Abstract
Continuous observations play an important role in the studies of solar variability. While such observations can be achieved from space with an almost 100% duty cycle, it is difficult to accomplish a very high duty cycle from the ground. In this context, we assess the duty cycle that has been achieved from the ground by analyzing the observations of a six station network of identical instruments, the Global Oscillation Network Group (GONG). We provide a detailed analysis of the duty cycle using GONG observations spanning over 18 yr. We also discuss the duty cycle of individual sites and point out various factors that may impact individual site or network duty cycles. The mean duty cycle of the network is 93%, however it reduces by about 5% after all images pass through the stringent quality-control checks. The standard deviations in monthly and yearly duty cycle values are found to be 1.9% and 2.2%, respectively. These results provide a baseline that can be used in the planning of future ground-based networks.
Export citation and abstract BibTeX RIS
Original content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence. Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
1. Introduction
Solar observations are now available for a few centuries, however most of the records prior to the 20th century are sparse. The coverage has significantly increased in the last few decades, primarily due to the advancement in observing capabilities from both space and ground. While observations from space are almost continuous with small disruptions, the ground-based observations are adversely affected by the diurnal cycle. There are several research areas in solar physics where uninterrupted observations for a long time are essential in order to achieve reliable results. Helioseismology is among one of these areas where solar oscillations below the surface are used to study phenomena occurring in the interior. These oscillations with a prominent period of 5 minutes were discovered in the early 1960s (Leighton 1960; Leighton et al. 1962) and the physical mechanism explaining these observations was suggested about a decade later (Ulrich 1970; Leibacher & Stein 1971). It was soon realized that continuous observations at high cadence for a longer time would be needed to reduce side lobes in the temporal power spectra, but this was not possible from a single ground station due to the diurnal cycle. There were two possibilities for continuous observations from the ground: either from the South Pole but for a limited time during the Austral Summer (Grec et al. 1980) or build a network of identical instruments at geographically separated locations around the Earth. This led to the inception of several networks for helioseismic observations; some are still operational even after more than 25 yr and others ceased operations after a few years. Observations from these networks allowed the measurement of solar oscillations from low to high degrees where the degree is the number of waves around the solar circumference.
Some of the short-lived networks were the International Research on the Interior of the Sun (Fossat 1991, IRIS) and the Taiwanese Oscillations Network (Chou et al. 1995, TON). Two still operating networks are the Birmingham Solar Oscillation Network (Chaplin et al. 1996, BISON) since 1975 and the Global Oscillation Network Group (Leibacher & GONG Project Team 1995, GONG) since 1995. While BiSON provides observations in solar-disk-integrated light best suitable for low spherical harmonic degree helioseismic modes, which travel deep into the core, GONG observations resolve acoustic modes covering a wider range from core to the near-surface layers. In addition, several observational campaigns were held at the South Pole from the early 1980s to the mid-1990s (Fossat 2013; Harvey 2013), however continuous observations were obtained for a maximum period of about a week. Occasional observation campaigns at the South Pole are still continuing with specific science goals (Finsterle et al. 2004; Jefferies et al. 2019) but continuous observations were acquired only for a few days. All of these observations played a crucial role in advancing the research in helioseismology and confirmed the solar variability below the surface as observed above the surface, first reported by Woodard & Noyes (1985). Furthermore, observations from the current or past space missions also complemented the ground-based observations (Pallé et al. 2015).
In this article, we assess the temporal coverage from the ground-based observatories using observations of the GONG network. Section 2 presents a brief history of GONG. Data used to compute the duty cycle is described in Section 3. Assessment of the duty cycle of individual sites and the network as a whole are discussed in Sections 4 and 5, respectively. We also provide a statistical analysis of the simultaneous observations between different sites in Section 6. Section 7 demonstrates the scientific need for a high duty cycle, especially in helioseismic studies. Finally, the summary is presented in Section 8.
2. A Brief History of GONG: From Helioseismology to Space Weather
One of the first tasks in establishing a network of observatories is to identify the number and locations of the sites suitable for substantially reducing the observed diurnal cycle. For this, a systematic study of potential sites for the GONG network started with estimating the performance of several hypothetical ground-based network composed of two to six existing observing sites distributed in both longitude and latitude. Results suggested that a network of six sites was needed to achieve >90% annual mean duty cycle (Hill & Newkirk 1985). This initiated a coordinated site survey with identical instruments (Fischer et al. 1986) placed at 15 sites around the globe, divided into six longitudinal bands to measure the cloud-cover percentage and transparency of the sky. Based on the detailed analysis of the collected data (Hill et al. 1994b) and the overall performance of each site (Hill et al. 1994a), six sites were finally selected in early 1991. Each selected site represents one of the six longitudinal bands that allows the network to make 24 hr-a-day observations of the Sun year round. These sites are shown in Figure 1 and their locations are given in Table 1.
Table 1. Locations of GONG Sites and Operation Start Dates
Site | Latitude | Longitude | Elevation | |
---|---|---|---|---|
Name | Identifier | (m) | ||
Learmonth, Australia | LE | S 022° 066 | E 114° 098 | 15 |
Udaipur, India | UD | N 024° 538 | E 073° 109 | 677 |
El Teide, Spain | TD | N 028° 030 | W 016° 430 | 2425 |
Cerro Tololo, Chile | CT | S 030° 042 | W 070° 197 | 2190 |
Big Bear, USA | BB | N 034° 372 | W 116° 171 | 2063 |
Mauna Loa, USA | ML | N 019° 101 | W 155° 333 | 3471 |
Download table as: ASCIITypeset image
The GONG instrument is based on measuring the line-of-sight Doppler shifts of the photospheric spectral line Ni i 676.8 nm. The key element of this instrument is a Michelson interferometer, the details of which are given by Harvey & GONG Instrument Team (1995). Deployment of the GONG instruments started in 1995 January with the TD site and the full network was completed in October of the same year after UD, the last site, became operational. Start dates of the operations at each site are also given in Table 1. Identical instruments and identical settings at all six sites operating in automated mode with minimal human intervention is the key to the network's success. The science-grade data of the network are available since 1995 May 7. At the beginning, the observations were carried out with low-resolution cameras that had rectangular pixels (aspect ratio of 1.28:1) and was oriented in such a way that the longer dimension was aligned with the solar axis of rotation. The resulting images were 204 × 239 pixels across with a pixel resolution of ≈10'' × 8''.
After observing successfully for the initial three years using low-resolution cameras (Harvey et al. 1996), the capabilities of the GONG network were expanded in 2001 by upgrading its cameras to 1024 × 1024 with square pixels (Harvey et al. 1998). Until 2001, the GONG, commonly known as GONG classic, data usage was primarily limited to global helioseismic studies. The upgrade provided data that is suitable for local helioseismic studies, e.g., internal conditions below small regions on the surface, hemispheric and latitudinal distributions of subsurface flows, mapping of the farside for the detection of active regions, etc. The upgraded network is commonly referred to as GONG+. This also served as the starting point to expand GONG's capabilities to space weather. In subsequent years, the GONG observations were extended to one-minute cadence full-disk magnetograms in 2006 (Hill et al. 2008; Petrie et al. 2008) and 2048 × 2048 pixels images of the Hα spectral line at a wavelength of 656.28 nm in mid-2010 (Harvey et al. 2011). Note that all Hα observations are taken at a cadence of one minute at each site but with the acquisition time offset between adjacent sites so that a new Hα image is, in principle, available every 20 s from the network. Various GONG data products are available at https://nso.edu/data/nisp-data/, and their applications in space-weather related studies and predictions are described by Hill (2018).
3. Data
We compute the duty cycle of individual sites and the network from fully calibrated images that have successfully passed through stringent quality checks as described in the Automated Image Rejection (Clark et al. 2004, AIR), and consist of, for example, active tracking, image stability, signal/noise level in individual images (Toussaint et al. 1995; Pintar & Toussaint 1998), temporal variations between images, and the correction for orientation for the best position-angle estimation (Toner 2001). As mentioned in Section 2, GONG has many different observables out of which we choose observations of Doppler velocities to determine the duty cycle. While the magnetograms and the Dopplergrams are part of the same observing sequence, the data products go through different processing pipelines and thus have different image rejection criteria as part of the quality control. Hα observations have been taken since 2010 using a different instrument setup and may provide a slightly different duty cycle.
The duty cycle is defined as the fraction of 24 hr day for which the observations are available. In addition, the network duty cycle is determined by combining all six site images for each minute in the same 24 hr period. There are several factors that significantly influence the observations at individual sites and, finally, the network duty cycle. These factors include, but are not limited to, severe weather conditions, instrument problems, downtime due to preventive maintenance and upgrades, etc. Although the network has been operational for more than 25 yr, we consider only a period of 18 yr from 2002 January to 2019 December for assessing the duty cycle, i.e., the period after the completion of camera upgrade. This period allows us to analyze consistent data set without any forced major disruptions. A total number of 6574 days, i.e., 9.47 × 106 minutes, are covered in this study. Since there are overlapping observations between different sites and GONG provides observations at a cadence of one minute, the analysis has been carried out using more than 13 million images collected at GONG sites over the period of 18 yr. Table 2 summarizes the details of the network input data and Table 3 provides the details of each site's data. In the next section, we describe the performance of individual sites starting from the East. Results for the full network are discussed in Section 5.
Table 2. Details of the Network Data Used in this Study
Data Type | Count | Duty Cycle | Details |
---|---|---|---|
Period covered | 18 yr | ||
(2002–2019) | |||
Total Number of days | 6574 | ||
Total Minutes covered | 9.47 × 106 | ||
Total Observations from all sites | 1.35 × 107 | ||
Total Minutes with observations | 8.36 × 106 | 88.33% | |
Total Minutes with no observations | 1.11 × 106 | 11.67% | |
Total minutes with independent observations | 4.12 × 106 | 43.49% | Table 3 |
Total minutes with simultaneous observations | 4.24 × 106 | 44.84% | Table 4 |
Download table as: ASCIITypeset image
Table 3. Details of Individual Site Data Used in this Study
Sites | Contribution to Total | Contribution to Network Duty Cycle | ||
---|---|---|---|---|
Observations | Total | Independent | Overlapping | |
LE | 18.6% | 26.6% | 12.9% | 13.7% |
UD | 11.2% | 15.9% | 5.1% | 10.8% |
TD | 17.8% | 25.7% | 8.7% | 17.0% |
CT | 21.1% | 30.1% | 8.0% | 22.1% |
BB | 16.5% | 23.6% | 4.7% | 18.9% |
ML | 14.8% | 21.2% | 4.1% | 17.1% |
Download table as: ASCIITypeset image
4. Individual Site Duty Cycle
4.1. Learmonth, Australia (LE)
The Learmonth Solar Observatory (LSO) situated on the western shore of the Exmouth Gulf on the North West Cape of Australia houses one of GONG's instruments. The observatory is co-administered by the Bureau of Meteorology, Australian Government, and the US Air Force. This GONG site became operational in 1995 April. We display the variation of the daily duty cycle in Figure 2, as an example, for the year 2009. Here different panels represent different calendar months. In addition to the seasonal variation in the duty cycle, there are some random gaps, which signify no observations for entire days. Longer gaps (e.g., from the last week of April to mid-May in Figure 2) generally arise due to instrument downtime or preventive maintenance periods while shorter gaps correspond to severe weather that may last from one to a few days. The maximum daily observations were obtained for 11.47 hr which gives a duty cycle of 47.8%. In total, LE provided observations for 5621 days out of 6574 days (i.e., 85.5% days). Various milestones achieved by individual sites are given in Table 4.
Download figure:
Standard image High-resolution imageTable 4. Various Performance Figures Achieved by Individual Sites
Criterion | Frequency | LE | UD | TD | CT | BB | ML |
---|---|---|---|---|---|---|---|
Total coverage [days] | All | 5621 | 3949 | 5280 | 5516 | 5532 | 5536 |
Mean Duty Cycle [%] | All | 26.6 | 16.0 | 25.4 | 30.2 | 23.6 | 21.2 |
Median Duty Cycle [%] | Monthly | 28.2 | 18.4 | 26.5 | 31.1 | 25.5 | 21.4 |
Yearly | 26.5 | 16.2 | 25.7 | 31.1 | 23.8 | 21.1 | |
Maximum Duty Cycle [%] | Daily | 47.8 | 46.7 | 49.4 | 50.3 | 48.5 | 47.7 |
Monthly | 34.2 | 27.0 | 39.4 | 43.6 | 36.3 | 26.1 | |
Yearly | 31.1 | 21.4 | 30.2 | 33.1 | 25.6 | 29.5 | |
Minimum Duty Cycle [%] | Daily | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 |
Monthly | 17.8 | 4.0 | 16.1 | 18.0 | 12.6 | 18.1 | |
Yearly | 23.1 | 9.9 | 20.5 | 26.4 | 19.8 | 14.1 | |
Standard Deviation [%] | Monthly | 6.0 | 9.7 | 8.3 | 9.7 | 7.3 | 2.6 |
Yearly | 2.0 | 3.0 | 2.6 | 2.2 | 1.4 | 3.3 |
Download table as: ASCIITypeset image
To examine the trends in LE observations, we display in Figure 3 the monthly variation in duty cycle for four selected years. A systematic seasonal variation is seen in all four years with a dip in the duty cycle in April–May, except for 2017, followed by an increasing trend. The higher duty cycle is achieved in later months every year with the beginning of Summer in the southern hemisphere due to the availability of sunshine over a larger fraction of the day. The overall trends in monthly and yearly variations averaged over the entire period are presented in Figures 4 and 5, respectively. We have included other GONG sites in the same figure for comparison. As mentioned above, the trend in the mean monthly values display strong seasonal variation with a standard deviation of 6.0%. When we consider yearly variation (see Figure 5), LE's contribution to the GONG network has been stable throughout the entire period and, as a result, the standard deviation reduces to about 2.0%.
Download figure:
Standard image High-resolution imageDownload figure:
Standard image High-resolution imageDownload figure:
Standard image High-resolution imageFrom Table 3, we find that the observations taken at LE contribute about 18.6% to the total observations and 26.6% to the network duty cycle. As mentioned earlier, all sites provide independent as well as overlapping observations and LE's contribution to the network duty cycle is almost equally split between these two categories. It is further noted that LE's independent contribution is the largest among all sites.
4.2. Udaipur, India (UD)
The Udaipur Solar Observatory (USO) is situated in a semi-arid highland region of Western India and houses the GONG's instrument in the Asian band. USO is administered by the Physical Research Laboratory (Ahmadabad), a unit of the Department of Space, Government of India. The Udaipur site experiences a strong weather pattern with an extended period of monsoon lasting for two–three months every year and the observations are stopped during this period. As a result, UD's contribution to the total observations is the lowest at 11.2%. As presented in Table 3, UD provides only 5.1% independent contribution to the network duty cycle while the total contribution is 15.9%. Two-thirds of UD observations have overlap with other sites and the details are given in Section 5.
To evaluate the daily variation throughout the year, we show the daily duty cycle for the year 2009 in Figure 6 and the monthly duty cycle for four selected years in Figure 7. Compared to LE, Udaipur observations have many longer gaps in 2009. The UD site was down from early July to mid-September due to the monsoon and the other two gaps, in late January and early October, are due to system failures and occurrence of preventive maintenance. The downtime due to the monsoon is seen in all monthly average plots presented in Figure 7. The duty cycle again decreases in December–January due to hazy conditions and less sunshine hours available during the Winter months. The maximum observations are normally obtained from February to May. During Spring/Summer time, the Udaipur site was observed for more than 11 hr and the maximum daily duty cycle during 2002–2019 reached 46.7%.
Download figure:
Standard image High-resolution imageDownload figure:
Standard image High-resolution imageThe average monthly and yearly duty-cycle trends, as shown in Figures 4(b) and 5(b), clearly demonstrate the effect of the monsoon on the duty cycle. The seasonal variation in the mean duty cycle for UD is more significant than for LE. As a result, the standard deviations, as summarized in Table 4, are also higher for both monthly and yearly averages. In addition, the mean duty cycle is notably lower. Total observations from UD are available for 60.1% or 3949 days during the period of 18 yr, the lowest among all sites. The lower coverage from UD was a combination of the monsoon season and several other long periods of system downtime, for example, a fire broke in the GONG instrument shelter in 2010 August and the site remained non-operational until the beginning of 2011.
4.3. El Teide, Canary Islands (TD)
The El Teide site is located on the island of Tenerife, in the Canary Islands off the west coast of Africa, hosted by the Instituto Astrofísica de Canarias (IAC). The IAC is administered by a public consortium which includes the Universidad de la Laguna. Similar to other sites, observations from TD are also affected by strong seasonal variations. Sometimes it is hit by an intense dust storm, known as the Calima, blowing off the Sahara Desert. During the Calima, strong winds are accompanied by an orange haze that significantly reduces the sky transparency and the observations, if any, are discarded. Even after being affected by the several months of poor weather conditions in Winter and Summer, TD provides good coverage of about 80.3% of the days.
The daily duty cycles for TD are presented in Figure 8 for 2009. It is evident that daily observations from TD are consistent for several months in Summer, thus providing good coverage to the GONG network. This is also apparent from Figure 9 where the duty cycle from April to September for all four years is consistently higher than other months. From Table 3, one can notice that TD's contribution to the network duty cycle is about 25% which is similar to LE. The independent contribution from the Teide site is 8.7% in the overall duty cycle. This site along with LE becomes notably more important when the UD instrument is down for the long monsoon season or any other down-time. During a clear day in Summer, good observations for almost 12 hr can be obtained from this site.
Download figure:
Standard image High-resolution imageDownload figure:
Standard image High-resolution imageSimilar to other sites, a strong weather-dependent variation in the TD duty cycle is apparent in Figure 4(c) where we show the monthly duty cycle averaged over the entire period considered in this study. It is noted that the duty cycle drops to almost one half in the Winter months and the standard deviation in the monthly mean duty cycle is moderately large with a value of 8.3%. However, the yearly variation is relatively small (Figure 5(c)) with a standard deviation of only 2.6% (Table 4).
4.4. Cerro Tololo, Chile (CT)
The second GONG site in the southern hemisphere is located on the mountain Cerro Tololo in northern Chile. The instrument is located on the grounds of the Cerro Tololo Interamerican Observatory (CTIO), a division of NSFs National Optical-Infrared Astronomy Research Laboratory (NSFs NOIRLab) headquartered in Tucson, USA. Due to excellent sky conditions, CTIO is one of the main sites for astronomical investigations of the southern skies. This is reflected in the number of observations from the individual sites summarized in Table 3. CT's contribution to GONG's cumulative observations is the highest and has 5516 observing days, that is about 84% of the total days. Despite a significantly large number of observations and about 30% contribution to the network duty cycle, the independent contribution from CT is only 8.0% while overlapping observations cover the remaining 22.1%, as evident from Table 3. This large fraction of overlapping observations are due to the proximity of adjacent sites BB and TD. Moreover, CT has the maximum overlap with the neighboring sites' observations.
Daily variation in duty cycles for the year 2009 is presented in Figure 10, which exhibits stability for several months. Maximum duty cycle of about 50% is achievable for several months, i.e., January, February, November, and December. From the monthly variation for four sample years (Figure 11), it is clear that the number of observations from CT reaches its lowest value in the month of June. This strong seasonal variations is also noticeable in Figure 4(d) and the standard deviation in the monthly duty cycle is about 9.7%. Aside from the seasonal variation, the yearly duty cycles from CT remain reasonably high as shown in Figure 5(d) and the values remain between 28% and 33% with a standard deviation of 2.2%, as given in Table 4.
Download figure:
Standard image High-resolution imageDownload figure:
Standard image High-resolution image4.5. Big Bear, USA (BB)
The only GONG site in the United States mainland is located in the southwest on the premises of Big Bear Solar Observatory (BBSO), managed by the New Jersey Institute of Technology (NJIT). The instrument is placed on the north shore of an artificial lake in the San Bernardino Mountains in California. Similar to other sites, BB also has strong seasonal variation in the duty cycle that peaks around June and reaches its lowest value in December–January. From the daily duty-cycle histograms, as shown in Figure 12, it is evident that the observations are sparse in the Winter months. Although the observations from BB are consistent in the Summer months, there are a few years when the duty cycle is relatively low and 2009 is among those years. As illustrated in Figure 13, the monthly duty-cycles for four years clearly show the variation from one year to another. The overall trend is more pronounced in Figure 4(e) where the average monthly duty cycle is shown. A gradual increase in the duty cycle is observed after February with a maximum in June, a slow decrease during July to September and finally, a sharp decline is noticed until December.
Download figure:
Standard image High-resolution imageDownload figure:
Standard image High-resolution imageWe note that the Big Bear site has been very stable throughout the operations of GONG and the year-to-year variation is the least among all six GONG sites (See Figure 5(e)). Therefore, the standard deviation in yearly values is also a minimum, i.e., only 1.4%, while the standard deviation in monthly values is 7.3%. Table 4 shows that the observations collected at BB spanned over 5532 days that is about 84.1% of the total observing days. The BB contribution to all GONG observations is around 16.5% with three-quarters of them have overlap with other sites as described in Table 3. Nevertheless, the independent coverage of 4.7% in total duty cycle makes this site equally important with other GONG sites.
4.6. Mauna Loa, USA (ML)
The sixth GONG site is located at the north side of the Mauna Loa volcano at the Mauna Loa Observatory (MLO), on the island of Hawai'i (Big Island). MLO is an atmospheric baseline station of the Global Monitoring Laboratory (GML) of the National Oceanic and Atmospheric Administration (NOAA). Local staff support is provided by the Mauna Loa Solar Observatory operated by the High Altitude Observatory, Boulder, Colorado. Unlike other GONG sites, Mauna Loa is minimally affected by strong seasonal variations. This is clearly visible in the daily duty-cycle plot for a year displayed in Figure 14. Although we do see variations from month to month, these are not as strong as in other sites and the daily values remain in a similar range throughout the year. Further, to evaluate the existence of this pattern in all years, we plot monthly values for four individual years in Figure 15. This shows no distinct trend in the monthly variation within a year, however there are random fluctuations, but these do not indicate any strong seasonal pattern. On averaging monthly values over the entire period presented in Figure 4(f), a mild pattern emerges with minor dips around both equinoxes. Finally, we find that the standard deviation in monthly values for ML is only about 2.6% which is the lowest among all sites (Table 4). However, the standard deviation in yearly values displayed in Figure 5(f) is higher than the monthly value and also the highest among all sites indicating that ML experienced the highest year-to-year variation in the duty cycle.
Download figure:
Standard image High-resolution imageDownload figure:
Standard image High-resolution imageIn addition, the number of observing days at ML at 84.2% level is comparable with CT and BB, but the contribution to the total network observations is the second lowest (see Table 3). It is worth pointing out that the independent observations from ML cover merely 4.1% which is the lowest among all six sites; more than three-quarters of ML observations overlap with other sites.
5. Network Duty Cycle
Section 4 clearly demonstrates that all sites except ML are affected by strong seasonal patterns. While sites in the northern hemisphere provide good coverage during the Summer months (humps in June–July for BB and TD in Figure 4), this trend reverses in the southern hemisphere (dips in June–July for LE and CT). Further, due to the low latitude location, the seasonal patterns are less apparent in ML observations. However, the network observations after integrating over those from individual sites (Toner et al. 2003) display a consistent coverage throughout the year with a significantly high duty cycle. Figure 16 presents the network daily duty cycle for a year exhibiting minimal weather-related patterns and the daily value reaches >90% in most cases. We do notice some days with duty cycles dipping down to 50% or lower but these are rare. To better understand the variation, if any, we compute mean monthly duty cycles for the network for four years and present them in Figure 17. We do not see any distinct pattern in all of these years, which strongly supports the concept of a ground-based network for uninterrupted observations with consistent duty cycles.
Download figure:
Standard image High-resolution imageDownload figure:
Standard image High-resolution imageFinally, we present monthly and yearly duty cycles averaged over the entire period in Figure 18. We notice a mild trend in the monthly duty cycle (left panel of Figure 18) throughout the year reaching its lowest value in July (84%) and the highest in December/January (91%). We also notice a mild variation in the yearly values. The yearly mean is relatively low in the first two years and then remains around the mean over the entire period for the rest of the years. The low standard deviations of 1.9% and 2.2% in monthly and yearly values, respectively exhibit excellent stability in the network duty cycle, even after more than 25 yr of its operations. In addition, Figure 5 also indicates that there are no long-term changes in the duty cycles of any of the six network sites though the contribution from each site is affected by the local weather as well as the instrument down time. The lack of long-term variations in duty cycles further implies that there are no significant changes in climate in any of the GONG network sites. These findings clearly demonstrate that a ground-based network can reliably study long term solar variability.
Download figure:
Standard image High-resolution imageThe mean duty cycle over 18 yr is 88.3% and the median value is 92.0%. Figure 19 displays the duty cycle corresponding to each day for the entire analysis period; smaller bars representing days where the duty cycle is lower than the median values. It is also observed that most days with lower duty cycles are around mid-year which is also apparent in monthly mean values displayed in Figure 18. Distribution of the daily values for bin-size of 2% is given in Figure 20. We find that only 82 days out of 6574 days covered in this study have the duty cycle less than 50%. Moreover, almost 13% of the total days, i.e., 865 days, have a duty cycle of 99% or greater of which 106 days have no gaps, i.e., 100% duty cycle. In general, the frequency of occurrence gradually decreases with decreasing duty cycle except when values lie between 97% and 98% where we notice a slight increase.
Download figure:
Standard image High-resolution imageDownload figure:
Standard image High-resolution imageIt is important to mention that the above analysis is based on the fully calibrated images where each and every individual site image is passed through the pre-defined quality control checks (Clark et al. 2004). During this process, some images are rejected every day affecting the site as well as network duty cycles. To estimate the impact of the rejection criteria on the network duty cycle, we display the distribution of the duty cycle computed from the partially calibrated site images in Figure 21. Note that these partially calibrated images do not pass through all quality checks and may be counted for initial observations. A comparison between the duty cycles for partially and fully calibrated data reveals that the mean duty cycle obtained from the partially calibrated images is higher with the mean value of 93.5% and the median value of 97%. This is in excellent agreement with the GONG site-survey study where (Hill et al. 1994a) suggested an achievable duty cycle value of 93.3%. This also indicates that the quality control matrices reduce the duty cycle by about 5% on an average, however there are a large number of days when the duty cycle is not affected in this process, primarily due to multiple-site observations for a given minute. Compared to fully calibrated data where only 13% of the days have duty cycle 99% or more, the partially calibrated data have about 40% of the days with duty cycle in this range. This reduction in duty cycle is also seen in another six-site network, BiSON, where the duty cycle of 82% reduces to 78% when fully calibrated images were used (Hale et al. 2016).
Download figure:
Standard image High-resolution image5.1. Comparison with Previous Results
Based on the hypothetical networks, Hill & Newkirk (1985) reported that two-, three-, four-, and six-site networks might achieve 60%, 75%, 82%, and 93% annual mean duty cycle. To check the feasibility of these duty cycle values, we also computed duty cycles for all possible speculative networks using GONG's fully calibrated site images and the results are displayed in Figure 22. The maximum duty cycles for two-, three-, four-, five-, and six-site combinations are 57%, 70%, 79%, 84%, and 88%, respectively. These values are smaller by a few percent than those reported in Hill & Newkirk (1985). These are shown by the open symbols. It should be noted that the duty-cycle values obtained in this study are also affected by the downtime of each site due to instrument repairs and preventive maintenance trips which were not considered in the simulated analysis. However, as discussed earlier, the use of fully calibrated images reduces the duty cycle by about 5% and thus the values obtained in this study are comparable with those reported by Hill et al. (1994a).
Download figure:
Standard image High-resolution imageTo study the impact of the number of sites on the duty cycle, we finally fit duty cycle values obtained for different site combinations to a function. The best fit is found for an exponential function of the form of f(n) = −1.20 e−0.22n +1.20, where f is the duty cycle, and n is the number of sites (Figure 22). It clearly displays that the duty cycle increases exponentially with the increasing number of sites, however the locations of the observing sites play a critical role in achieving desired duty cycle. We also find that adding another site to the six-site network will improve the duty cycle notably.
6. Simultaneous Observations from Multiple Sites
In a well-organized network, it is important that simultaneous observations are carried out at adjacent sites. This will not only minimize interruptions in observations but also helps in a better data calibration. In ideal conditions, the GONG network should be providing overlapping observations all of the time. However as presented in Table 2, almost 11% of the period considered in this study was not covered by any site and the rest of the coverage was equally divided among single-site and overlapping observations. Table 5 illustrates the details of simultaneous observations from two, three, and four sites. As expected, the maximum simultaneous coverage comes from two-site overlap. This is in agreement with the site-survey study by Hill et al. (1994a) where they also achieved a maximum overlap of 44.8% between two sites. Further, the single-site observations were limited to 28.9%, which is lower than this study, and three- and four-site simultaneous observations contributed to 18.5% and 1.7% of the total observing time.
Table 5. Multi-site Simultaneous Contribution to the Network Duty Cycle
Simultaneous | Count | Duty Cycle | Details |
---|---|---|---|
Observations Type | (minutes) | ||
Total | 4.24 × 106 | 44.84% | |
2 sites | 3.37 × 106 | 35.60% | Table 6 |
3 sites | 8.20 × 105 | 8.66% | Table 7 |
4 sites | 5.45 × 104 | 0.58% | Table 8 |
Download table as: ASCIITypeset image
6.1. Simultaneous Observations from Two Sites
Table 5 shows that the two-site simultaneous observations provide coverage of about 35.6% over the entire period. We present in Table 6 the details of the two-site observations and their coverage. In addition, every site has some fraction of overlapping observations with adjacent sites. It can be easily seen that two-site observations provide coverage between 34% and 56%, however the simultaneous coverage ranges between <0.1% and 7.3%. It is obvious that the simultaneous coverage depends on the locations of the sites in the network. The maximum overlap is observed between two pairs of sites, i.e., LE and UD, and TD and CT, with a contribution of 7.3% to the total duty cycle. However, LE and UD have more overlapping observations, i.e., 21.6% of their total coverage compared to TD and CT with an overlapping observation of 16.4%.
Table 6. Simultaneous Coverage from Two Sites in the Network
Site 1 | Site 2 | Total Coverage | Simultaneous Coverage | |||
---|---|---|---|---|---|---|
(minutes) | Duty Cycle | (minutes) | Duty Cycle | |||
LE | UD | 3.2 × 106 | 34.4% | 6.9 × 105 | 7.3% | |
TD | 4.8 × 106 | 50.5% | 9.7 × 104 | 1.0% | ||
CT | 5.4 × 106 | 56.8% | 211 | <0.1% | ||
BB | 4.6 × 106 | 49.0% | 6.9 × 104 | 0.7% | ||
ML | 4.1 × 106 | 43.8% | 3.1 × 105 | 3.3% | ||
UD | TD | 3.6 × 106 | 38.4% | 2.2 × 105 | 2.3% | |
CT | 4.3 × 106 | 45.9% | 1.0×104 | 0.1% | ||
BB | 3.7 × 106 | 39.5% | 34 | <0.1% | ||
ML | 3.5 × 106 | 36.7% | 1.7 × 104 | 0.1% | ||
TD | CT | 4.2 × 106 | 44.8% | 6.9 × 105 | 7.3% | |
BB | 4.2 × 106 | 44.0% | 1.4 × 105 | 1.5% | ||
ML | 4.3 × 106 | 45.2% | 1.6 × 104 | 0.1% | ||
CT | BB | 4.0 × 106 | 42.4% | 4.0 × 105 | 4.2% | |
ML | 4.1 × 106 | 43.5% | 2.8 × 105 | 3.0% | ||
BB | ML | 3.3 × 106 | 34.9% | 4.7 × 105 | 4.4% |
Download table as: ASCIITypeset image
6.2. Simultaneous Observations from Three Sites
Compared to two-site observations, three-site simultaneous coverage is less frequent and the contribution is about 8.66% in the total duty cycle. Table 7 presents the coverage between three sites where we have included only those sites that observed simultaneously at any point in the entire period. We find only two three-site combinations that observed simultaneously for >1% in the total coverage. Among these, the maximum overlap is obtained between CT, BB, and ML.
Table 7. Simultaneous Coverage from Three Sites in the Network
Site 1 | Site 2 | Site 3 | Total Coverage | Simultaneous Coverage | |||
---|---|---|---|---|---|---|---|
(minutes) | Duty Cycle | (minutes) | Duty Cycle | ||||
LE | UD | TD | 5.3 × 106 | 55.9% | 5.4 × 104 | 0.6% | |
LE | UD | ML | 4.9 × 106 | 51.4% | 2.7 × 104 | 0.3% | |
LE | CT | BB | 6.4 × 106 | 67.8% | 2.0 × 102 | <0.1% | |
LE | CT | ML | 6.3 × 106 | 66.0% | 3.9 × 102 | <0.1% | |
LE | BB | ML | 5.4 × 106 | 56.7% | 4.4 × 104 | 0.5% | |
UD | TD | CT | 5.5 × 106 | 57.7% | 1.2 × 104 | 0.1% | |
TD | CT | BB | 5.2 × 106 | 55.1% | 2.4 × 105 | 2.5% | |
TD | CT | ML | 5.5 × 106 | 57.6% | 2.5 × 104 | 0.3% | |
TD | BB | ML | 5.2 × 106 | 54.9% | 3.9 × 104 | 0.4% | |
CT | BB | ML | 4.8 × 106 | 50.4% | 3.8 × 105 | 4.0% |
Download table as: ASCIITypeset image
6.3. Simultaneous Observations from Four Sites
For a tiny fraction in network duty cycle, i.e., <1% , four GONG sites observed simultaneously. As presented in Table 8, the maximum contribution came from TD, CT, BB, and ML. These sites covered 63% of the available time together in 6574 days of which only 0.57%, i.e., 54,423 minutes observed simultaneously. There are two other site combinations in this category, however their contribution to the network duty cycle is less than 0.01%.
Table 8. Simultaneous Coverage from Four Sites in the Network
Sites | Total Coverage | Simultaneous Coverage | |||
---|---|---|---|---|---|
(minutes) | Duty Cycle | (minutes) | Duty Cycle | ||
LE, CT, BB, ML | 6.8 × 106 | 72.3 % | 328 | <0.01% | |
LE, UD, BB, ML | 6.1 × 106 | 64.3 % | 14 | <0.01% | |
TD, CT, BB, ML | 6.0 × 106 | 63.0 % | 54423 | 0.57% |
Download table as: ASCIITypeset image
7. Significance of High Duty Cycle
GONG's uninterrupted full-disk observations have been widely used in numerous science projects as well as space-weather operations/monitoring where duty cycle plays a pivotal role. In particular, helioseismic studies have been greatly benefited from these long span of observations. For example, the continuous data with consistently high duty cycle have allowed the study of temporal variability in global (Jain et al. 2009, 2011; Broomhall 2017) as well as local oscillation mode characteristics (Tripathy et al. 2013, 2015) and their connection with the 11 yr cyclic behavior of the surface magnetic activity. In addition, these data have also revealed other periodicities in helioseismic data, e.g., the quasi-biennial oscillations in oscillation frequencies (Simoniello et al. 2013). Further, long consistent helioseismic data are also crucial in constraining the fundamental stellar properties, particularly in asteroseismology where most stellar observations cover relatively short time intervals (Howe et al. 2020).
Another benefit of Doppler observations with high duty cycle is the reliable inferences of subsurface flows which are crucial in understanding the plasma motion below the surface and the solar dynamo. Several studies have been carried out where torsional oscillation patterns (Howe et al. 2000, 2018), subsurface zonal and meridional flows (Komm et al. 2018; Basu & Antia 2019; Gizon et al. 2020), and deep meridional flows (Jackiewicz et al. 2015) have been reliably inferred. To emphasize the importance of high-duty cycle in helioseismic studies, we compare horizontal flows (Figure 23) computed utilizing the technique of ring diagrams (Hill 1988) where 1664 full-disk Doppler images taken at 1 minute cadence are used. In the figure, arrows show the flow magnitude and the direction, and the background depict the uncertainties in these estimates. The left panel shows the horizontal flows with a 100% duty cycle while the middle and right panels display the flows measured with reduced duty cycles of 79% and 66%, respectively. Lower duty cycles are computed by truncating observations from 100% duty cycle data. From a comparison of these three panels, we notice that the flow magnitudes as well as the errors increase with the decrease of the duty cycles. Thus, the reliable flow measurements require a higher duty cycle.
Download figure:
Standard image High-resolution imageGONG has also become an important data source for uninterrupted high-cadence magnetic-field and Hα observations, primarily due to significantly high duty cycle. These are being used in various space-weather forecast systems (Petrie et al. 2008; Arge et al. 2010) and the studies related to flare and filament eruptions (Luna et al. 2018). In addition, maps of the invisible-side (farside) of the Sun, computed from GONG Dopplergrams, identifying active regions before they rotate toward Earth are another important data product (González Hernández et al. 2007, 2010). An example of consequences of low-duty cycle in farside mapping is illustrated in Figure 24. Here we display farside maps for three different duty cycles using 24 hr GONG observations for 2014 June 21. Top panel shows the map from 100% duty cycle as observed and other two panels display maps for lower duty cycles by artificially introducing gaps in the observations. It is clearly visible that reducing duty cycle reduces signal to noise and also increases random noise. It is also important to note that the uninterrupted synoptic observations are crucial in advancing our knowledge of the phenomena related to solar activity whose origin is still not fully understood (Elsworth et al. 2015).
Download figure:
Standard image High-resolution image8. Summary
Continuous observations of the Sun play an important role in the studies of long-term as well as short-term solar variability. In this context, GONG comprising of six ground-based observatories has been providing consistent and continuous observations spanning over more than two solar cycles with a stable temporal coverage. In this paper, we perform a detailed analysis of all the six sites and assess the duty cycle that has been achieved by individual as well as combined sites.
The study reveals that the maximum observations have been recorded at two sites located in the southern hemisphere and the lowest number of observations were obtained at the UD site, predominantly owing to the yearly site closure during the Indian monsoon. Further, there is significant overlap between observations from different sites, which is important for better calibration and for creating a seamless, long time series by merging observations from various sites. The mean and median duty cycle of the network is 88% and 92%, respectively,
Our study further demonstrates that strong seasonal trends are present at all sites except the Mauna Loa (ML) site, probably due to its geographical location. However, these trends disappear in the network when the observations from all sites are combined/merged. Although the down time of each site varies from year to year, we do not find long-term change in climate at any of the network sites that may affect the duty cycle.
It should be noted that the ground-based networks are less expensive as compared to space missions and have the ease of repairing/upgrading any instrument though the image quality might be affected by the atmospheric seeing. The detailed analysis presented here demonstrates that a significantly high duty cycle values can be achieved from a well-designed network of ground-based observatories and can provide an important platform for long-term synoptic observations.
The authors would like to thank Jack Harvey, John Leibacher, and Charles Lindsey for critically reading the manuscript and their suggestions. K. Jain thanks Valentín Martínez Pillet for many useful discussions. GONG has received support from hundreds of people at the GONG headquarters at the National Solar Observatory and around the world, and their dedication is pivotal for maintaining the GONG operations for such a long time. The authors would like to thank all those who are, or have been, associated with GONG from its planning to the present. This work utilizes data from the National Solar Observatory Integrated Synoptic Program, which is operated by the Association of Universities for Research in Astronomy, under a cooperative agreement with the National Science Foundation and with additional financial support from the National Oceanic and Atmospheric Administration, the National Aeronautics and Space Administration, and the United States Air Force. The GONG network of instruments is hosted by the Big Bear Solar Observatory, High Altitude Observatory, Learmonth Solar Observatory, Udaipur Solar Observatory, Instituto de Astrofísica de Canarias, and the Cerro Tololo Interamerican Observatory.