Introduction

Oxygen is one of the most essential gases in the earth’s atmosphere, approximately 20.95% by volume. Its concentration directly affects living organisms’ survival and ecosystem stability1. At high altitudes, the partial pressure of oxygen decreases significantly due to lower atmospheric pressure and reduced air density. In some areas of the Tibetan Plateau, the oxygen partial pressure is only approximately 50% of that at sea level2. Such low oxygen levels compromise human health (residents and workers) and challenge the survival and adaptation of plants, animals, and microorganisms3. High Altitude Sickness (HAS) caused by a low oxygen environment mainly manifests as headache, nausea, fatigue, and insomnia, among other symptoms. Studies have shown that hypoxemia risk increases significantly when humans are exposed to > 2500 m altitudes4. Prolonged exposure to low oxygen may also lead to Chronic Mountain Sickness (CMS), characterized by polycythemia, pulmonary hypertension, and cardiac abnormalities5,6. These environmental health problems become more severe with more prolonged exposure periods.

The Tibetan Plateau is the largest and highest plateau globally. Its peculiar geography has attracted much attention to the near-surface atmospheric oxygen concentration7. Previous studies have indicated that the near-surface oxygen concentration of the Tibetan Plateau is closely related to the combined effects of elevation, topography, climate, watershed, vegetation, and soil8,9. Moreover, various regions and seasons also have significantly different oxygen concentrations10. In summer, the oxygen level decreases by 0.15% for every 1000 m increase in elevation, while a 10 °C increase raises the oxygen level by 0.21%. Additionally, a 10% increase in vegetation cover raises the oxygen concentration by 0.02%. Generally, oxygen concentration in summer is higher than in winter11. At the same time, the oxygen levels differ among various land use and land cover areas12. The potential correlation between oxygen concentration and precipitation, as well as the differences in the response mechanisms of oxygen concentration to precipitation between two types of bare land (dry bare land and ice-snow melting bare land), remains to be thoroughly investigated. Addressing this issue will enhance our understanding of the environmental drivers regulating oxygen concentration in high-altitude regions.

The Qinghai-Tibet Plateau has a large proportion of bare land (i.e., 618,400 km2, only second to grassland, 1,512,200 km2). The bare land includes sand, Gobi desert, saline land, bare land, bare rock, bare gravel land, and other vegetation covers (< 10%)13,14. In recent years, global warming has accelerated glacier retreat and snow melting, thus significantly increasing the area of ice-snow melting bare land in the Tibetan Plateau region. The ice-snow melting bare land is a bare land formed by the melting of the covered ice and snow and the retreat of the snow line due to the influence of climate warming on the permanent snow and glaciers in the mountains15. For example, the spatial and temporal dynamics of the bare land in the ice-snow melting area of Yushu, located in the hinterland of the Tibetan Plateau, shows a continuously expanding bare land in the ice-snow melting area (by 2,450.34 km2 in the past 30 years), accounting for 2% of the total area of the prefecture (including Tanggula Town, administered by Golmud)16. This change alters surface coverage and profoundly impacts the ecosystem, particularly on near-surface atmospheric oxygen concentration17. Additionally, extensive areas of dry bare land exist in the Qinghai-Tibet Plateau. For instance, the northwestern parts of the Tibetan Plateau and the Qaidam Basin are located in the inland hinterland, where precipitation is scarce and the climate is cold and dry. These conditions result in the distribution of an extensive range of dry bare land in these areas, thus, significantly hindering the ecosystem’s oxygen production18,19. Oxygen is essential for life and ecosystem stability. Due to minimal vegetation and photosynthesis, bare land contributes little oxygen, potentially intensifying localized hypoxia and impacting human health and biodiversity. Investigating oxygen variations across different bare land types on the Tibetan Plateau enhances understanding of plateau hypoxia and informs assessments of climate change effects on ecosystem function.

The bare land areas of the Qinghai-Tibet Plateau are predominantly characterized by cold and dry climates, resulting in low vegetation cover and limited photosynthesis20,21. Therefore, this paper aimed to obtain systematic near-surface oxygen concentration (hereinafter referred to as oxygen concentration) data of the Tibetan Plateau through field research, and combine it with statistical analysis to reveal changes in oxygen concentration at various bare land types (ice-snow melting and dry bare lands) under various elevations, temperatures, and precipitations. This study would help to better understand the ecological service function of bare land on the Tibetan Plateau and provide scientific evidence to address the health risks of hypoxia in the plateau region.

Study area

The Qinghai-Tibet Plateau is characterized by complex and diverse topography, with an average elevation of 4000 m above sea level. The region exhibits various climatic conditions, influenced mainly by elevation and topography. High altitude areas experience low temperatures, with annual averages below zero, while the river valleys at lower altitudes are relatively warm. Most plateau areas are cold and dry, with short summers and long winters. The thinness of the air and the diversity of climatic conditions in the region significantly impact the spatial and temporal distribution of oxygen levels. There, hypoxia is a crucial environmental hazard factor for the region’s population and economic systems. In this paper, the Qaidam Basin and the Yushu Plateau were selected as typical regions for study (Fig. 1). The Qaidam Basin is a highland dry zone with sparse precipitation (a multi-year average of < 100 mm) and a widely distributed dry and bare land. The Yushu Plateau in Qinghai belongs to the typical plateau alpine climate, with an average temperature of 0.4 °C, precipitation of 487.7 mm, and an annual snowfall period from October to June22. Glaciers and snow cover are widespread in the region. However, due to the impact of global climate warming, the melting of ice-snow has formed ice-snow melting bare land. Therefore, the study area was a representative region for investigating the spatial and temporal patterns of oxygen concentration in high elevation dry and ice-snow melting bare lands.

Fig. 1
figure 1

Location of the study area (Maps generated by ArcGIS 10.7(https://www.esri.com/en-us/arcgis/products/arcgis-desktop/overview)).

Full size image

Data and methods

Data collection

The data used in this paper (Table 1) comprised the following: (1) existing oxygen concentration spatial distribution data, including the average relative oxygen concentration in January, and July, and the annual average; (2) field-measured oxygen concentration data in the dry bare land area; (3) dry and ice-snow melting bare land data; (4) topographic and meteorological data, including elevation, temperature, and precipitation; and (5) supplementary data, such as administrative divisions of the study area and the boundary data of the Qinghai-Tibet Plateau. To ensure consistency and facilitate subsequent analysis, the spatial resolution of all aforementioned datasets was standardized to 1 km through resampling, aligning with the spatial resolution of the oxygen concentration data.

Table 1 List of data used in this study.
Full size table

Oxygen concentration spatial distribution data

The spatial distribution dataset of oxygen concentration was obtained through field observations during our research group’s Second Tibetan Plateau Scientific Expedition. Between 2017 and 2022, oxygen concentration data and corresponding geographical and environmental data were collected from 807 measured sampling points across the Tibetan Plateau. The temporal and spatial patterns of near-surface relative oxygen concentration across the Tibetan Plateau were estimated from these data23.

Measured oxygen concentration data

From April 30 to May 6, 2024, our group conducted field measurements in the Qaidam Basin (Fig. 1). Based on road accessibility, 133 samples were measured at 10 km intervals. At each measurement point, latitude, longitude, and elevation were recorded using a GPS monitor; oxygen concentration was measured using three units of TD400-SH-O2 portable oxygen concentrators (all measurements in this study are relative oxygen concentration), with a 0.01% accuracy; similarly, three units of DPH-103 digitally measured the temperature, humidity, and pressure.

To minimize measurement errors, the mean of the triplicate measurements was taken as the final result for each point. Because the oxygen was measured instantaneously in the quasi-static state of the air, the data were obtained in a windless environment as much as possible and read after the instrument was stabilized to minimize the effect of convective disturbances on the data.

Bare land data

The bare land data included dry and ice-snow melting data (Fig. 1). The dry bare land was a synthesized product of our group study using MOD09A1 data for 8 days, with a spatial resolution of 500 m till August 29, 2023. The projection was redefined to the Albers Conic Equal Area projection using the MODIS Reprojection Tool (MRT). In ENVI 5.3 software, the pixel dichotomy model calculated the vegetation coverage. The confidence intervals for NDVI soil and NDVI vegetation were determined by selecting the cumulative 2% and 98% percentiles of the NDVI estimates. Then the threshold of 15% of vegetation cover (less than or equal to) was regarded as dry bare land.

Ice-snow melting bare land was the bare land data formed after snow and ice melting in 1990–2020. We used the S3 snow index model and NDVI to calculate the snow and ice cover data. The S3 snow index model takes into account the reflection characteristics of ice, snow and vegetation in different bands, effectively reducing the impact of vegetation cover on ice-snow pixel recognition, and at the same time increasing the distinction between snow and clouds, and further improving the mapping accuracy of snow cover under vegetation cover. When S3 ≥ 0.29, and when 0.08 ≤ S3 < 0.29 and NDVI ≥  − 0.12, we can estimate that a pixel is an ice-snow pixel. There were some missing points in individual spots with low snow and ice coverage and low thickness. Multiple situations involved tributaries of rivers and lakes. A man–machine interactive visual interpretation was subsequently used to revise the missing and multi-divided spots. The difference set of the two periods of snow and ice covered data for the ice-snow melting area. Combined multi-indicator factor composite analysis and object-oriented classification obtained the ice-snow melting bare land data. In order to verify the accuracy of the classification, we used the DJI UAV M210 and GPS field sampling. The overall accuracy of this method was 83.59% and the Kappa coefficient was 0.79 for the extraction of bare land in the ice-snow melting area of the Yushu Plateau16.

Topographic and meteorological data

The ArcGIS 10.7 preprocessed the DEM data, primarily involving image mosaic data, cropping, and projection transformation. The preprocessing of meteorological data involved the MATLAB R2019a platform writing codes for processing, while the temperature data enabled the statistical processing of annual mean, winter, and summer temperatures. The precipitation data determined the statistical processing of annual, winter, and summer precipitations. From these data, the correlation coefficients of the temperature and precipitation dataset were higher than 0.95 and 0.90, respectively 24,25.

Analysis methods

Statistical methods

The statistical methods used in this study were simple linear regression analyses and the linear mixed model (LMM). Simple linear regression investigated the correlation between oxygen concentration and various geo-environmental factors (elevation, temperature, and precipitation). This method is suitable for quantitatively assessing the strength of the effect of each independent variable on oxygen concentration, revealing the influence of geo-environmental factors on the variation of oxygen concentration under varied conditions26. In this analysis, oxygen concentration was the dependent variable while geo-environmental elements were the independent variables. The linear relationship between each geo-environmental factor and oxygen concentration was assessed by calculating the regression coefficients, and the model’s fit was evaluated by the coefficient of determination (R2).

The LMM was used to analyze the effects of geo-environmental factors such as elevation, temperature, and precipitation on oxygen concentration and assess their relative contributions. The strength of LMM lies in its ability to account for both fixed and random effects, capturing unmeasured between-group variation and reducing systematic errors. Fixed effects estimate the influence of variables at the aggregate level, while random effects model the hierarchical structure or between-group variation, enhancing inference robustness. Failure to account for random effects may lead to sample correlation within regions, violating the assumption of independent errors and compromising parameter estimation reliability.

To evaluate the relationships among variables and assess potential multicollinearity, a correlation analysis was performed on elevation, temperature, and precipitation data. The results revealed a significant negative correlation between elevation and temperature, while the correlations between elevation, temperature, and precipitation were relatively weak. Furthermore, variance inflation factor (VIF) analysis was conducted to detect collinearity, and all VIF values were below 5, indicating that multicollinearity was not a concern in the model. Therefore, in the final model, the oxygen concentration was the dependent variable, while elevation, temperature, and precipitation were the fixed-effect independent variables, and the geo-environmental region was a random effect. The model formula is as follows:

$$y=X\beta +Zb+\epsilon$$
(1)

where \(y\) represents oxygen concentration, \(X\) is the design matrix for the geo-environmental factors, \(\beta\) is the regression coefficient for the fixed effects, \(Z\) is the design matrix for the random effects, \(b\) is the random effect, and \(\epsilon\) is the error term. Through this model, we quantified the relative contribution of each geo-environmental factor to oxygen concentration and decomposed the explanatory rate to determine the relative impact of elevation, temperature, and precipitation across the environments.

Validation of the accuracy of existing oxygen concentration data

To verify the accuracy of an existing dataset on the spatial distribution of annual mean oxygen concentration, the oxygen concentration data of 133 field measurements in the dry bare land area served as validation samples. The correlation between the measured data and the pre-existing dataset was evaluated using linear regression analysis (Fig. 2). The coefficient of determination (R2) was 0.71, indicating a strong positive correlation; the P-value was less than 0.001, indicating the correlation was highly statistically significant. These results showed that the existing oxygen concentration dataset reliably reflected the actual distribution of oxygen levels in the study area, meeting the requirements for further analysis.

Fig. 2
figure 2

Correlation between existing oxygen concentration data and measured oxygen concentration data.

Full size image

Results

Spatial distribution of oxygen concentration in bare land types

In the dry bare land

The overall oxygen concentration of dry bare land ranged from 19.84 to 20.40%. Spatially, the areas with relatively high oxygen concentrations were mainly in the flat basin floor of the Qaidam Basin. In contrast, the Altun Mountains in the northwest, the Qilian Mountains in the north, and Kunlun Mountains in the south had relatively low oxygen concentrations due to high topography (Fig. 3).

Fig. 3
figure 3

Spatial distribution of near-surface oxygen concentration in dry bare land (Maps generated by ArcGIS 10.7(https://www.esri.com/en-us/arcgis/products/arcgis-desktop/overview)).

Full size image

The annual average oxygen concentration in dry bare landranged from 19.93 to 20.24%, averaging 20.16%, with 60.56% of the values falling between 20.15 and 20.20 and 20.74% between 20.10 and 20.15%. Concentrations above 20.20% accounted for 11.74%, while those below 20.10% comprised 6.96%.

In January, the oxygen concentration ranged from 19.84 to 20.10%, averaging 20.05%, 0.11% lower than the annual mean. Most January concentrations were between 20.05 and 20.10% (64.22%), followed by 20.00 to 20.05% (26.08%). Only 0.72% of the concentrations exceeded 20.10%, and 8.96% were below 20.00%.

In July, the oxygen concentration ranged from 20.03 to 20.40%, averaging 20.27%, 0.11% higher than the annual mean. Concentrations above 20.25% dominated (76.50%), followed by those between 20.20% and 20.25% (17.93%). Only 5.57% of the concentrations were below 20.20%.

In ice-snow melting bare land

The ice-snow melting bare land was primarily located on the high mountain ridges and those of the plateau. With relatively small spatial variation, the oxygen concentration (19.61–20.29%) in these areas was typically lower than that in dry bare land (Fig. 4).

Fig. 4
figure 4

Spatial distribution of near-surface oxygen concentration in ice-snow melting bare land (Maps generated by ArcGIS 10.7(https://www.esri.com/en-us/arcgis/products/arcgis-desktop/overview)).

Full size image

The annual average oxygen concentration in ice-snow melting bare landranged from 19.71 to 20.11%, averaging 19.88%, 0.28% lower than that of dry bare land. Most of the annual oxygen concentrations (85.56%) fell between 19.80 and 19.95%, while 9.49% were above 19.95%, and 4.95% were below 19.80%.

In January, the oxygen concentration ranged from 19.61 to 20.01%, averaging 19.79%, 0.09% lower than the annual mean for ice-snow melting bare land and 0.26% lower than that of dry bare land. Concentrations between 19.75 and 19.85% accounted for 73.45%, with 12.20% above 19.85% and 14.35% below 19.75%.

In July, the average oxygen concentration ranged from 19.82 to 20.29%, averaging 19.99%, 0.11% higher than the annual mean for ice-snow melting bare land but 0.28% lower than that of dry bare land. Concentrations between 19.90 and 20.05% accounted for 77.38%, while 18.82% are above 20.05%, and 3.80% are below 19.90%.

Relationship between bare land oxygen concentration and elevation

In dry bare land

A significant negative correlation existed between oxygen concentration and elevation in dry bare land for the annual average, January, and July. The R2 values were 0.91, 0.88, and 0.68, respectively, with the p-values less than 0.001 (Fig. 5).

Fig. 5
figure 5

Relationship between oxygen concentration and elevation in dry bare land. The elevation and oxygen concentration frequency are shown on the top and right sides, respectively. *** Indicates that the correlation reaches a significance level of P < 0.001.

Full size image

The elevation range in dry bare lands was relatively small (2675–3641 m). For every 1000 m increase in elevation, the annual average oxygen concentration in dry bare land decreased by 0.133% (Fig. 5a).

In January, the oxygen concentration decreased by 0.132% for every 1000 m increase in elevation (Fig. 5b), slightly lower than the rate for the annual average. In July, the decrease was 0.119% per 1000 m (Fig. 5c), lower than the of annual average and January. This difference may be attributed to higher temperatures and more active air circulation in summer, which mitigated the suppressive effect of elevation on oxygen concentrations.

In ice-snow melting bare land

The annual average, January, and July oxygen concentrations in ice-snow melting bare land also correlated negatively and significantly with elevation, with respective R2 values of 0.80, 0.78, and 0.57, and the p-values less than 0.001 (Fig. 6).

Fig. 6
figure 6

Relationship between oxygen concentration and elevation in ice-snow melting bare land. The elevation and oxygen concentration frequency are shown on the top and right sides, respectively. *** Indicates that the correlation reaches a significance level of P <0.001.

Full size image

The elevation range in ice-snow melting bare land was significant (4438–6188 m). On average, for every 1000 m increase in elevation, the oxygen concentration decreased by 0.134% (Fig. 6a), slightly higher than the decrease rate for the annual average oxygen concentration in dry bare land. This difference may be attributed to lower atmospheric pressure and temperature during snow and ice melting, making the oxygen level drop more pronounced at higher elevations.

In January, the oxygen concentration decreased by approximately 0.132% for every 1000 m increase in elevation (Fig. 6b). This rate was similar to that observed in dry bare land during the same month, indicating that elevation consistently impacted the oxygen concentration across both winter environments. This similarity may stem from the stable atmospheric conditions during cold weather, leading to a uniform degree of air thinning at high elevations.

In July, the oxygen concentration decreased by 0.126% for every 1000 m increase in elevation (Fig. 6c), slightly higher than the decreasing rate of 0.119% for dry bare land. Although the reduction in oxygen concentration in July for ice-snow melting bare land was less pronounced than in January, and was the annual average, the influence of elevation remained significant compared to dry area.

Relationship between bare land oxygen concentration and temperature

In dry bare land

The annual average, January, and July oxygen concentrations in dry bare land were significant and positively correlated with temperature (Fig. 7). The R2 of annual mean oxygen concentration with temperature was 0.92, 0.89 in January, and 0.67 in July; all correlations were highly statistically significant (P < 0.001), indicating that a temperature increase raised the oxygen level significantly in dry bare land. The statistical analyses results suggested that the monthly oxygen concentration response to temperature changes varied. On an annual average basis, within the same area, a 10 °C increase in temperature raised the oxygen concentration by 0.166% in dry bare land (Fig. 7a). In January, within the same area, a 10 °C rise resulted in an approximate 0.162% increase in oxygen concentration (Fig. 7b), slightly lower than the annual average. In July, within the same area, a 10 °C increase corresponds to an approximate 0.156% rise in oxygen concentration (Fig. 7c). These results indicated that temperature was crucial to the oxygen concentrations in dry bare land, with varying impacts across various seasons.

Fig. 7
figure 7

Relationship between oxygen concentration and temperature in dry bare land. Temperature and oxygen concentration frequency are shown on the top and right sides, respectively. *** Indicates that the correlation reaches a significance level of P < 0.001.

Full size image

Ice-snow melting bare land

The annual average, January, and July oxygen concentrations in ice-snow melting bare land correlated significantly and positively with temperature (Fig. 8). The R2 of annual mean oxygen concentration with temperature was 0.83, 0.87 in January, and 0.56 in July. All correlations were statistically significant (P < 0.001), indicating that a temperature increase significantly increased oxygen concentration in the ice-snow melting bare land. On average annually, within the same area, a 10 °C temperature increase would raise oxygen concentrations by 0.163% in ice-snow melting bare land (Fig. 8a). In January, within the same area, a 10 °C rise resulted in an approximate 0.151% increase in oxygen concentration (Fig. 8b), slightly lower than the annual average. In contrast, oxygen levels in July were more sensitive to temperature changes, within the same area, with a 10 °C increase corresponds to an approximate 0.176% rise in oxygen concentration a (Fig. 8c). These results indicated that temperature was crucial to oxygen concentration in ice-snow melting bare land, with varying impacts across the seasons. The positive effect of temperature rise on oxygen concentration was most pronounced during the summer.

Fig. 8
figure 8

Relationship between oxygen concentration and temperature in ice-snow melting bare land, with temperature and oxygen concentration frequency shown on the top and right sides, respectively. *** Indicates that the correlation reaches a significance level of P < 0.001.

Full size image

In Figs. 5, 6, 7 and 8, the relatively lower R2 values for the relationship between July oxygen concentration and elevation/temperature are primarily attributed to higher summer temperatures, which enhance vegetation and microbial activity, as well as intensify atmospheric convection, leading to increased fluctuations in oxygen content. Additionally, seasonal variations in meteorological conditions (e.g., precipitation, humidity) further contribute to the variability in oxygen levels, making the relationship between oxygen concentration and elevation/temperature more complex.

Relationship between bare land oxygen concentration and precipitation

In dry bare land

The annual average, January, and July oxygen concentrations in dry bare land correlated with precipitation significantly and negatively, with R2 values of 0.34, 0.32, and 0.27, respectively. Here, all p-values were below 0.001 (Fig. 9). It showed that precipitation affected the oxygen concentration to an extent. From the average value, within the same area, for every 100 mm increase in precipitation, the oxygen concentration decreased by 0.023% in dry bare land (Fig. 9a).

Fig. 9
figure 9

Relationship between oxygen concentration and precipitation in dry bare land, with precipitation and oxygen concentration frequency shown on the top and right sides, respectively. *** Indicates that the correlation reaches a significance level of P < 0.001.

Full size image

In January, within the same area, a 100 mm increase in precipitation corresponded to a 0.678% decrease in oxygen concentration (Fig. 9b) at a rate higher than the annual mean oxygen concentration. In July, within the same area, a 100 mm increase in precipitation lowered the oxygen concentration by 0.036% (Fig. 9c), and was lower than in January.

In the dry bare land area, precipitation rapidly increased soil moisture, and wet soils had lower gas diffusion coefficients, lowering the near-surface oxygen levels. Lower temperatures and a smaller temperature gradient between soil and air further reduced oxygen diffusion rates during winter. In contrast, in summer, higher temperatures promoted active gas exchange, increasing oxygen diffusion rates. The elevated temperature also accelerated moisture evaporation, rapidly reducing soil moisture and alleviating the suppressive effect of precipitation on soil oxygen levels.

Ice-snow melting bare land

The oxygen concentrations in ice-snow melting bare land showed no significant correlation with precipitation annually, in January or July, with R2 values of 0.01, 0.03, and 0.01, and 0.73, 0.12, and 0.16 p-values, respectively (Fig. 10). It showed that precipitation did not significantly affect the oxygen content of ice-snow melting bare land.

Fig. 10
figure 10

Relationship between oxygen concentration and precipitation in ice-snow melting bare land, with temperature and oxygen concentration frequency shown on the top and right sides, respectively.

Full size image

This land type was primarily distributed in high mountain regions above 5000 m, where the climate was sub-normal, the temperature was low year-round, and the soil was frozen for extended periods, except for brief thaw periods. In such environments, soil physical structure and moisture conditions were primarily governed by the ice-snow melting process rather than precipitation. The water supply from ice-snow melting greatly exceeded the contribution from precipitation, thus minimizing its impact on oxygen concentration. At elevations above 5000 m, soils experienced frequent freeze–thaw cycles, a physical process that significantly influenced soil structure and oxygen diffusion capacity. Freeze–thaw cycles created compact soil structures, which reduced gas diffusion rates. However, these structural changes were predominantly temperature driven, making variations in precipitation relatively minor in their impact on soil oxygen concentration.

Contribution of elevation, temperature, and precipitation to oxygen concentration in bare land

The LMM results indicated that elevation, temperature, and precipitation collectively determined 77.23% and 79.42% of oxygen concentration variations in dry and ice-snow melting bare land, respectively. This result suggested that these three environmental variables significantly explained oxygen concentration changes in the two types of bare land (Fig. 11). Specifically, in dry bare land, the elevation, temperature, and precipitation rates on oxygen concentration were 42.89%, 27.98%, and 6.36%, respectively. In contrast, they accounted for 46.33%, 32.50%, and 0.59%, respectively, in ice-snow melting bare land. Elevation and temperature were the primary drivers of oxygen concentration variation in these two types of bare land. In contrast, the influence of precipitation was more limited, particularly in ice-snow melting areas. This difference may be attributed to the climatic characteristics, elevation, and precipitation forms associated with each type of bare land. In dry bare land, the impact of precipitation on oxygen concentration was often indirect and delayed; increased precipitation promoted vegetation growth, enhanced ecosystem photosynthesis, and raised atmospheric oxygen levels. However, vegetation could not be established in ice-snow melting bare land, which mainly consisted of exposed bedrock after ice-snow melting. Most precipitation occurred in solid form, making it unlikely to affect oxygen concentration changes, directly.

Fig. 11
figure 11

Contribution of elevation, temperature, and precipitation to oxygen concentration in bare land.

Full size image

Comparative analysis of oxygen concentration between bare land and other surface cover types

In this field sampling, comparative observation samples were set up in the dry bare area of the “Ring of Qaidam Basin” to better understand the differences in oxygen concentration between bare land and other surface cover types. Arid bare land can be categorized into natural bare land and artificial bare land. Natural bare land is defined as the exposed surface formed primarily through the influence of natural factors, encompassing types such as sand, Gobi desert, Yardang, salt crust, bare rock, bare sand, and bare soil. In contrast, artificial bare land refers to exposed surfaces resulting directly or indirectly from human activities, typically associated with land use changes, resource exploitation, or engineering construction, including construction site, road, and railway. The basic principles for setting comparative samples were as follows: the distance between paired observation points within each comparison group was less than 1 km, and the time interval between measurements was less than 30 min (except for temperature comparisons). Based on the above principles, eight groups of comparative observation sample data were measured, six groups of data were comparative data on oxygen concentration of bare land and other land types, and two groups of data were data on the effect of temperature change on oxygen concentration of the same bare land data. The measurement points are listed in Table 2.

  1. a.

    Comparing the oxygen concentration between artificial bare land and grassland: Fig. 12a,b show that the oxygen concentration in artificial bare land (20.56%, in road) was lower than the same as that in grassland (i.e., 20.57% in desert grassland in the early greening stage).

  2. b.

    Comparing the oxygen concentration between natural bare land and grassland: Fig. 12i–l evince that the oxygen concentrations in natural bare land (bare sand and bare soil) and grassland (saline dry reed) were the same (20.42% and 20.46%).

  3. c.

    Comparing the oxygen concentration among bare land subtypes: Fig. 12e,f illustrate that the oxygen concentrations in bare land subtypes, salt crust (dry salt flats) and yardang (bare soil), were identical (i.e., 20.44%).

  4. d.

    Comparing the oxygen concentration between bare land and water: Here, one data set compared bare gravel and ice surface (Fig. 12c,d with the oxygen concentration slightly higher (by 0.01%) in bare land than on the ice surface. The other dataset (Figs. 12g,h also indicates that the bare land oxygen concentration was 0.01% higher than that of the lake.

  5. e.

    Comparing the effect of temperature variation on oxygen concentration at the same location: One dataset showed higher oxygen concentration (by 0.37%) on artificial bare land in the evening (21.30 pm) than in the morning (7:39 am) with a temperature difference of 6.30 °C the evening concentration at the same location. The other dataset compares the oxygen concentration of bare sand over a 60-min interval, indicating that, in the absence of vegetation, an increase in temperature by 0.67 °C corresponded to a 0.03% increase in oxygen concentration.

Table 2 Comparative observation data of oxygen concentration between bare land and other surface cover types.
Full size table
Fig. 12
figure 12

Field sampling photos comparing oxygen concentrations between bare land and other types of surface cover (These photographs were captured by JIA Wei during field surveys).

Full size image

Meanwhile, from July 19 to 28, 2023, our group obtained five datasets comparing oxygen concentrations between bare land and other surface cover types in the “surrounding Qilian Mountains and Qinghai Lake.” Among these, three datasets were compared between bare land and grassland. The primary grassland types included alpine grassland, freeze–thaw grassland, and overgrazed grassland. Results showed that the oxygen concentration in grassland (20.66%) was consistently higher than that in bare land (20.64%). The other two datasets compared oxygen concentrations between bare land and wetland, where the wetland type was alpine wetland grassland. The oxygen concentration in wetlands (20.57%) was higher than in bare land (20.55%). Thus, the oxygen concentration in vegetated areas was higher than in bare land regions.

Discussion

Shi et al. quantified the relative contributions of elevation, temperature, and vegetation cover to oxygen concentration on the Tibetan Plateau as − 47%, 32%, and 3%, respectively, explaining 82% of the contribution to the oxygen concentration, as elevation is not an absolute controlling factor9. This paper focuses on the spatial distribution characteristics of oxygen concentration in the bare land of the Tibetan Plateau and its primary influencing factors. Given the negligible or absent vegetation cover in bare land, vegetation was excluded as a variable, while precipitation was incorporated into the analysis. The results demonstrate that the factors influencing oxygen concentration in bare land are generally consistent with those reported by Shi et al. Elevation remains the most significant contributor, followed by temperature. The strong negative impact of elevation on oxygen concentration was closely influenced by thinner air and decreased atmospheric pressure at higher elevations. Additionally, atmospheric pressure variations driven by temperature changes played a crucial role in oxygen distribution. Specifically, higher temperatures led to the formation of thermal low-pressure systems, whereas lower temperatures resulted in cold low-pressure systems. These pressure systems contributed to localized oxygen concentration accumulation and dispersion.

In a comparison of oxygen concentrations in bare land with other surface cover types, it was found that bare ground has relatively low oxygen concentrations, and these differences can be rationalized based on the very low oxygen production of bare land. The low oxygen concentration in bare land could be primarily ascribed to its inability to produce oxygen and possible oxygen depletion processes. First, as a land type lacking vegetation cover, the bare land ecosystem experienced negligible photosynthesis, a process by which plants absorb carbon dioxide and release oxygen, making it one of the significant sources of oxygen on Earth27. However, this vital oxygen production was almost nonexistent in bare land due to the lack of green vegetation. Therefore, bare ground could not produce and replenish atmospheric oxygen through photosynthesis as in vegetated areas28.

We deduced that the oxygen concentration of bare land was only dependent on the original atmospheric oxygen level and could not be replenished by localized photosynthesis29. Second, bare land may also be subject to rapid soil organic matter decomposition and carbon release due to soil exposure and wind erosion. This process may consume some oxygen and reduce oxygen levels over bare land. Meanwhile, the oxygen concentration of the ice-snow melting bare land was lower than that of the dry bare land, mainly because the ice-snow melting bare land was usually located at higher elevations, with lower temperatures and relatively lower air density. This condition diffused the oxygen concentration, thereby lowering the oxygen concentration. In addition, many microorganisms may accompany ice-snow melting, and these microorganisms consume oxygen when decomposing organic matter, reducing the oxygen content in the localized area30.

The influences of wind speed, soil moisture, microbial activity, and local atmospheric circulation on oxygen concentration cannot be overlooked. Wind speed plays a crucial role in regulating oxygen exchange and transport within the near-surface layer by affecting gas diffusion rates and atmospheric mixing, thereby exerting a significant impact on oxygen concentration. Soil moisture can influence oxygen availability by affecting soil porosity, aeration, and microbial respiration processes-higher moisture levels may reduce soil oxygen diffusion, while also enhancing microbial activity. Microbial activity, particularly soil microbial respiration, consumes oxygen and releases carbon dioxide through the decomposition of organic matter. The metabolic intensity of these microorganisms is regulated by environmental factors such as temperature, humidity, and organic matter availability, indirectly influencing local oxygen concentrations. Furthermore, local atmospheric circulation redistributes oxygen spatially and temporally by driving cyclic air movements. However, compared to dominant factors such as elevation and temperature, the contributions of these variables are relatively minor. Additionally, due to the lack of direct observational data, this study does not incorporate them into the analysis. Future research should explore the combined effects of wind speed, air exchange, soil moisture, and microbial activity on oxygen concentration at larger spatial and temporal scales.

Additionally, there exists a temporal mismatch between the time span of the oxygen concentration dataset used in this study (2017–2022) and the formation period of ice-snow melting bare land (1990–2020). The evolution of ice-snow melting bare land is primarily driven by long-term processes such as climate change, and ice- snow melting, whereas oxygen concentration is influenced by short-term meteorological conditions and biogeochemical processes. Consequently, discrepancies in the temporal resolution of these datasets may introduce uncertainties in the analysis. Future studies should focus on optimizing data integration methods to enhance spatial and temporal consistency, thereby improving the reliability of the results.

Meanwhile, although this study utilized data collected over multiple years from various regions, the lack of repeated, multi-seasonal field measurements at fixed locations introduces uncertainty in evaluating the model’s temporal stability and generalizability. This limitation is mainly due to the logistical and environmental challenges associated with conducting continuous in situ monitoring in the harsh and remote terrain of the Tibetan Plateau. Future work should aim to incorporate long-term, multi-temporal field campaigns to more comprehensively assess oxygen concentration dynamics under different seasonal and climatic conditions.

Conclusions

This study revealed the spatial distribution patterns and responses of near-surface oxygen concentrations in dry bare land and ice-snow melting bare land to elevation, temperature, and precipitation in the Tibetan Plateau. Furthermore, the data quantified the contributions of various environmental factors to oxygen concentration.

The annual average oxygen concentration in dry bare land ranged from 19.93 to 20.24%, with the highest values concentrated in the relatively flat basin areas, while lower concentrations were found in the high-elevation mountainous regions. The oxygen concentration in ice-snow melting bare land (19.71 to 20.11%) was marginally lower than in dry bare land. Additionally, the oxygen concentrations in bare land types were higher in summer than winter.

Elevation negatively impacted the oxygen concentration. For dry bare land, the annual average oxygen concentration decreased by 0.133% for every 1000 m increase in elevation but was reduced by 0.132% and 0.119% in January and July, respectively. In ice-snow melting bare land, the annual average oxygen concentration decreased by 0.134%, with a 0.132% and 0.126% decrease in January and July, respectively. Conversely, temperature correlated positively with oxygen concentration. For dry bare land, within the same area, the annual average oxygen concentration increases by 0.166% for every 10 °C rise in temperature, with increases of 0.162% in January and 0.156% in July, respectively. Meanwhile, in ice-snow melting bare land, within the same area, the annual average oxygen concentration increased by 0.163%, 0.151% in January, and 0.176% in July for every 10 °C rise in temperature. Precipitation negatively affected oxygen concentration in dry bare land, particularly in winter, when it significantly inhibited oxygen diffusion, while its impact in ice-snow melting bare land was insignificant.

Consequently, elevation and temperature were the primary driving factors affecting oxygen concentration in bare land. The total explanatory rates of elevation, temperature, and precipitation for oxygen concentration in dry bare land was 77.23%, with elevation, temperature, and precipitation, contributing 42.89%, 27.98%, and 6.36%, respectively. Comparatively, the elevation and temperature rates were higher in ice-snow melting bare land, i.e., 46.33% and 32.50%, respectively, but lower for precipitation (0.59%).