Lenghu on the Tibetan Plateau as an astronomical observing site – Nature

Geographic information

The Lenghu site is at a local summit on Saishiteng Mountain, which is located to the east of the Altyn Mountains and on the northern edge of the Qaidam Basin. Its geographical coordinates are 38.6068° N, 93.8961° E, and it has an elevation of 4,200 m. The Lenghu site occupies a unique geographic position in the Eastern Hemisphere and bridges the huge gap between Mauna Kea (155.8246° W), Atacama (70.4042° W) and the Canary Islands (17.8577° W). This will form a perfect network of ground-based, high-quality observatories ready for great scientific discoveries, including searching signs of life on exoplanets, electromagnetic counterparts of gravitational wave outbursts, high-value transient events alerted by space-borne triggers that need to be done in very narrow time window^32,33 and much more.

According to the climate record collected at three local weather stations for past 30 years, the average annual precipitation is around 18 mm, with over 3,500 h of annual sunshine^34,35. Land transport from the site to the local supporting base, Lenghu Town, and then to the developed areas of China by road and railway networks is convenient. The nearest international airport, highway access and the cargo railway stations of Dunhuang are all within only 250 km of the town. The altitude of Lenghu Town is only 2,700 m and is 80 km away from the site, which provides comfortable conditions for a supporting base for the site. This infrastructure enables good logistics for future activity at the site (Extended Data Fig. 1).

AOT statistics

The clear night is derived by using both LH-Cam images analysis and the smoothness of integrated sky brightness records of the SQM. Lenghu has almost no artificial light pollution. Therefore, on LH-Cam images, any cloud will block the star background on a new moon night or when the moon is 18° or more under the horizon, leaving a dark patch on the image. When the moon is in the view of the camera, clouds will be directly visible. On sky brightness curves of the SQM, clouds make darkening or brightening fluctuations patterns depending on the moon age, and the amplitudes of such variations are correlated well with cloud coverage/thickness on the images¹⁴, as demonstrated in Extended Data Fig. 2. In the left panel, three LH-cam images show a typical clear time (right), small clumpy cloud passage between clear time (left) and overcast (middle) cases, on the night of 6 October 2019, together with the SQM light curve. The right panel shows the distribution of clear time (cyan) in 2019 as an example of annual observing time statistics, based on the method described above.

A comparison of Tibetan sites

In addition to Lenghu, other sites, namely Ngari, Muztagh Ata and Daocheng, on the Tibetan Plateau were also tested during an earlier general site survey by the Large Optical/infrared Telescope team. An intensive site testing programme was carried out for the Chinese 12-m telescope from 2016 to 2018 at the three sites. The testing results are concluded in an overview paper¹⁸. The Ngari site was later found to be good for the primordial gravitational wave project⁴. The concerns regarding strong wind, cloud cover in summer and light pollution from the nearby Shiquanhe Town³⁶ are potential challenges to further development of optical/infrared astronomy at the Ngari site. The other two sites are now being developed for different purposes. It turns out that Lenghu has the best observing conditions of all the sites tested on the plateau. A direct comparison of the key parameters of AOT and seeing are shown in Extended Data Table 2 and Extended Data Fig. 4, respectively. As the seeing data offered by the Large Optical/infrared Telescope team for Ngari, Muztagh Ata and Daocheng were all truncated at 3.0 arcseconds, the seeing data of Lenghu are processed accordingly, as presented in Extended Data Fig. 4.

In Extended Data Table 2, we adopted the method of the Large Optical/infrared Telescope team for AOT calculation, based on all-sky camera images. They divide the total visible sky by two circles with zenith angles of 44.7° and 65°, namely the inner and the outer circles. When there is no cloud in the inner and the outer circles, it is defined as ‘clear’ (or photometric in ref. ¹⁸); when only the inner circle is clear, it is defined as ‘outer’ (spectroscopic).

Site quality matrix scores

Extended Data Table 3 shows the site quality matrix for the Lenghu site based on all the nights that have both DMn and cAOT statistics. The DMn and cAOT are divided into five levels and four levels, respectively. Each element is assigned a weight according to the values of the DMn and cAOT (in parenthesis). The total score of the site is denoted by the ratio between the weighted summation of the number of nights and the total number of nights (457 in our case). For Lenghu, this score is 65%. An ideal site with all nights in yellow would score 100%. For the Delingha site, the median of all seeing measurements is 1.58 arcseconds¹³. Half of the AOT for Delingha (about 250 in total¹³) would be in blue (score 0.5) with the other half in brown (score 0.3), and the total score would be approximately 40%, which is typical for current existing classical observatories in China. For the Xinglong site (150 km from Beijing), where the Large Sky Area Multi-Object Fibre Spectroscopic Telescope (LAMOST) is hosted, the expected score would be even lower than the Delingha site for the same level of seeing, but with less observable time.

Using the publicly available seeing and AOT data from the European Southern Observatory (ESO) website, the sites at Cerro Paranal and La Silla are evaluated on the site quality matrix scale. Both the seeing and AOT data released on the ESO website are monthly average values. For the AOT of the ESO sites, a photometric night calls for six or more hours of consecutive photometric night. Therefore, only the third column of Table 1 (cAOT > 6 h) is used for the site quality estimations for the Cerro Paranal and La Silla sites. The monthly fractions of photometric nights are transferred to cAOT nights, and then the nights are divided into five levels according to the seeing divisions in Table 1. Finally, the total scores for Cerro Paranal (1999–2012) and La Silla (2000–2008) are both 66%. It is noted that these scores are upper limits for the sites as the nights with cAOT < 6 h (therefore, lower scores) cannot be assessed owing to the lack of daily weather data.

Turbulence profiles

To understand the local meteorological pattern at the mountain region where our site is located, three balloon experiments were conducted at the Lenghu weather station. Once at 23:15 UT on 12 August 2020, and twice on 16 November 2020, at 11:18 UT and 23:44 UT. These balloon missions provided a vertical spatial resolution of 6.4 m. The mean potential temperature profile θ(h) is calculated by

where h is the altitude, T(h) is the temperature profile in K and P(h) is the pressure profile in hPa. The structure function of the temperature fluctuation ({C}_{T}^{2}) is evaluated by the AXP model³⁷. The refractive index structure constant ({C}_{N}^{2}) is then estimated by the Gladstone formula

The turbulence profiles calculated using the parameters obtained during the balloon flights are shown in Extended Data Fig. 5. Above 11 km, ({C}_{N}^{2}) decreases monotonously with no seasonal pattern. ({C}_{N}^{2}) is around 10^−17.5 and 10⁻¹⁷ between 4 km and 11 km. On 16 November, the two turbulence profiles show a similar trend, but the turbulence strength at night (red profile) is lower than that in the morning (grey profile). At an altitude of 6–9 km, the turbulence profile shows a clear difference in August and November, which suggests possible seasonal changes.

PWV

The PWV can be calculated by the equation

where ρ is the density of liquid water, g is the acceleration of gravity, p_z is the pressure of the ground and q is the specific humidity. The value of q is calculated by the water vapour pressure e by the equation

 The saturation water vapour pressure is usually converted from temperature by the Goff–Gratch formula³⁸. We used the temperature, pressure and humidity of the ground weather station to estimate the amount of PWV. Here we adopted a temperature drop rate of 6.5 K km⁻¹, an exponential decay of air pressure with temperature, and the height of the tropopause is 11 km as measured by the balloon experiments (Extended data Fig. 5). The mean and median values of PWV modelled for the whole testing period are 3.13 mm and 2.01 mm, respectively (Extended Data Fig. 6).

By checking the data, we found that PWV changes substantially with season. We calculated the mean PWV by month and compared it with the PWV of La Palma and Mauna Kea (Fig. 3). Our two-year PWV values show a similar trend, that is, PWV values in winter are much lower than those in summer. The standard deviation in each month is about half of the average monthly PWV. From October to March, the mean PWV value is 1.55, which is 27% and 73% of the PWV values in La Palma and Mauna Kea, respectively²⁸.

To explore the possible deviations of our PWV, we also adopted the empirical equation between PWV and specific pressure of water vapour, PWV = a₀e + a₁. The coefficients a₀ and a₁ change with elevation and latitude. Here we adopted the coefficients of Tibetan Plateau²⁵ (assuming an elevation of 4,200 m) and Ngari (also called Ali) site²⁶ (southwest part of the Tibetan Plateau) to re-estimate the PWV. Comparing the PWV estimates using the two sets of coefficients, our modelling of the PWV of the Lenghu site is consistent, but slightly overestimated by 0.15 mm and 0.01 mm, respectively.

Dust grains

Dust and aerosol above an observing site can create problematic extinction for astronomical observations, and their presence in the ground layer can be troublesome for both optical surfaces and mechanical bearings. To measure local dust and aerosol, we implemented a dust meter (GRIMM EM180) in December 2019. We have so far collected a full year of uninterrupted data regarding the dust grains and aerosols of the site, with a temporal resolution of 5 min. The mean and median values of particulate matter with a diameter smaller than 10 μm (PM₁₀) density are 20.7 μg m⁻³ and 11.7 μg m⁻³, respectively. The ambient dust level of the Lenghu site is comparable to the sites in Atacama³⁹. Twice in 2020, a high value of PM₁₀ was recorded, of around 100 μg m⁻³, during sandstorms that originated from the Taklimakan and local Gobi deserts. Dust grain densities higher than 50 μg m⁻³ occurred 31 times in 2020, with a typical duration of several hours. Owing to the high altitude of the site, dust is less serious than at the La Palma site⁴⁰, which suffers from proximity to the Sahara Desert, and precautions can be implemented at the Lenghu site to protect the equipment for the few days per year affected by dust.