be first removed. For the satellite IFBs, they are available online and easily downloadable from several sources such as the international GNSS service [ftp://igscb.jpl. nasa.gov/] and the center for orbit determination in Europe (Center for Orbit Determination in Europe) at the University of Berne [ftp://ftp.unibe.ch/aiub/code/]. In fact, the satellite IFBs will be canceled in the ionospheric delay gradient estimation. For the receiver IFB, there are several methods for the receiver IFB calibration [see, for example, Otsuka et al., 2002; Ma and Maruyama, 2003; Komjathy et al., 2005; Rideout and Coster, 2006]. The extensive list of previous studies [Lee et al., 2010, 2011c; Kim et al., 2012; Jung and Lee, 2012] proposed the long-term ionospheric anomaly monitoring for GBAS. The receiver IFB calibration method based on Ma and Maruyama [2003] is recommended in this approach since it is simpler and faster to estimate a single receiver IFB than the supertruth method which is used in Datta-Barua et al. [2010]. Since the leveling uncertainties of adjusted STEC are affected by the multipath and also intraday receiver IFB variation effect [Ciraolo et al., 2007], it can cause the errors of differential STEC calibrated by this method. Figure 1 shows an example of the uncertainty offset of differential STEC between nearby stations (STFD-KMIT: 12 km baseline and AERO-KMIT: 4 km baseline). Different colors represent the GPS satellites. The uncertainty offset of differential STEC can vary from 2 to 5 TECU. The ionospheric delay gradient estimation accuracy can therefore be affected by these uncertainties. Figure 2. Illustration of ionospheric delay gradient monitoring stations. 2.2. Receiver Bias Estimation In this work, we propose to use a simple method whereby the uncalibrated bias of adjusted differential STEC (dstec) is a summation of the natural differential STEC and the differential receiver IFB. The clarification of this concept is shown in Figure 2 and from d STEC k ¼ STEC k 1 STECk 2 (5) þðb R1 B R2 Þ where STEC k 1 STECk 2 is the natural differential STEC measured from stations RUNGRAENGWAJIAKE ET AL. IONOSPHERIC DELAY GRADIENT IN THAILAND 1078

Figure 3. Comparison of STEC and dstec measured from PRN2 on 1 September 2011 during quiet and disturbed ionospheric conditions. 1 and 2 viewing at the same satellite k. Note that the satellite IFB is already canceled so the bias term is only the differential receiver IFB (B R1 B R2 ). For the differential receiver IFB removal, we assume that the slant TECs of a short baseline monitoring stations are similar during the quiet ionospheric conditions. Figure 3 shows an example of STEC as well as dstec patterns measured from PRN2 on 1 September 2011, which contained both the quiet and disturbed ionospheric conditions. The constant level of dstec during the quiet ionospheric condition (10:00 13:00 UTC) is assumed to be the differential receiver IFB (B R1 B R2 ). Although this concept can also be utilized for the plasma bubble occurrence detection by comparing the dstec during the quiet and disturbance ionospheric condition, it is difficult to find the setting of proper threshold due to the uncertainty of the differential receiver IFB. So the dstec computation is only computed for the differential receiver IFB estimation. 2.3. Ionospheric Anomaly Detection For the ionospheric disturbance detection, the long-term ionospheric anomaly monitoring for GBAS collects two indices of global geomagnetic activity from the space weather databases, which are the planetary K (Kp) and disturbance storm time (Dst). These indices are used to evaluate the potential period of ionospheric anomaly. However, they may not be suitable for the plasma bubble occurrence detection. The previous studies [Pi et al., 1997; Beach and Kintner, 1999; Nishioka et al., 2008] have utilized the variation of TEC values to detect the occurrence of plasma bubbles. The variation of TEC values in terms of the rate of TEC change index (ROTI) is used to indicate a plasma bubble occurrence. ROTI is a standard deviation of rate of TEC change (ROT), i.e., and ROTðÞ¼ i STECði þ 1 Þ STEC ðþ i ; (6) t iþ1 t i vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi u1 X N ROTI ¼ t 2 ROTðÞ ROT i : (7) N i¼1 The ROT is computed from the differential STEC at every minute and the 5 min time window is used to compute ROTI in this work. Note that the satellite and receiver IFB are already canceled since these biases are almost constant during a short sampling interval period. So the monitoring stations can compute the ROTI without biases consideration which is reducing the computation load and screening the potential plasma bubble occurrence data for the differential receiver IFB calibration. Figure 4 shows an example of ROTI on 1 September 2011, when the plasma bubbles possibly occurred after sunset. The ROTI suddenly increased from the background levels during the period of plasma bubble occurrence. In order to detect the plasma bubble occurrence, Nishioka et al. [2008] suggested using the difference of ROTI value between daytime (no plasma bubble occurrence) and nighttime (with plasma bubble occurrence). The observed ROTI values during daytime is generally lower than 0.1 TECU/min and exceeds 1 TECU/min during the plasma bubble occurrences on nighttime. To avoid the fault detection due to the measurement noise at low elevation angles, we set the ROTI threshold at 0.5 TECU/min to indicate the plasma bubble occurrence. In fact, RUNGRAENGWAJIAKE ET AL. IONOSPHERIC DELAY GRADIENT IN THAILAND 1079

Figure 4. Rate of TEC change index (ROTI) values computed at KMIT station on 1 September 2011. The different colors indicate the different GPS satellites. this threshold value can be modified by considering the ROTI observations when changing the GPS receivers or observed locations. 2.4. Ionospheric Delay Gradient Estimation After removing the differential receiver IFB, the ionospheric delay gradient can be estimated from a ratio of the differential adjusted STEC measured from the same satellite between two stations and the baseline distance, which can be expressed as [Datta-Barua et al., 2010] I k ¼ Ik 1 Ik 2 d ¼ 40:3 STEC k 1 STECk 2 f 2 d where d is the baseline distance (km) and f is the frequency of L-1 GPS signal which is the currently used for civil aviation. The ionospheric delay gradient is generally expressed in mm/km. 2.5. Proposed Data Analysis Procedure Figure 5 is a summary of the proposed data analysis procedure for equatorial and low-latitude ionospheric delay gradient monitoring stations. The daily raw RINEX (Receiver Independent Exchange Format) format data obtained from each monitoring station are first processed to find and fix discontinuities in the carrier phase observations. This preprocessing is done by the open source software called GPS toolkit or GPSTk [Tolman et al., 2004]. More than 600 s of good data points with less than 300 s of adjacent-point time gap of measurements are first filtered and then performed by the cycle slip detection. The cycle slip detection module of GPSTk is based on the works of Blewitt [1990]. This algorithm utilizes the geometry-free (GF) and wide lane (WL) linear combination to find cycle slips. In this work, the GF and WL variation are set to 16 and 1.5 m, respectively. After cycle slip detection, the preprocessed pseudorange and carrier phase measurements are (8) Figure 5. Proposed data analysis procedures to compute ionospheric delay gradients due to ionospheric disturbance at equatorial and low-latitude stations. RUNGRAENGWAJIAKE ET AL. IONOSPHERIC DELAY GRADIENT IN THAILAND 1080

Figure 6. Three GPS monitoring stations near the Suvarnbhumi international airport, Bangkok, Thailand. used to determine the adjusted STEC. Next, the plasma bubble occurrence is determined from the ROTI levels. The advantages of using ROTI as the ionospheric anomaly indicator are the needless external information (Kp and Dst) gathering and the efficiency of data processing because the biases of STEC data are automatically removed in the ROTI computation. In this work, the ROTI level more than 0.5 TECU/min is flagged to indicate the plasma bubbles occurrence. Then, only the flagged data are used to compute the ionospheric delay gradient. In order to mitigate the leveling uncertainty effects, the differential receiver IFB calibration is considered for each satellite. Note that this approach may lose the ionospheric delay gradient information during the ionospheric quiet condition caused by the physical separation between reference station and the equatorial ionization anomaly, which is the large-scale electron density gradient existing in the equatorial and low-latitude regions. However, these two factors exhibit less significant effects on the gradients than the plasma bubble occurrence. Next, the maximum I value of each station pair is collected and then still need to be confirmed by manual validation to prevent the systematic errors caused by the GPS receivers. Finally, the confirmed maximum I value of each day will be updated to the bound of the ionospheric threat model. 3. Experimental Setup The RINEX (receiver-independent exchange) format data at 1 Hz sampling interval are collected from three dual-frequency GPS receivers at the monitoring stations near the Suvarnabhumi airport in Bangkok, Thailand. One is located on the runway of the airport (AERO: 13.6945 N, 100.7608 E). The others are located at King Mongkut s Institute of Technology Ladkrabang (KMIT: 13.7278 N, 100.7726 E) and Stamford International University (STFD: 13.7356 N, 100.6612 E) are shown in Figure 6. The Novatel ProPak-V3 GPS receiver is used at the AERO and STFD stations, while the KMIT station used the Novatel DL-V3 GPS receiver. Both types of receiver are based on the same OEMV-3 receiver board. The KMIT and STFD baseline (~12 km) can be assumed as the east-west direction and the AERO and KMIT baseline (~4 km) for the north-south direction. In this work, we selected the data with the ROTI level over 0.5 TECU/min during September equinox season of 2011 and 2012 for the analysis. 4. Results and Discussions The 31 days of flagged data during September 2011 and 2012 are analyzed. The dstecs showing both the quiet and disturbed ionospheric condition are used for the differential receiver IFB calibration, which is separately considered for each satellite and each station pair. Both ionospheric conditions are used to distinguish the differential IFB (which is generally constant) from the absolute dstec value. The overall intraday variation of the differential receiver IFB can vary around 1 3 TECU. In Figure 7, we show the histogram of differential receiver IFBs between AERO-KMIT and STFD-KMIT station pair of all flagged data. The differential receiver IFBs can vary from 4.11 to 10.20 TECU for AERO-KMIT and from 1.37 to 8.69 TECU for STFD-KMIT station pair with the average and standard deviation of 7.243 and 1.63 TECU for AERO-KMIT and 5.083 and 1.75 TECU for STFD-KMIT station pair, respectively. The differences in the average differential receiver IFB (7.243 TECU for AERO-KMIT and 5.083 TECU for STFD-KMIT baseline) could be caused by the multipath environment at each location and also the temperature dependent on the other parts of the GPS receiver such as the antenna preamplifier and the cable used to connect the receiver to the antenna. Since the differential receiver IFBs can be considered as the natural differential receiver IFB combined with the uncertainty offsets due to the multipath and intraday receiver IFB variation effects, the assumption that the receiver IFB is a daily constant RUNGRAENGWAJIAKE ET AL. IONOSPHERIC DELAY GRADIENT IN THAILAND 1081

Figure 7. Histogram of differential receiver IFB between (top) AERO-KMIT station and (bottom) STFD-KMIT station during September equinox 2011 and 2012. can cause the error of ionospheric delay gradient estimation. Therefore, the standard deviations of the differential receiver IFBs can reflect the accuracy of the ionospheric delay gradient ( I) estimation if the previous work method is used. The results show the standard deviations of both baseline are similar, but the baseline distances are 3 times different (~4 km for AERO-KMIT and ~12 km for STFD-KMIT baseline), so the accuracy in terms of I are 66.17 mm/km and 23.68 mm/km for AERO-KMIT and STFD- KMIT station pair, respectively. In this work, we separately consider the differential receiver IFBs for each satellite and each baseline so these errors can be mitigated. Next, the summary of maximum I values during the September equinox season in 2011 and 2012 are shown in Tables 1 and 2, respectively. These values are free from the uncertainty offset due to the new differential receiver IFB assumption. The STFD-KMIT baseline can be assumed as the west-east gradient and AERO-KMIT for the south-north gradient. The maximum I can vary from about 28 to 178 mm/km, while the overall maximum I of the STFD-KMIT baseline is higher than the AERO-KMIT baseline. The higher gradient at west-east direction could be explained by the alignment combined with motion of equatorial plasma bubble, whereby the low electron density region elongates in the north-south direction and moves in the eastward direction. However, the highest maximum I was found on the AERO-KMIT baseline on 22 September 2011 reaching 178 mm/km. Figure 8 shows the highest maximum I occurred on the AERO-KMIT baseline measured by PRN9 on that day. The uncalibrated STEC patterns of both stations indicate the multiple low electron density regions or multiple plasma bubbles during the satellite passing. The peak reaches the elevation angle of 38.90 at the first-pass plasma bubble during 13:00 13:30 UT or 20:00 20:30 LT (UT + 7). Note that we found some data gap of STEC on AERO station due to loss of lock tracking during plasma bubble occurrence. The previous studies [Kim et al., 2012; Jung and Lee, 2012] recommended the polynomial fitting method to merge the adjacent data gap of STEC. Since the STEC patterns observed during plasma bubble occurrences are more fluctuated corresponding with the complex shape of low electron density regions, the data merging by the polynomial fitting may cause the aliasing of natural STEC. Table 1. Summary of Maximum Ionospheric Delay Gradient in 2011 a Day Max ( I) (STFD-KMIT) Elevation Max ( I) (AERO-KMIT) Elevation 1/09/2011 74.1423 60.4182 107.9654 37.3313 2/09/2011 48.0014 56.5176 46.9092 56.6751 3/09/2011 47.0926 38.5982 41.5379 39.7136 5/09/2011 83.9019 57.1261 71.9907 48.8708 15/09/2011 98.227 41.8706 75.0675 41.2403 16/09/2011 71.5266 45.4079 78.6378 45.9666 18/09/2011 99.1812 80.7503 74.3446 81.4172 21/09/2011 77.6459 51.477 58.7644 77.6769 22/09/2011 115.0557 41.5181 178.6774 38.9015 23/09/2011 29.6133 47.7109 28.5965 59.0788 24/09/2011 131.9525 42.4043 116.358 44.1382 25/09/2011 82.0485 81.9626 65.5578 72.1774 26/09/2011 85.988 82.3252 59.4376 83.352 27/09/2011 53.4031 76.3584 62.0129 77.2927 30/09/2011 113.6279 89.4067 80.3121 88.3635 a Italic values indicate maximum observed gradient. RUNGRAENGWAJIAKE ET AL. IONOSPHERIC DELAY GRADIENT IN THAILAND 1082

Table 2. Summary of Maximum Ionospheric Delay Gradient in 2012 Day Max ( I) (STFD-KMIT) Elevation Max ( I) (AERO-KMIT) Elevation 1/09/2012 53.4342 34.3633 30.8231 38.793 2/09/2012 67.5578 75.0022 63.5875 78.143 14/09/2012 48.831 47.3317 44.6458 47.2502 15/09/2012 81.0166 47.2157 76.8431 47.6037 17/09/2012 138.9088 88.0301 82.8074 88.0301 24/09/2012 48.654 65.274 37.9204 39.8029 25/09/2012 154.0052 53.2874 84.2283 63.2328 2/10/2012 60.287 32.3555 40.7745 44.6948 5/10/2012 55.2212 55.222 51.5206 47.8623 15/10/2012 112.9938 42.5551 55.9498 44.234 18/10/2012 91.7162 37.688 77.2257 38.4194 19/10/2012 55.2068 58.9205 57.1736 58.9611 23/10/2012 79.8716 57.7942 105.3635 39.3346 24/10/2012 47.0182 53.6035 35.5489 71.1678 30/10/2012 126.6001 44.7716 94.376 45.1767 31/10/2012 101.5447 39.3794 76.8858 39.4056 However, we have not investigated this point in this work. For higher data reliability, the STECs at the elevation angles above 30 are used to estimate ionospheric delay gradients. The results from Tables 1 and 2 are shown as a function of the elevation angles in Figure 9. The highest maximum I values are 154.01 and 178.677 mm/km for STFD-KMIT and AERO-KMIT baseline, respectively. The relationship between the maximum I and elevation angle is not clear. Although the maximum I observed in this study are well bounded by the extremely rare event observed in CONUS, the observed 20 100 mm/km of I during plasma bubble season may potentially degrade the availability of the current CAT-I GBAS operation in this area; if the broadcast fault-free bounding parameters such as σ vig, σ pr_gnd,andp value are inflated to bound these gradients (σ vig is the standard deviation of a normal distribution associated with the residual ionospheric uncertainty due to the ionospheric spatial decorrelation; σ pr_gnd is the 1 sigma differential correction error generated by the ground stations for each satellite; P value is the ephemeris decorrelation parameter for each satellite). In addition, this could make bounding of these more Figure 8. Maximum ionospheric delay gradient of the AERO-KMIT baseline observed by PRN29 on (bottom) 22 September 2011, (top) elevation tracking of PRN29, and (middle) uncalibrated STEC of KMIT and AERO station. RUNGRAENGWAJIAKE ET AL. IONOSPHERIC DELAY GRADIENT IN THAILAND 1083

200 180 STFD-KMIT AERO-KMIT 160 140 I (mm/km) 120 100 80 60 40 20 0 30 40 50 60 70 80 90 Elevation (degree) Figure 9. Maximum ionospheric delay gradients as a function of elevation angles. Acknowledgments Theauthorsaregratefultothe National Institute of Information and Communication Technology (NICT) for numerous technical discussions related to this research. This work is supported by the Thailand Research Fund and King Mongkut s Institute of Technology Ladkrabang through the Royal Golden Jubilee Ph. D. Program (grant PHD/0022/2555). In addition, it is partially supported by King Mongkut s Institute of Technology Ladkrabang (grant 2559A11802187). The proprietary GPS data were collected in the ionospheric and GNSS data center (http://iono-gnss.kmitl.ac.th) of King Mongkut s Institute of Technology Ladkrabang. This work is a part of collaboration of King Mongkut s Instituteof Technology Ladkrabang, Stamford International University, Aeronautical Radio of Thailand and Electronic Navigation Research Institute, Japan. frequent events difficult when applying the position-domain geometry screening by using these existing broadcast parameters (for example, the current maximum σ vig scale is just 25.5 mm/km), when the bounded differential errors resulting from these events presents a challenge to availability. The current SARPs (Standards and Recommended Practices) require that the minimum detectable gradient with a probability of 1 10 9 and within 1.5 s shall be 300 mm/km for a category III capable GBAS facility. Recently, these SARPs have modified this requirement so as to limit the remaining differential errors when the aircraft reaches the runway threshold to a maximum value of 2.75 m with a probability of 1 10 9. So the frequently observed I could reach this maximum error if they affect more than two satellites simultaneously. In this work, although we present only 2 years of data due to availability during high solar cycles with the active plasma bubble months. In order to assure the safety of GBAS operation in this area, the monitoring stations should continue to investigate the data as long term monitoring to develop the up-to-date ionospheric threat model. More data analysis in the other durations of solar cycle and validation of the observation in other equatorial and low-latitude regions are also important for improvement of the system performance under the ionospheric anomaly conditions. 5. Conclusions The ionospheric delay gradient is an important parameter for high-precision GBAS operations. However, the extensive previous studies mostly focused on the ionospheric irregularities existing in the midlatitude regions, but the observations in the equatorial and low-latitude regions are studied in a few areas. In this work, we investigate the ionospheric delay gradient observed in Thailand, which is located in a low-latitude region. In addition, the data analysis procedure suitable for the equatorial and low-latitude monitoring stations is proposed. The results show that the ionospheric delay gradient can vary from about 28 to 178 mm/km, while 20 to 100 mm/km is frequently observed during the plasma bubble occurrences. For the future works, we will include more monitoring parameters such as the plasma bubble speed, the spatial scale transition region (front width), and the number of loss of lock satellite during plasma bubble occurrence for the monitoring system. In addition, we will expand this monitoring system to install in the other important airports in Thailand. References Beach, T. L., and P. M. Kintner (1999), Simultaneous Global Positioning System observation of equatorial scintillations and total electron content fluctuations, J. Geophys. Res., 104(A10), 22,553 22,565, doi:10.1029/1999ja900220. Blewitt, G. (1990), An automatic editing algorithm for GPS data, Geophys. Res. Lett., 17(3), 199 202, doi:10.1029/gl017i003p00199. RUNGRAENGWAJIAKE ET AL. IONOSPHERIC DELAY GRADIENT IN THAILAND 1084

Ciraolo, L., E. Azpilicueta, C. Brunini, A. Meza, and S. M. Radicella (2007), Calibration errors on experimental slant total electron content (TEC) determined with GPS, J. Geodes., 81, 111 120. Datta-Barua, S., T. Walter, S. Pullen, M. Luo, J. Blanch, and P. Enge (2002), Using WAAS ionospheric data to estimate LAAS short baseline gradients, Proceedings of the 2002 National Technical Meeting of the Institute of Navigation, pp. 523 530. Datta-Barua, S., J. Lee, S. Pullen, M. Luo, A. Ene, D. Qiu, G. Zhang, and P. Enge (2010), Ionospheric threat parameterization for local area global-positioning-system-based aircraft landing systems, J. Aircraft, 47(4), 1141 1151. Dautermann, T., and C. Mayer (2010), Equatorial plasma depletions observed over Brazil Impact on safety critical GNSS navigation, Proceedings of the 23rd International Technical Meeting of The Satellite Division of the Institute of Navigation, pp. 2583 2591. Ene, A., D. Qiu, M. Luo, S. Pullen, and P. Enge (2005), A comprehensive ionosphere storm data analysis method to support LAAS threat model development, Proceedings of the 2005 National Technical Meeting of the Institute of Navigation, pp. 110 130. Garner, T. W., T. L. Gaussiran II, B. W. Tolman, R. B. Harris, R. S. Calfas, and H. Gallagher (2008), Total electron content measurements in ionospheric physics, Adv. Space Res., 42(4), 720 726. International Civil Aviation Organization (2010), International Standard and Recommended Practices, Annex 10 to the Convention on International Civil Aviation, Volume I, Radio Navigation Aids. Jung, S., and J. Lee (2012), Long-term ionospheric anomaly monitoring for ground based augmentation systems, Radio Sci., 47, RS4006, doi:10.1029/2012rs005016. Khanafseh, S., S. Pullen, and J. Warburton (2012), Carrier phase ionospheric gradient ground monitor for GBAS with experimental validation, J. Inst. Navigat., 59(1), 51 60. Kim, M., J. Lee, and S. Pullen (2012), Data quality improvements and applications of long-term monitoring of ionospheric anomalies for GBAS, Proceedings of 25th International Technical Meeting of the Satellite Division of the Institute of Navigation, pp. 2159 2174. Komjathy, A., L. Sparks, B. D. Wilson, and A. J. Mannucci (2005), Automated daily processing of more than 1000 ground-based GPS receivers for studying intense ionospheric storms, Radio Sci., 40, RS6006, doi:10.1029/2005rs003279. Konno, H. (2007), Dual-frequency smoothing for CAT III LAAS: performance assessment considering ionosphere anomalies, Proceedings of the 20th International Technical Meeting of the Satellite Division of the Institute of Navigation, pp. 424 437. Konno, H., S. Pullen, J. Rife, and P. Enge (2006), Ionosphere monitoring methodology for hybrid dual-frequency LAAS, Proceedings of the 19th International Technical Meeting of the Satellite Division of the Institute of Navigation, pp. 409 424. Lee, J., M. Luo, S. Pullen, Y. S. Park, P. Enge, and M. Brenner (2006), Position-domain geometry screening to maximize LAAS availability in the presence of ionosphere anomalies, Proceedings of the 19 th International Technical Meeting of the Satellite Division of The Institute of Navigation, pp. 393 408. Lee, J., S. Jung, E. Bang, S. Pullen, and P. Enge (2010), Long term monitoring of ionospheric anomalies to support the local area augmentation system, Proceedings of 23 rd international technical meeting of the satellite division of the Institute of Navigation, pp. 2651 2660. Lee, J., J. Seo, Y. S. Park, S. Pullen, and P. Enge (2011a), Ionospheric threat mitigation by geometry screening in ground-based augmentation systems, J. Aircraft, 48(4), 1422 1433. Lee, J., S. Datta-Barua, G. Zhang, S. Pullen, and P. Enge (2011b), Observations of low-elevation ionospheric anomalies for ground-based augmentation of GNSS, Radio Sci., 46, RS6005, doi:10.1029/2011rs004776. Lee, J., S. Jung, and S. Pullen (2011c), Enhancements of long term ionospheric anomaly monitoring for the ground-based augmentation system, Proceedings of the 2011 international technical meeting of the Institute of Navigation, pp. 930 941. Luo, M., S. Pullen, D. Akos, G. Xie, S. Datta-Barua, T. Walter, and P. Enge (2002), Assessment of ionospheric impact on LAAS using WAAS supertruth data, Proceedings of the 58th Annual Meeting of the Institute of Navigation and CIGTF 21st Guidance Test Symposium, pp.175 186. Luo, M., S. Pullen, J. Dennis, H. Konno, G. Xie, T. Walter, and P. Enge (2003), LAAS ionosphere spatial gradient threat model and impact of LGF and airborne monitoring, Proceedings of the 16th International Technical Meeting of the Satellite Division of the Institute of Navigation, pp. 2255 2274. Luo, M., S. Pullen, S. Datta-Barua, G. Zhang, T. Walter, and P. Enge (2005), LAAS study of slow-moving ionosphere anomalies and their potential impacts, Proceedings of the 18th International Technical Meeting of the Satellite Division of the Institute of Navigation, pp. 2337 2349. Ma, G., and T. Maruyama (2003), Derivation of TEC and estimation of instrumental biases from GEONET in Japan, Ann. Geophys., 21, 2083 2093. Misra, P., and P. Enge (2011), Global Positioning System: Signals, Measurements, and Performance, revised 2nd ed., Ganga-Jamuna Press, Lincoln. Nishioka, M., A. Saito, and T. Tsugawa (2008), Occurrence characteristics of plasma bubble derived from global ground-based GPS receiver networks, J. Geophys. Res., 113, A05301, doi:10.1029/2007ja012605. Otsuka, Y., T. Ogawa, A. Saito, T. Tsugawa, S. Fukao, and S. Miyazaki (2002), A new technique for mapping of total electron content using GPS network in Japan, Earth Planets Space, 54, 63 70. Pi, X., A. J. Mannucci, U. J. Lindqwister, and C. M. Ho (1997), Monitoring of global ionospheric irregularities using the worldwide GPS network, Geophys. Res. Lett., 24(18), 2283 2286, doi:10.1029/97gl02273. Pullen, S., Y. S. Park, and P. Enge (2009), Impact and mitigation of ionospheric anomalies on ground-based augmentation of GNSS, Radio Sci., 44, RS0A21, doi:10.1029/2008rs004084. Ramakrishnan, S., J. Lee, S. Pullen and P. Enge (2008), Targeted ephemeris decorrelation parameter inflation for improved LAAS availability during severe ionosphere anomalies, Proceedings of the 2008 National Technical Meeting of The Institute of Navigation, pp. 354 366. Rideout, W., and A. Coster (2006), Automated GPS processing for global total electron content data, GPS Solut., 10, 219 228. Saito, S., T. Yoshihara, and N. Fujii (2009), Study of effects of the plasma bubble on GBAS by a three-dimensional ionospheric delay model, Proceedings of the 22nd International Technical Meeting of The Satellite Division of the Institute of Navigation, pp. 1141 1148. Srinivas, V. S., A. D. Sarma, A. S. Reddy, and D. K. Reddy (2014), Investigation of the effect of ionospheric gradients on GPS signals in the context of LAAS, Prog. Electromagn. Res. B, 57, 191 205. Tolman, B. W., R. B. Harris, T. Gaussiran, D. Munton, J. Little, R. Mach, S. Nelsen, B. Renfro, and D. Schlossberg (2004), The GPS Toolkit Open Source GPS Software, Proceedings of the 17th International Technical Meeting of the Satellite Division of the Institute of Navigation, pp. 2044 2053. Walter, T., S. Datta-Barua, J. Blanch, and P. Enge (2004), The effects of large ionospheric gradients on single frequency airborne smoothing filters for WAAS and LAAS, Proceedings of the 2004 National Technical Meeting of the Institute of Navigation, pp. 103 109. Yoshihara, T., N. Fujii, K. Matsunaga, K. Hoshinoo, T. Sakai, and S. Wakabayashi (2007), Preliminary analysis of ionospheric delay variation effect on GBAS due to plasma bubble at the southern region in Japan, Proceedings of the 2007 National Technical Meeting of The Institute of Navigation, San Diego, pp. 1065 1072. RUNGRAENGWAJIAKE ET AL. IONOSPHERIC DELAY GRADIENT IN THAILAND 1085