Taishan Station (72.01° S, 92.08° E; 2621 m a.s.l.) is located on the Antarctic ice sheet (Fig. 1). It is situated on the high Antarctic Plateau, an area characterized by extreme environmental conditions. The station experiences strong katabatic winds and is positioned in a region with limited access to logistics and power resources, making it a challenging site for long-term atmospheric monitoring. The climate at Taishan Station is typical of the high Antarctic interior, where temperatures are consistently low throughout the year. The long-term data from September 2012 to July 2021 revealed an annual average air temperature of -35.4 °C, with a relative humidity of 67 %, an air pressure of 699 hPa, and an average wind speed of 10.9 ms-1 (Ding et al., 2022). The temperature exhibits significant seasonal variation, with a clear “coreless” winter, meaning that there is no distinct temperature minimum, which is common in the interior of Antarctica due to the unique seasonal sunlight patterns (Fig. 2). Furthermore, temperature variability is much greater during the austral winter (2.46 °C) than during the austral summer (1.67 °C), reflecting the greater frequency of weather events and atmospheric disturbances during the colder months (Ding et al., 2020b, 2022).

The CRUX-1.0 fully automated unattended observation system was deployed at Taishan Station from 10 February to 9 March 2024. This system has achieved simultaneous monitoring of CO2 and O3 and obtained continuous observation data. Specifically, the CRUX-1.0 system is installed in a yellow observation cabin, which is located approximately 100 m to the southeast of the green-colored unattended generator cabin (Fig. 1). Given that easterly winds prevail at Taishan Station, the air inlet of the observation cabin is positioned on the upwind side of the generator cabin, which effectively minimizes the potential interference of the monitoring results from the exhaust emissions of the generator cabin. Detailed technical parameters and operational mechanisms of this system will be elaborated in Sect. 3.

This study adopts four representative Global Atmosphere Watch (GAW) background stations covering polar, mid-latitude coastal and alpine regions for atmospheric comparative analysis. The Amundsen-Scott South Pole Station (SPO; 90° S, 24.8° W; 2837 m a.s.l.) is a NOAA-operated Antarctic background station, performing hourly in situ CO2 and O3 observations with biweekly flask sampling and weekly NOAA-2016 scale calibration (Mahesh et al., 2003). It maintains a 98.8 % data continuity rate and provides benchmark Antarctic atmospheric background data (Prinn et al., 2018). Barrow Station (BRW; 71.32° N, 156.61° W; 11 m a.s.l.), a key NOAA GMCC Arctic coastal station, conducts hourly CO2 and O3 monitoring and biweekly flask cross-validation. Influenced by North American emissions and Arctic atmospheric circulation, BRW captures both pristine Arctic background air masses and regional pollution episodes, with all data standardized to 10 min intervals via NOAA quality control protocols (Heintzenberg, 1989). The long-term NOAA-operated Mauna Loa Station (MLO; 19.54° N, 155.58° W; 3397 m a.s.l.), established in 1958, hosts the world’s longest continuous greenhouse gas observation records. It implements hourly WMO GAW-calibrated CO2 and O3 monitoring and daily flask sampling, and its remote marine location minimizes anthropogenic interference to ensure reliable mid-latitude baseline observations (Keeling, 1976). As a high-altitude alpine background station under the ACTRIS network, Jungfraujoch Station (JFJ; 46.54° N, 7.96° E; 3580 m a.s.l.) monitors CO2 and O3 at 6 min intervals, with weekly flask sampling and monthly systematic calibration, effectively reflecting mid-latitude European atmospheric trace gas variations. The station maintains a long-term data completeness of approximately 92 % under routine quality assurance (Sturm et al., 2012). Notably, systematic biases between NOAA and AGAGE datasets are less than 1 % for CO2 and 3 % for O3, which are sufficiently small to be negligible for regional and global comparative analyses (Prinn et al., 2018).

The temperature control module is designed to ensure the stable operation of the monitoring equipment under extreme Antarctic conditions, especially the freezing temperatures at Taishan Station. This module includes a temperature control cabinet, a gas flow controller, and an uninterruptible power supply (UPS) (see Fig. 3-IIb-➃, ➄). The temperature control cabinet is equipped with precise sensors and adjustment mechanisms that maintain a stable internal temperature range for the equipment, protecting it from extreme cold at the station.

To meet the appropriate temperature range for the operation of CO2 and ozone monitoring while reducing the energy consumption of temperature control, the temperature control accuracy is ±1.0 °C, and the heating response time is 5 min, and the minimum heating temperature of the temperature control cabinet is preset to 10 °C. When the internal temperature drops below 10 °C, the system begins to heat; when the temperature exceeds 15 °C, heating stops, ensuring that the equipment remains within a safe and optimal operating temperature range. This design ensures that the monitoring instruments are effectively protected in the extremely low-temperature environment of Taishan Station (which can reach as low as -40 °C). The environmental monitoring and control system, which includes temperature regulation, is the highest power-consuming component in the entire system, with a peak power consumption of 250 W. Given the power constraints at Taishan Station, energy efficiency is a crucial consideration in the design of the temperature control system. This is balanced with the need to maintain a stable environment for the equipment.

Since the experimental system is powered by Taishan Station's external energy cabin, an uninterruptible power supply (UPS) system is installed to prevent sudden power outages or voltage instability from negatively impacting the system's operation (Fig. 3-IId). The UPS provides stable and reliable power to the entire monitoring system (see Fig. 3-IId). In the event of an external power failure, the UPS can quickly switch to backup power. The battery backup can last up to 48 h, ensuring the continuous operation of the monitoring system and uninterrupted data collection. This design significantly enhances the system's reliability and fault tolerance, providing a safeguard against power disruptions in the remote environment.

The satellite communication system consists of a customized bracket (with a height of 3 m) and an integrated transmitting antenna (Fig. 3-IIa-➀), enabling high-speed and stable data transmission to a remote data center. This module uses ARGOS (full system: Argos Data Collection System, named after the all-seeing giant Argus from Greek mythology) satellite communication technology, with a data transmission interval of 1 h, ensuring the real-time transmission of monitored data processed by the data logger in the harsh Antarctic environment, which can avoid potential interruptions and delays that could arise with traditional ground communication methods. The stable applicability of this satellite communication module in Antarctica has been verified through testing at the Prydz Bay-Amery Ice Shelf-Dome A (PANDA) profile automatic weather station (Ding et al., 2022).

The data processing and storage system uses the CR1000X measurement and control data logger (see Fig. 3-IIc), a low-power device primarily used for sensor measurements, direct/remote communication connections, data analysis, external device control, and storage of data and programs. It features a sealed design to shield against radio frequency interference and a stainless steel housing. Equipped with a precise clock, it supports the BASIC-like programming language commonly used by Campbell Scientific data loggers, which includes data processing and analysis capabilities.

The CR1000X connection panel has two 12 V terminals, 16 analog measurement terminals, and detachable connectors. It is responsible for receiving raw data from two monitoring sensors, performing preprocessing, analysis, and storage. This module has powerful data processing capabilities, enabling it to filter out and correct anomalous data points in real time, ensuring the reliability of the data quality. Additionally, the module supports remote data access and download functions, allowing researchers to retrieve and analyze monitoring data at any time.

From 10 February to 12 March 2024, the observation system at Taishan Station operated autonomously under unattended low-temperature conditions, with good performance, and satellite remote data transmission was normal. By analyzing the transmitted data, we found the following: in the environment at Taishan Station, where the average temperature was below -25 °C in February, the internal temperature of the equipment cabinet was able to be stably maintained at 13.9 ± 2.5 °C (Fig. 4a), meeting the environmental temperature requirements for both types of optical cavity equipment used in this experiment.

The hourly average CO2 mixing ratio observed at Taishan Station was 422.6 ± 2.8 ppm (Fig. 4b). To ensure the accuracy of observational data, we first analyzed concurrent meteorological parameters including wind speed and wind direction, and verified that generator exhaust exerted no direct interference on the monitoring data (Fig. S1 in the Supplement). To constrain instrumental drift under unattended field conditions, the system executes an automatic 10 min standard gas calibration every 11 h; this customized calibration interval comprehensively balances drift correction efficiency, limited on-site power supply and automatic operation requirements. All CO2 raw datasets underwent standardized multi-step QA/QC procedures sequentially: exhaust interference exclusion based on sampling layout and field wind rose statistics, wind-speed-dependent pollution screening, quantitative mixing ratio calibration, and final 48 h moving average smoothing for noise reduction. The daily-averaged calibrated CO2 mixing ratio was determined as 420.7 ± 0.7 ppm. The ozone raw datasets were calibrated using the five-point cross-mixing ratio formula (y=0.994⋅x-0.495, R2=0.9998). This calibration function was established in laboratory before instrument shipment via systematic calibration on the Model 205 ozone analyzer with a Thermo 49ips standard ozone calibrator. Upon completion of the QA/QC procedures, the daily-averaged calibrated surface ozone mixing ratio at Taishan Station was 20.1 ± 0.8 ppb. Full technical details of each QA/QC step are provided in the Supplement.

We evaluated system performance using WMO/GAW required indicators: CO2 (zero drift ≤0.1 ppm per month, span drift ≤0.2 % FS, accuracy ±1.5 ppm); O3 (zero drift ≤0.5 ppb per month, span drift ≤1 % FS, accuracy ±0.8 ppb), plus standard deviation and coefficient of variation (CV) (Global Atmosphere Watch [GAW], 2001). The standard deviation provides insight into the dispersion of measurement data, reflecting the precision and stability of the system. On the other hand, CV, which normalizes the standard deviation by the mean value, allows for a comparison of variability across different parameters and conditions, regardless of their magnitude. Using these two metrics, we evaluated the precision of O3 and CO2 mixing ratios, as well as the cabinet temperature, over time.

Prior to the Antarctic field campaign, we had been completed laboratory performance intercomparison between the LICOR-830 analyzer used in this work and Picarro G2301 by means of repeatability, long-term drift and three-point cross-mixing ratio calibration with traceable CO2 standard gases, with related test plots (Figs. S2–S5) attached in the Supplement. The test results indicated that LICOR-830 had a repeatability of 0.4 ppm versus 0.04 ppm for Picarro G2301, alongside drift biases of -4.22 % and 0.82 % respectively; after three-mixing ratio standard gas calibration for 1 h, LICOR-830’s precision was markedly optimized to 0.026 ppm (bias: -0.11 %), while Picarro G2301’s precision changed slightly to 0.0341 ppm with an accuracy of 0.03 %.

In addition, for the ozone measurement module equipped with a 2B Model 205 dual-beam UV analyzer, multi-point calibration conducted before and after field deployment meets the data quality requirement for 1-year Antarctic observation (Bauguitte et al., 2011; Ding et al., 2020a). The laboratory comparison test (Fig. S6) between Model 205 and WMO reference Thermo Model 49i proved excellent measurement consistency (R2=0.9998), with most absolute deviations below 4 ppb, and regular mixing ratio-dependent bias can be fully corrected by linear fitting using pre- and post-campaign calibration results. Laboratory aging experiments demonstrate the intrinsic monthly zero drift of Model 205 is less than 0.6 ppb; after adding extra thermal drift from periodic cabinet temperature fluctuation, the total monthly drift remains below 0.7 ppb, within the WMO/GAW allowed total uncertainty of ±0.8 ppb for O3.

The O3 mixing ratio (black line) consistently has a low standard deviation and minimal fluctuations, indicating high measurement stability. In contrast, the original CO2 mixing ratio (yellow line) displays substantial variation, with a peak standard deviation of 1.8, suggesting lower precision (Fig. 5a). The cabinet temperature (blue line) maintains a stable standard deviation, although a stepwise decrease over time suggests some control, albeit with potential for further improvement. In terms of the CV, the O3 mixing ratio remains low (<6 %) and continues to decrease over time, reflecting enhanced measurement consistency. As illustrated in Fig. 4, the ambient surface temperature at Taishan Station fluctuates widely between -45.0 and -20.5 °C with prominent diurnal cycles, whereas the active thermal cabinet limits internal instrument temperature within 8.0–19.0 °C with a CV of 16 %–20 % (±2.5 °C fluctuation). Comparative time-series analysis of CO2 and O3 in Fig. 4b confirms that such cabinet temperature variation only yields tiny measurement drift (≤0.2 ppmh-1 for CO2, ≤0.1 ppbh-1 for O3), with no significant impact on measurement results (Fig. 5b). Postcalibration, the CO2 mixing ratio (red line) significantly decreases in CV, approaching 0 %, demonstrating the effectiveness of calibration in improving measurement accuracy.

Traditional manned stations (e.g., South Pole Station) have irreplaceable advantages: real-time troubleshooting, flexible parameter adjustment, and weekly on-site calibration (measurement uncertainty <0.5 ppm, better than CRUX's ±1.0 ppm for CO2). However, manned observations face high logistics costs and data gaps due to extreme weather evacuations. CRUX compensates for these shortcomings: its unattended operation mode eliminates the need for resident staff and frequent resupply, substantially reducing routine logistical costs; its 48 h UPS backup also ensures data continuity during short-term power outages. The good consistency with South Pole Station data (CO2: 420.7 ppm vs. 418.2 ppm; O3: 20.1 ppb vs. 20.5 ppb) indicates reliable measurement quality under the current short-term stable deployment, providing supporting evidence for its potential to complement manned observations in the future.

As shown in Table 1, CRUX's strengths lie in integrated optimization for Antarctic inland conditions: (1) Balanced performance: it reduces power consumption by 92 % compared to the initial 3 kW version, achieving <350 W operation while maintaining multi-parameter detection, adapting to polar green energy constraints. (2) Extreme cold reliability: active temperature control isolates -25 to -40 °C external temperatures, and upwind intake with particle filters mitigates blowing snow interference. (3) High autonomy and accuracy: twice-daily auto-calibration reduces CO2 CV to near 0 %. Field tests revealed minor limitation flaws: (1) Temperature control lag: A 5 min heating response delay leads to 16 %–20 % cabinet temperature CV, potentially causing 0.2 ppmh-1 drift of the LICOR-830 analyzer. (2) Insufficient long-term power: 48 h UPS backup. (3) Limited adaptability: fixed-time calibration cannot respond to sudden sensor drift; only two parameters are monitored.

To verify the global representativeness of measurements from CRUX-1.0 at Taishan Station, quantify the regional differences in atmospheric CO2 and O3 between polar and mid-latitude background regions,

To present the latitudinal gradient of global atmospheric background concentrations and verify the reliability of CRUX-1.0 data against the same-region reference station, comparative analysis was conducted on concurrent CO2 and O3 datasets from five stations. It should be noted that stations with different latitudes and environments are not used to directly validate system performance, but to provide a global background context. The selected stations cover distinct geographical locations (polar vs. mid-latitude) and operational modes (unattended vs. manned), including Taishan Station (Antarctic inland, unattended), South Pole Station (Antarctic coastal, manned), Barrow Station (Arctic coastal, manned), Mauna Loa Station (mid-latitude marine, manned), and Jungfraujoch Station (mid-latitude alpine, manned) (Fig. 6a).

Comparative results revealed distinct regional and operational differences in both CO2 and O3 observations. For CO2, polar stations (Taishan and South Pole; mean: 420.7 and 418.2 ppm; hereafter means) exhibited significantly lower baseline mixing ratios than mid-latitude counterparts (Mauna Loa: 424.9 ppm; Jungfraujoch: 427.6 ppm), primarily due to the absence of anthropogenic emissions in the Antarctic inland. Taishan Station also showed smaller fluctuation amplitude (±1.5 ppm), attributed to the homogeneous air masses in the Antarctic inland circulation and interference-free operation of the unattended system, whereas mid-latitude stations displayed larger fluctuations driven by regional emission transport and diurnal cycles. For O3, polar stations (Taishan: 20.1 ppb; South Pole: 20.5 ppb) had much lower mixing ratios than mid-latitude stations (Mauna Loa: 48.9 ppb; Jungfraujoch: 47.1 ppb), which was associated with weak photochemical activity in the Antarctic late summer., which was associated with the low mixing ratios of ozone precursors in Antarctica (due to the dual scarcity of natural and anthropogenic sources) impair the ozone formation potential, while the intense sunlight in summer, chemical depletion triggered by trace amounts of NO, and stable boundary layer conditions jointly lead to surface ozone in Antarctica being dominated by depletion during summer (Tian et al., 2022).

The 1-month validation in this study only confirms the short-term feasibility of CRUX-1.0, which is a common constraint for proof-of-concept polar observation systems. From a broader perspective, long-term autonomous atmospheric observation in Antarctica generally faces three core common challenges: energy supply limitations under polar nights, long-term data quality assurance via autonomous calibration, and extreme environmental adaptability (e.g., blowing snow and ultra-low temperature).

For energy supply, CRUX-1.0 currently relies on Taishan Station’s diesel power infrastructure, which is limited by seasonal manual fuel replenishment. For long-term measurement accuracy, the single-point calibration of CRUX-1.0 is sufficient for short campaigns but has limited drift correction capacity over multi-month deployments.Targeting these challenges, we are conducting iterative upgrades of the CRUX-2.0 system, including a standalone PV-battery-methanol hybrid power system and autonomous multi-concentration calibration capability. The prototype is currently undergoing field validation at Saishiteng Mountain (Qinghai), with detailed parameters and test progress provided in the Supplement (Figs. S7 to S10).