The Regulating Role of Meteorology in the Wetland-Air CO₂ Fluxes at the Largest Shallow Grass-Type Lake on the North China Plain

Abstract

Lakes are hot spots of carbon cycles in inland aquatic systems. As a vital factor, meteorology, including air temperature, precipitation, wind speed and evapotranspiration, is profoundly affecting or even regulating the wetland-air CO₂ exchanges. Compared with some other similar lakes in China, the largest shallow grass-type Baiyangdian Lake (BYDL) acts as a vital CO₂ sink on the North China Plain. The purpose of this study is to reveal the effects of meteorology on the process of CO₂ flux variation. Based on the method of the eddy covariance, the daily average wetland-air CO₂ flux at the BYDL over the monitoring period from April 2019 to November 2020, reached −0.63 μmol m⁻² s⁻¹, and the annual average reached −0.71 μmol m⁻² s⁻¹ from 12 April 2019 to 12 April 2020. The CO₂ sink fluxes varied with the seasons and reached the maximum in summer. Temperature and evapotranspiration are two major driving factors, whose higher values can positively improve the wetland CO₂ sinks. Precipitation generally coincides with the CO₂ sinks, but the relatively larger summertime precipitation (0.39 m in 2020, compared with that of 0.17 m in 2019) inhibits the CO₂ uptakes on longer timescales. A moderate wind speed in the range of 1.6~3.3 m s⁻¹, promoted the CO₂ sinks for the shallow grass-type lake. Compared with previous studies at the same or similar wetlands, consistent CO₂ sink fluxes are found. Further in this study, the variation trends of CO₂ sinks with the changing meteorological factors are revealed for the first time in this type of wetland. Once meteorology is determined under both the anthropogenic and climatic impacts, the evaluation and prediction of the lacustrine carbon cycling could be more precise. Generally, this study will serve as an important data point into the global understanding of lake carbon fluxes.

1. Introduction

Lakes are significant regulators in the terrestrial carbon cycle, which act as either sources or sinks of CO₂ [1,2]. Dissolved CO₂ concentrations in inland waters are usually supersaturated relative to the atmosphere. This leads to most inland lakes to be characterized by high CO₂ emissions [3,4]. However, eutrophic lakes are generally undersaturated for CO₂ and tend to be a sink for the atmospheric CO_2, due to their high primary production [5,6]. In highly populated plains with a high intensity of industrial and agricultural activity, the anthropogenic pollution (including large amounts of nutrients) inputs and the manual water volume regulation deeply alter the hydrochemical equilibrium and the carbon cycles in lakes.

The Baiyangdian Lake (BYDL) is the largest shallow grass-type freshwater lake on the North China Plain (NCP), where both eutrophication and artificial water diversion are ongoing. Nevertheless, as the core water environment of the Xiongan New Area, it is a hot spot for regulating the regional CO₂ cycle. Previous studies on the wetland-air CO₂ flux at the BYDL found that the reed vegetation was abundant, and the sedimentation rate was much greater than that of other lakes, which was overall, a strong CO₂ sink area [7]. During the growing season, from June to October, the CO₂ absorption by the BYDL ecosystem was greater than its release. The nighttime CO₂ release rate was <2.33 g C m⁻² d⁻¹, and the daytime CO₂ absorption rate can be as high as 7.84 g C m-2 d-1, leading to a net absorption rate of 4.90 g C m⁻² d⁻¹ [8]. Niu et al. [9] measured the CO₂ release rate at the BYDL littoral zones from March to November and found them to be 0.12~3.49 g C m⁻² d⁻¹, using the closed static chamber method. Zhao and Wang [10] used static chamber-gas chromatography to obtain the CO₂ release fluxes from different wetland sub-zones in the BYDL, from May to October, which were consistent with the above results. They pointed out that areas with reeds exhibited lower CO₂ release fluxes than those without plants. The previous results show that the CO₂ release rate of the BYDL is much lower than the CO₂ absorption rate, indicating a net CO₂ sink area. Compared with other inland wetlands in China, the BYDL functions as an outstanding carbon sink [8].

In such a typical carbon sink lake with intensive human activities, how local meteorological factors regulate the wetland-air CO₂ fluxes is still unclear. Especially, the connections between the meteorological elements and the wetland-air CO₂ fluxes, urgently need to be revealed, in order to obtain the more precise predictive evaluation for the regional lacustrine carbon sinks. We hypothesize that meteorology plays a regulating role in the wetland-air CO₂ exchanges. To clarify the regulating influence, we quantify the air temperature, precipitation, wind speed, evapotranspiration, and wetland-air CO₂ fluxes over 18 months, during 2019~2020. In addition, we analyze the correlation between the above terms. We build on the literature by (1) applying the eddy covariance method to reveal the controlling effects of meteorology on the wetland-air CO₂ fluxes; and (2) by assessing the level of the CO₂ sinks at the shallow grass-type lake, and comparing with other lake-wetland systems.

2. Methods

2.1. Study Site

The Baiyangdian Lake (BYDL), known as the “Pearl of North China” is located in Xiongan New Area, Hebei province (Figure 1). It was the largest shallow freshwater eutrophic wetland in the NCP, bounded by embankments with a maximum area of about 366 km². Without significant inlets or outlets in the system, the wetland was under very weak flow conditions. Aquatic plants are widely inner-distributed, with the reed growing area of 94 km², accounting for 54.6% of the total wetland area [11]. Their growing season mainly covers from May to October. The study site is in a temperate continental monsoon climate, with an arid and windy spring, a hot and rainy summer, a cool and refreshing autumn, and a cold and dry winter. The average annual temperature is 12.1 °C, and the average precipitation is about 500 mm a⁻¹, with 80% concentrated in June to September. The multi-year average potential and the actual evapotranspiration at the BYDL basin are 1031.1 mm and 461.1 mm, respectively [12]. The study site is located in the north of the BYDL, where the reeds are widely distributed except for in the limited waterways. The annual water level variation is within 1 m (Figure 2), with a small flooded area change around the tower, during the observation period. The measurements at this typical location can be highly representative for the whole wetland.

2.2. In-Situ Monitoring of the Meteorological Data

The in-situ measurement, tower, and site design considerations followed Aubinet et al. [13] and Burba [14]. In 2019, a concrete platform was built in the Shaochedian Lake (the largest sub-lake in the north of the BYDL), with a distance of 1.5 km from the lakeshore. The LI-COR 7500 eddy covariance analysis system (Licor Inc., Lincoln, NE, USA) was set up on this platform, with a height of 10 m. The system includes a three-dimensional ultrasonic anemometer (Gill sensor, Lymington, UK), air temperature and pressure sensors, open-path infrared gas analyzer (LI-7500 DS), and rain gauges (Figure 1). The CO₂ flux on the lake surface (the positive value indicates the upward release, and the negative value is the downward absorption), and the meteorological data, including evapotranspiration, air temperature, wind direction/speed (u, v, w), and precipitation can be real-time monitored and transmitted online. The monitoring period started on12 April 2019 until 18 November 2020, covering a total of 586 days, and the actual number of days for obtaining the complete data was 574 days. All of the output data are averaged by half an hour, with a sampling frequency of 10 Hz and a fetch area of 1 km around the tower. The wetland surface cover under the tower was almost homogeneous, based on the investigation and the satellite image around the tower.

2.3. Eddy Covariance Approach

The daily evapotranspiration rates and the CO₂ fluxes at the lake-air interface can be obtained directly from the eddy covariance daily data. The daily total evapotranspiration/CO₂ fluxes can be best described as a function of the product of the vertical wind speed and vapor pressure/CO₂ concentration difference between the water surface and atmosphere. The vertical flux of the water vapor (F_ET) and CO₂ (F_c), can be written as [15]:

$$F_{ET}=\left(1+\mu \sigma \right)\left[\overline{w^{\prime }{\rho }_v^{\prime }}+\left(\overline{{\rho }_v}/\overline{T}\right)\overline{w^{\prime }{T}^{\prime }}\right]$$

(1)

$$F_c=\overline{w{\rho }_c}=\overline{w^{\prime }{\rho }_c^{\prime }}+\overline{w} \overline{{\rho }_c}\approx \overline{w^{\prime }{\rho }_c^{\prime }}$$

(2)

where μ = m_a/m_ν and σ = ρ_ν/ρ_a; m_a and m_ν are the molecular masses (‘weight’) of dry air and water vapor constituents, respectively; ρ_ν and ρ_a are the density of the water vapor and the dry air constituents, respectively; w is the vertical velocity of the dry air constituent; T is the absolute temperature; ρ_c is the density of CO₂ in the air. Covariances were calculated by first filtering the spikes of the 10 Hz sampled data and then using a 30 min block average. The coordinate frame was rotated using the planar-fit method [16] frequency domain corrections for the path length averaging and the sensor separation were applied [17], and the density fluctuations were accounted for in the calculation of the fluxes. The sonic temperature was used to calculate the sensible heat flux using the method suggested by Paw U et al. [18] which accounts for a missing energy balance term associated with the expansion of air during evaporation under a constant pressure. The fluxes were measured when the wind was blowing from the direction within ±5° of the back of the anemometer. Approximately 2% of the data were omitted due to possible interference from the anemometer support and the IRGA mounted behind the anemometer.

2.4. Data Process and Correlation Analysis

Due to the meteorological influences, instruments and human disturbance, etc., the daily evapotranspiration sequence obtained by the field monitoring often loses data or exist gaps. In order to eliminate the impact of the data volatility of the time series, multiscale moving averages can reveal the characteristics of the land surface’s physical or physiological processes inherent in different cycles of the sequence, while inhibiting the randomness of the features [19]. The original data lost 456 items, leading to a deletion rate of 1.62%. When obtaining the dynamic curves of the meteorological elements, including the CO₂ flux, a 10-day sliding average after excluding the spikes is performed to eliminate the sudden change of the special cases or the data loss (Figure 2). The smooth and continuous sliding average curve can clearly reflect the temporal variation trend. The quality control and data filtering also followed Morin et al. [20].

However, when analyzing the meteorological characteristics on a monthly or seasonal scale, the daily mean after the elimination of the spike is directly used, to exclude the staggered time from the sliding average approach. With the help of SPSS statistics software (IBM^®), the correlation analyses were conducted between the meteorological factors and the wetland-air CO₂ exchange fluxes.

Last but not least, it should be clarified here that though the emission or absorption of CO₂ by an ecosystem is basically related to its interior physiological processes, the meteorological factors are focused in this paper to reveal the meteorological influencing trends on the wetland-air CO₂ fluxes. It is defined as the total CO₂ flux through the interface between the wetland, not only the water, and air. Thus, plant (i.e., reeds shown in Figure 1) photosynthesis and its fundamental factor of the solar radiation are not analyzed or emphasized.

3. Results

3.1. Meteorological Conditions

The Baiyangdian Lake (BYDL) is located in a typical temperate semi-arid monsoon climate. From the 10-day sliding average monitoring data of the temperature, precipitation, wind speed, and evapotranspiration (Figure 3A), it can be seen that during the monitoring period, the rain and heat were concentrated in the same period, mainly in summer (Figure 3B,C and Figure 4E). Wind speed is greater in spring than in other seasons, which decreases with the temperature rising and falling (Figure 3D and Figure 4C). Daily evapotranspiration is positively correlated with the daily average temperature and wind speed (Figure 4A,B). The relatively heavier precipitation corresponds to a wind speed range of 1.9~3.4 m s⁻¹, with a lighter precipitation under either faster or slower wind speeds (Figure 4D). For the two gray bands in Figure 4D,E, it is shown that the relatively high precipitation occurred on relatively hot days (temperature range of 23~28 °C) and light breezes (wind speed range of 1.9~3.3 m s⁻¹). At a smaller precipitation, the daily evapotranspiration rate fluctuates in a large range. As the rainfall increases (>80 mm), the evapotranspiration tends to converge to the annual average (4.39 mm d⁻¹, Figure 4F). Dividing the four seasons by month (winter: December to February, spring: March to May, summer: June to August, autumn: September to November), the seasonal variations of the above four meteorological elements are characterized (Figure 3B–E). The temperature, precipitation, and evapotranspiration show a good consistency of summer > spring > autumn > winter, while the wind speed is different, with a variation trend of spring > summer > autumn > winter.

3.2. Temporal Variations of the Wetland-Air CO₂ Exchange Fluxes

Based on the monitored data from the LI-COR 7500 eddy covariance system, it was found that the BYDL is a net CO₂ sink area overall, with a monitoring period averaged absorption rate of −0.63 μmol CO₂ m⁻² s⁻¹ (Figure 2). The monthly and seasonal average wetland-air CO₂ exchange fluxes present regular temporal fluctuations (Figure 5A). From May to September, it acts as a net sink of CO₂, especially in the summer when the reeds grow vigorously, while the remaining months, from October to April, show a weak source and sink switches. The seasonal average wetland-air CO₂ net fluxes (unit in μmol m⁻² s⁻¹, Figure 5B) decrease in the order of summer (−1.57) > autumn (−0.40) ≈ spring (−0.36) > winter (−0.10).

3.3. Relations between the Wetland-Air CO₂ Fluxes and the Meteorological Variables

The wetland-air CO₂ flux is affected by the meteorological elements in different ways. From Figure 6, the relationship between them is revealed: (1) For daily average air temperature (Figure 6A): the sources and sinks of CO₂ are basically balanced in spring and autumn (10~22 °C) and winter (<10 °C), while in summer (>22 °C), it behaves as a strong sink. (2) For the daily rainfall (Figure 6B): there is no obvious trend for the wetland-air CO₂ flux, as the precipitation is rising. Nevertheless, the CO₂ sinks are generally more common than the sources during the precipitation events, of varying intensities. (3) For the daily average wind speed (Figure 6C): the CO₂ absorption is slightly stronger than the release when the wind is light air (0.3~1.5 m s⁻¹) or a gentle breeze (3.4~5.4 m s⁻¹), while the sinks are obviously more than the sources, when the wind is a light breeze (1.6~3.3 m s⁻¹). The CO₂ releases tend to be dominant when the wind is at a moderate breeze (5.5~7.9 m s⁻¹). (4) There is a general negative correlation between the wetland-air CO₂ flux with the daily evapotranspiration (Figure 6D). When the evapotranspiration is relatively weak (<5 mm), the sources and sinks of CO₂ are basically balanced. When the evapotranspiration is moderate (5~10 mm), the absorption is significantly larger than its release. When the evapotranspiration is strong (>10 mm), they are overwhelming CO₂ sinks.

Further, Pearson correlation analyses (

Table 1. Correlation coefficients between the wetland-air CO₂ fluxes and the meteorological factors.

	Temperature	Precipitation	Wind Speed	Evapotranspiration	Wetland-Air CO2 Flux
Temperature	1
Precipitation	0.099 *	1
Wind speed	0.062	0.093 *	1
Evapotranspiration	0.732 **	0.025	0.468 **	1
Wetland-air CO2 flux	−0.485 **	−0.012	0.051	−0.406 **	1

Note: * and ** indicate the significant correlation at the 0.05 and 0.01 level, respectively.

) obtain results consistent with Section 3.1. There is a good positive correlation between the temperature and precipitation, and between the precipitation and wind speed (p < 0.05). Evapotranspiration is significantly positively correlated with the air temperature and wind speed (p < 0.01, Figure 4A,B). The wetland-air CO₂ fluxes are significantly inversely correlated with the temperature and evapotranspiration (p < 0.01), which can also be obtained from Figure 6A,D.

4. Discussion

4.1. The Role of Meteorology in the Wetland-Air CO₂ Exchanges

From Section 3.3, we can see that the relatively higher temperature (>22 °C, summer), moderate wind speed (1.6~3.3 m s⁻¹, light breeze) and stronger evapotranspiration (>5 mm d⁻¹, especially >10 mm d⁻¹) will jointly promote the wetland CO₂ absorption. For different level precipitations, from light to downpour, it is generally, but not obviously, beneficial for the wetland CO₂ absorption. The monitoring period covered two whole summers, in which seasons the CO₂ exchange rates are much higher than in the other seasons. By comparing the daily average CO₂ flux at the wetland-air interface during the two summers (

Table 2. Daily average meteorology and wetland-air CO₂ flux at the Baiyangdian Lake (BYDL) in the summers of 2019 and 2020.

Summer of the Year	Daily Average Temperature (°C)	Daily Average Precipitation (mm)	Daily Average Wind Speed (m s−1)	Daily Evapotranspiration (mm)	Daily Wetland-Air CO2 Flux (μmol m−2 s−1)
2019	26.22	1.94	2.16	7.32	−2.10
2020	25.94	4.26	2.15	6.84	−1.03
Ratio of 2019 to 2020	1.01	0.46	1.01	1.07	2.03

), it was found the flux of 2019 was twice as high as that of the 2020. The temperatures, wind speeds, and evapotranspiration are close to each other in both summers, but the precipitation of the former is less than 50% of that of the latter. It is likely that the heavier precipitation generally corresponds to weaker light conditions. On the one hand, this hinders the stomata opening of the reeds and reduces the evapotranspiration or CO₂ absorption. On the other hand, the respiration can be enhanced, inhibiting the foliar absorption of CO₂. On the daily scale, precipitation is beneficial to the carbon sink of the reed wetlands, however, on a longer time scale of the quarter, more precipitation reduces evapotranspiration to a certain extent, and obviously weakens the carbon sink effect.

4.2. Comparisons of the Wetland-Air CO₂ Fluxes with Previous Studies

The CO₂ absorption or release intensity varies in the different lakes, and the flux direction may be reversed. Even in the same lake, the CO₂ fluxes in different lake areas or seasons may be reversed. The values for the global lakes range from −0.17~0.54 μmol m⁻² s⁻¹ with a mean of 0.18 μmol m⁻² s⁻¹ (

Table 3. Comparison of the wetland-air CO₂ absorption fluxes between the BYDL and other similar lakes.

Site	Wetland-Air CO2 Flux (μmol m−2 s−1)	Reference
Dongtinghu Lake	−0.04 (daily average)	[23]
Chaohu Lake	−0.10 (daily average)
Hongzehu Lake	−0.17 (daily average)
Erhai Lake	−0.01 (daily average)
Dianchi Lake	−0.12 (daily average)
Donghu Lake	0.37 ± 0.29 (winter), −0.02 ± 0.06 (spring) −0.04 ± 0.11 (summer), 0.04 ± 0.09 (autumn)	[22]
Lake Daming	−0.10~0.04 (summer)	[24]
Taihu Lake	−2.23~0.19 (annual average: −0.73)	[25] (Monthly statistics)
Lake Batur	−0.03	[26]
Ngoring Lake	−0.70~−0.13 (June to September)	[27]
Yindeer Lake	−0.84~0.24 (annual average: −0.26)	[28]
Panjin reed wetland	−2.75 (June~September 2004) −3.04(June~September 2005)	[21]
Qinghai Lake	−0.84 ± 0.37 (Ice-covered) −0.40 ± 0.34 (Ice-free)	[29]
BYDL	−2.34~−0.33 (average: −1.29, June~October)	[8]
BYDL	−2.24~−0.74 (average: −1.72, June~October 2019) −1.79~0.34 (average: −0.57, June~October 2020)	This study (Monthly statistics)
BYDL	−0.10 (winter), −0.36 (spring) −1.57 (summer), −0.40 (autumn) −0.63 (average over the monitoring period) −0.71 (annual average, from 12 April 2019 to 12 April 2020)	This study (Seasonal statistics)

in [4]). The CO₂ absorption flux at the BYDL is compared with those of other lakes with CO₂ sinks (Table 3). It was found that the BYDL has a relatively higher CO₂ sink capacity on the annual scale. The CO₂ sink fluxes (−2.24~0.34 μmol m⁻² s⁻¹) from June to October are comparable to that (−2.34~−0.33 μmol m⁻² s⁻¹) from a previous study at the BYDL [8]. This CO₂ absorption flux of −1.57 μmol m⁻² s⁻¹ in summer, is on the same order of magnitude as that of −2.75~−3.04 μmol m⁻² s⁻¹ in the Panjin reed wetland in the growing season (June–September, [21]). On the seasonal or annual average levels, the CO₂ absorption flux at the BYDL is higher than or close to those at other similar lakes, as shown in Table 3. For the seasonal variations of the different lakes, the summers show a stronger absorption or weaker releases of CO₂ than the winters ([22] and this study).

4.3. Limitations of the Current Study

In this study, the meteorological elements are mainly analyzed to reveal the relations between them and the wetland-air CO₂ fluxes at the largest shallow grass-type lake. Actually, some other factors, including eutrophication [30], the hydrological situation [31], the spatial distribution [32], the global climate change [33], the biogeochemical function, etc., can also influence the wetland-air CO₂ fluxes. Eutrophication causes clear lakes to become a strong CO₂ sink, whereas humus lakes become a stable CO₂ source after the addition of external nutrients [34]. Hard-water lakes in central North America gradually changed from CO₂ sources to sinks with an increase in the annual mean pH value [35]. Previous monitoring data indicate that the water of the Baiyangdian Lake is alkaline and eutrophicated, which may be other reasons for its carbon sinks. Thus, further wetland CO₂ flux evaluations and regulations should consider the more complicated multi-parameter interactional ecosystem networks. Under the dual effects of global climate change (e.g., the temperature increase may enhance carbon sinks) and the artificial water diversion, which can change the lacustrine chemical environment and may promote carbon releases, it is necessary to further observe and judge how the meteorology and CO₂ flux at the BYDL will change and be controlled.

5. Conclusions

This study clarified the temporal variation of the CO₂ flux at the wetland-air interface at one typical site of the Baiyangdian Lake (BYDL), and its relationship with four main meteorological factors, including the air temperature, precipitation, wind speed, and evapotranspiration. Based on the eddy covariance method and the correlation analysis, the temporal variations of the CO₂ flux at the largest shallow grass-type lake on the North China Plain and its response laws to the local meteorological factors are revealed. Among the meteorological factors, temperature and evapotranspiration were significantly correlated with the wetland-air CO₂ fluxes. The relatively higher temperature (in summer) and stronger evapotranspiration contributed to a greater CO₂ absorption than in other seasons. However, the too slow or fast wind speeds will not help the wetland CO₂ sinks. Precipitation basically corresponds to the carbon sinks, while on an annual scale, more rainfall will hinder the CO₂ sink by decreasing evapotranspiration and increasing respiration. These findings are of scientific significance for the lacustrine eco-environmental evaluation and the carbon sink evolution trend analysis. In order to more comprehensively depict the CO₂ flux at the wetland-air interface in lakes under an anthropogenic effect and global climate change, further studies relating to the more complex physical or biogeochemical processes need to be taken into account.