Description of the Study Area

The Liaohe River Delta is one of the eight major river deltas in the world, with rich water resources and diverse wetland resources. Panjin Wetland Paddy Field Ecological Observation Station is located in Dawa District, Panjin City, which is Panjin Rice Field Experimental Station (121°57′E, 40°55′N, altitude 4.0 meters) of Shenyang Atmospheric Environment Research Institute of China Meteorological Administration. The observation field covers an area of 25×25 m, and the crop planted is japonica rice. The research area is located north of Bohai Bay, belonging to the continental semi–humid monsoon climate of the north temperate zone, with the rainy and hot seasons in the same period and four distinct seasons. The average annual precipitation is 631 mm, the sunshine duration is 2780 hours, and the annual average temperature is 9.5℃. The highest and lowest temperatures occur in July and January, respectively [23, 24]. Statistically, the dominant surface wind direction from 2020 to 2022 is S, SSW, followed by NE (Fig. 1).

Observation Data and Methods

Collecting carbon flux data for the rice paddy comes from the open–path eddy covariance flux observation system. The observation site is located in an open and flat area without obstructions from buildings. The ground crops are all rice, and the available area for the experiment is 30 mu (about 2 hectares). The eddy covariance carbon flux observation system consists of a three-dimensional ultrasonic anemometer (CSAT3), an open-path CO₂/H₂O analyzer (Li-7500), an open-path CH₄ analyzer (Li-7700), and a data acquisition unit (Li-7550). The sensors are positioned at a height of 4.2 meters above the ground, with a sampling frequency of 10 Hz. The 30-minute time series data was obtained through data processing using EddyPro (v7.0.9).

The eddy covariance method can directly measure the fluxes of CO₂, CH₄, and other gasses between vegetation and the atmosphere. However, during nights when the atmospheric structure is relatively stable and turbulence is weak, the diffusion of fluxes may not reach the measurement height of the instrument, reducing the accuracy of the measurements. Additionally, open-path eddy covariance systems are susceptible to weather conditions such as rainfall and fog, as their infrared optical paths are exposed. This can lead to anomalous and missing values in the observational data. We identify an anomaly for the anomalous and missing values in the data when the absolute difference between the data at a certain moment and the average data from the previous five moments is greater than five times the variance. Based on the actual conditions at this site and existing related research, the reasonable range of carbon flux values is set to -60 μmol•m^-2•s^-1 to 60 μmol•m^-2•s^-1 [25]. At the same time, linear interpolation is used to supplement data gaps of less than 2 hours. For data gaps greater than 2 hours, based on different months, the carbon flux is interpolated separately for daytime and nighttime by establishing relationships between carbon flux and radiation, as well as between carbon flux and ground temperature. When radiation and ground temperature data are also missing, relationships with meteorological factors are established for interpolation. The interpolated data account for 4.8% to 8.3% of the total data volume (Fig. 2). When meteorological data from the ecological station is missing, data from a meteorological station located 11.4 kilometers away is used as a supplement.

Fig. 2.

Description of the missing NEE data by day and hour from 2020 to 2022. The left column (Figure a, Figure c, Figure e) represents the missing data due to power and/or instrument failure or low data quality; the right column (Figure b, Figure d, Figure f) represents the interpolated data.

Calculation of Ecosystem Parameters

Ecosystem gross primary productivity (GPP) refers to the amount of CO₂ fixed into organic matter by green plants through photosynthesis per unit area and time. Ecosystem respiration (Reco) includes aboveground plant respiration and soil respiration. This process releases most of the carbon fixed by primary productivity in the ecosystem into the atmosphere in the form of CO₂, which is an important way for the exchange between the soil carbon pool and the atmospheric carbon pool, playing a key role in maintaining the carbon balance of terrestrial ecosystems [26]. The sensitivity of ecosystem respiratory temperature (Q₁₀) refers to the multiple increases in the ecosystem respiration rate when the temperature rises by 10℃, which is used to study the response of ecosystem respiration to climate change [27]. The specific calculation formula is as follows:

Where: T is the soil temperature; a and b are coefficients to be determined.

Annual Changes in Carbon Fluxes and Meteorological Factors

Meteorological elements are important environmental factors that affect crop growth. Crop growth is very sensitive to changes in meteorological conditions, especially temperature, precipitation, and sunshine duration [28, 29]. From 2020 to 2022, the annual total NEE at the ecological station was negative but was positive during the non-growing season. The growing season was defined as April 16^th to October 14^th (same below, see Table 1). Among them, the CO₂ absorption value in 2021 was the largest, reaching −432.89 gCm^-2•y^-1, while the carbon absorption values in 2022 and 2020 were lower than −360 gCm^-2•y^-1. The carbon flux during the growing season also showed the largest absorption value in 2021, with 2020 slightly lower than 2022. The variation in carbon absorption during the growing season was up to 79.8 gCm^-2•y^-1, significantly higher than that during the non–growing season. Contrary to the global warming trend, the temperature in the Panjin wetland has decreased in the past three years, while the annual precipitation has been increasing, reaching a maximum of 895.7 mm in 2022. The annual sunshine duration was highest in 2020, reaching 3372.7 hours, which was significantly higher than in 2021 and 2022. The average wind speed was lowest in 2021 and highest in 2022, with a variation of 0.2–0.4 m•s^-1. NEE is largest in 2021 in the three–year comparison; the factors that contributed to its promotion were higher temperatures and more precipitation. The impact of meteorological elements on carbon flux will be analyzed in detail in the following context.

Characterization of Daily Changes in Carbon Fluxes

We respectively calculated the diurnal variation of NEE in spring (March to May), summer (June to August), autumn (September to November), and winter (December and January to February) from 2020 to 2022 (Fig. 3). The seasonal differences in the diurnal variation of NEE are significant. The diurnal variation of NEE in summer and autumn exhibits a "decreasing first and then increasing" pattern, while the variation in winter and spring is less obvious. The daily accumulation of NEE is negative in summer and autumn, which means it acts as a carbon sink. The daily value in summer is much higher than in autumn, with a daily accumulation of −4.28 to −4.82 gCm^-2•d^-1 in summer and −0.35 to −0.61 gCm^-2•d^-1 in autumn. The daily cumulative value in spring and winter is positive, indicating a carbon source.

The diurnal variation of carbon flux in the three months of spring showed obvious differences; sometimes, the value was positive at night and turned negative in the midday of the day. Sometimes, there was no negative value during the day, and the value was positive all day. To describe the characteristics of carbon flux variation in spring in detail, the time periods are divided based on the appearance of significant negative values during the daytime. Each year, significant negative values during the daytime typically occur after late May. The dates when negative values appear in other periods are not fixed. For example, in 2020, negative values appeared in early March, while in 2021, they occurred in early May. In 2022, there were fewer days with negative values. The time when the negative value of carbon flux occurs in Panjin in spring is relatively late, mostly after late May. The main reason is that Panjin is located in the north, with long winters, low temperatures in spring, and a short growing season. The diurnal variation of carbon flux in summer remains relatively consistent over many years. At 6:00 am, with the increase in solar radiation, the photosynthesis of vegetation is enhanced, and the amount of CO₂ absorbed by photosynthesis exceeds the amount released by the ground. At this time, the amount of CO₂ on the ground surface is converted from release to absorption, and NEE is converted from positive to negative. The carbon absorption reaches its maximum at 12:00 noon. At 18:00 in the evening, NEE changes from negative to positive, and the ground surface changes from carbon absorption to carbon emission. In July, the negative value of NEE is the largest, reaching −20.6 to −26.0 μmol∙m^-2∙s^-1, slightly lower in August than in July and smaller in June.

Fig. 3.

Average diurnal variation of NEE in different seasons from 2020 to 2022. The NEE value (μmol∙m^-2∙s^-1) represents the average hourly value of carbon flux in the paddy fields in Panjin within the specified date range. Among them, Spring: Figures a1-a3; Summer: Figures b1-b3; Autumn: Figures c1-c3; Winter: Figures d1-d3.

Fig. 4.

Comparison of Daily NEE Values in the Growing and Non-Growing Seasons of Panjin Paddy Fields from 2020 to 2022.

After mid–October in autumn, there were no significant negative NEE values, and the dates varied slightly each year. In 2020, the date was earlier, on October 15^th, while in 2022, it was later, on October 20^th. In 2021, it was October 19^th. The NEE value turned from positive to negative at 7:00 am and from negative to positive at 17:00 pm, one hour later and earlier than in summer. The maximum carbon absorption occurred at 11:00 or 12:00, reaching −8.2 to −9.4 μmol∙m^-2∙s^-1. During the entire winter, the negative values of NEE during the daytime were relatively small. NEE turned negative from 8:00 to 9:00 am and positive at 16:00 pm. The maximum value occurred between 11:00 and 13:00, ranging from −0.2 to −0.4 μmol•m^-2•s^-1.

Statistics on the annual variation of daily carbon flux over a three–year period have been conducted (Fig. 4). Centered on summer, the distribution of the daily value of NEE throughout the year shows a "decreasing first and then increasing" pattern. However, the daily cumulative value of NEE did not show a negative value starting from April 16^th, the beginning of the growing season, and only after May 20^th did the daily value of NEE turn negative. After October 15^th, the daily value of NEE shifted from negative to near zero. It is worth noting that from April to mid–May, the positive value of NEE was slightly higher, indicating that during this period, the crop's aerobic respiration was stronger, but photosynthesis was weaker. This feature was also observed in late October and early November but was weaker than in the spring. The maximum daily cumulative value of NEE occurs in the middle and late July, ranging from −11.3 to −16.0 gCm^-2•d^-1.

Relationship between NEE and Meteorological Factors

Sunshine, temperature, precipitation, etc., are very important natural conditions for the growth of surface vegetation, and the total amount of precipitation and the time distribution of precipitation in the growing season determine the size and duration of the net carbon absorption function of the ecosystem [30, 31]. Fig. 5 shows the daily variation comparison of NEE, precipitation, temperature, sunshine, and other meteorological factors during the active period of carbon absorption (from May 20^th to October 14^th). NEE reached its maximum in late July, and during the period of high NEE values in summer, the daily variation was relatively large, with a difference between the high and low values ranging from 16.65 to 24.30 gCm^-2•d^-1. Precipitation was high in early 2021; for example, it exceeded 75 mm on June 10^th and June 27^th, but it was low from late July to early August and then increased again in the later period. The precipitation was low in early 2022 but increased significantly in late July and then decreased again afterward. Although the total precipitation in 2022 was the largest, the NEE value was the highest in 2021, indicating that the effect of early– season precipitation on carbon flux was greater than the annual precipitation. The distribution performance of temperature change and NEE are in the opposite phase. The peak values from late July to early August correspond to high values of NEE. From July 23^rd to August 1^st, 2020, when NEE exceeded −9 gCm^-2•d^-1, the daily temperature was higher than 26.7℃. From July 24^th to August 8^th, 2021, when NEE exceeded −9 gCm^{- 2}•d^-1, the daily temperature was higher than 26.0℃. And from July 18^th to August 2^nd, 2022, when NEE exceeded −9 gCm^-2•d^-1, the daily temperature was higher than 26.1℃. When the temperature is low, the corresponding NEE is also low.

Fig. 5.

Comparison of Daily NEE Values (gCm^-2•d^-1) in Panjin Paddy Fields from May 20^th to October 14^th Each Year from 2020 to 2022 with Changes in Daily Values of Precipitation (mm), Temperature (℃), Wind Speed (m/s), Specific Humidity (100%), and Sunshine Duration (h).

The variation of specific humidity is similar to temperature, and the corresponding specific humidity is larger when precipitation and temperature are higher, but the variation range is lower than that of temperature. Sometimes, high NEE values occur even when the temperature is not particularly high but the specific humidity is relatively large. For example, on July 17, 21, and 24, 2022, the specific humidity exceeded 79.5%, and the NEE value exceeded −10.0 gCm^-2•d^-1, while the daily temperature was below 24.3℃. The variation pattern of wind speed is not significant. The average wind speed in July over the three years ranged from 2.14 to 2.60 m/s, while the average wind speed in September ranged from 1.91 to 2.87 m/s. The wind speed characteristics are not significant when the NEE value is high. The average wind speed from July 15^th to August 15^th, 2020, is 3.02 m/s, and the average wind speed is 3.82 m/s when NEE is greater than −9 gCm^-2•d^-1. The average wind speed from July 15^th to August 15^th, 2021, is 1.99 m/s, and the average wind speed is 2.21 m/s when NEE is greater than −9 gCm^-2•d^-1. The average wind speed from July 15^th to August 15^th, 2022, is 2.27 m/s, and the average wind speed is 1.64 m/s when NEE is greater than −9 gCm^{- 2}•d^-1. It can be seen that the correspondence between wind speed and high values of NEE is not significant. The sunshine value corresponding to the high value of NEE is also high. For example, from July 15^th to August 15^th, 2021, the average sunshine duration was 5.27 hours, and the average sunshine duration when NEE was higher than −8 gCm^-2•d^-1 was 9.02 hours. From July 15^th to August 15^th, 2022, the average sunshine duration was 5.52 hours, and the average sunshine duration when NEE was higher than −8 gCm^-2•d^-1 was 7.43 hours. The slightly lower sunshine values were due to precipitation on 3 days.

Temperature, precipitation, sunshine, and specific humidity all promote NEE. Among several meteorological factors, temperature has the most significant correspondence with changes in NEE over a long period, followed by precipitation and specific humidity. The main reason for the high carbon absorption value in the growing period in 2021 was the higher precipitation in the early stage and the higher temperature in the later stage of 2021.

The NEE values were compared between the growing and non–growing seasons for temperature ranges of 15– 30℃ with an interval of 5℃ over a period of 3 years (Table 2). In the non–growing season, temperatures below 15℃ accounted for the majority, reaching up to 96.15 %, while the remaining only accounted for 3.85%. The NEE values for all temperature ranges during the non–growing season were positive, indicating carbon release. During the growing season, the proportion of temperatures below 15℃ was 15.79%, with the highest proportion of 36.38% falling between 20–25℃, and the lowest proportion of 2.42% above 30℃.

Fig. 6.

Comparison of daily variations in hourly GPP (μmol∙s^-1∙m^-2) values in rice paddies in Panjin from 2020 to 2022 in different months, with Figures a-d representing June to September, respectively.

During the growing season, when the temperature is below 15℃, the NEE value is positive, similar to the non–growing season. However, the NEE values are negative for other temperature ranges, indicating carbon absorption. The NEE value increased significantly with the rise of temperature during the growth season, and the mean value of NEE was the largest at −8.079 μmol∙m^-2∙s^-1 when the temperature was higher than 30℃. It is worth noting that the mean value of NEE increased similarly when the temperature rose by 5℃ within the range of 15℃ to over 30℃, specifically, −2.36, −2.74, and −2.68 μmol∙m^-2∙s^-1.

Characteristics of Daily Changes in GPP

The hourly daily changes of GPP from June to September of 2020–2022, which are the months when crops grow most vigorously and concentratedly, were counted respectively (Fig. 6). The diurnal variation of GPP was significantly different in each month. The diurnal variations from June to September all exhibit a pattern of "increasing first and then decreasing", with June being less prominent. The GPP value was the highest in July, with an average daily accumulation of 10.69–13.55 gCm^-2•d^-1, followed by August, with an average daily accumulation of 10.13–10.89 gCm^-2•d^-1, and the lowest in June, ranging from 3.68–5.43 gCm^{- 2}•d^-1. The months of July and August in summer are the hottest and most humid periods of the year, when the photosynthesis capacity of crops is the strongest, leading to the highest proportion of gross primary productivity. According to statistics, the total GPP in July accounts for 28.6–31.1% of the annual total, while August accounts for 24.1–29.7%. The total GPP for July and August accounts for 52.7–60.8% of the annual total. During the diurnal variation, the GPP value gradually increases starting at 5:00, reaches its peak around 12:00 noon, and then begins to decline, falling to a low value between 19:00 and 20:00. The interannual differences in GPP's nighttime variation are relatively small, while the diurnal variation differences are significant. Among them, the diurnal variation in July is the most significant, with the diurnal variation in July 2021 being higher than in other years. Correspondingly, the total GPP in 2021 was also the highest.

Fig. 7.

Comparison of daily GPP (gCm^-2•d^-1) changes in Panjin rice paddies during the growing and non-growing seasons from 2020 to 2022.

In the annual variation of daily GPP values (Fig. 7), the trend is "increasing first and then decreasing", centered on summer. The high values appear from late July to early August, with the maximum daily value ranging from 20.75 to 23.51 gCm^-2•d^-1. Starting from May 20^th, the daily GPP value gradually increases, with the average value reaching its maximum on July 23^rd and declining continuously, dropping to a low value by October 15^th. In the early growing season, from April 16^th to May 20^th, the total primary productivity of crops was relatively low, with an average GPP of 0.35–1.23 gCm^-2•d^-1. After October 15^th, the GPP value remained at a relatively low level, averaging 0.68 gCm^-2•d^-1.

Relationship between Carbon Fluxes and Net Radiation

Solar shortwave radiation is the main energy source of the atmosphere, and net radiation is the main controlling factor for the diurnal variation of NEE in the growing season. We have statistically analyzed the correlation between the diurnal variations of NEE and net radiation from June to September 2020–2022 (Fig. 8). The relationship between NEE and net radiation is a quadratic polynomial with negative growth. From July to September, the correlation values between the two variables are relatively high, with the highest correlation value occurring in July, where the R² of each year is above 0.98. In June, however, the correlation values are lower, with R² ranging from 0.64 to 0.83. During the day, when net radiation is greater than 0, NEE is negative, indicating that the vegetation absorbs CO₂ through photosynthesis. At night, when net radiation is less than 0, NEE becomes positive, indicating that the vegetation releases CO₂ through aerobic respiration. These two processes alternate. Starting at 6:00 in July, net radiation turned positive (ranging from 7.2 to 50.3 W/m²), indicating that photosynthesis in the vegetation exceeded respiration, switching from releasing CO₂ to absorbing it. The NEE value ranges from −1.21 to −6.54 μmol•m^-2•s^-1. After that, as net radiation increases, the ability of vegetation to absorb CO₂ gradually enhances, and the NEE value also increases, reaching a peak between 11:00 and 12:00. At this time, the net radiation value is 370.78–382.83 W/m2, and the NEE value is −20.57 to −25.95 μmol•m^-2•s^-1. Then it begins to decline, and at 18:00, net radiation turns negative, and NEE turns positive between 18:00 and 19:00. The variation of NEE is sometimes lagging behind and sometimes ahead of net radiation, with most of the time lagging behind and sometimes ahead in the morning. For example, in August 2021, the peak of net radiation appeared at 11:00 (310.26 W/m²), while the peak of NEE appeared at 12:00 noon (−21.94 μmol•m^-2•s^-1). In August 2022, net radiation turned positive at 7:00 (64.75 W/m²), and NEE turned negative at 6:00 (−1.75 μmol•m^-2•s^-1).

Fig. 8.

Corresponding relationship between net radiation (W/m²) and NEE (μmol∙m^-2∙s^-1). The symbols of triangle, square, and plus represent the corresponding values of the two in different years, and the curve represents the quadratic polynomial relationship between the two. Figures a-d represent the period from June to September.

Ecosystem Respiration (Reco) as Affected by Soil Temperature

Soil temperature is crucial for ecosystem respiration. The study pointed out that when precipitation is no longer a limiting factor for vegetation growth, temperature has a greater impact on plant respiration [32, 33]. There is a clear exponential relationship between the two (Fig. 9), where Reco increases with the rise in soil temperature. The correlation between Reco and soil temperature in each season is significant, passing the 0.01 correlation test. Among them, the respiratory intensity of summer crops was relatively high, ranging from 4.54 to 5.68 gCm^-2•d^-1, with the highest respiratory intensity in the summer of 2020 reaching 5.68 gCm^-2•d^-1. Respiratory intensity in summer is 0.7 to 6.7 times higher than in other seasons; autumn is 0.4 to 2.6 times higher than spring and winter, and winter has the weakest respiratory intensity.

The amplitude of the respiratory intensity of summer crops with soil temperature is higher than in other seasons. The variation of summer ground temperature in the three years ranged from 18.2 to 29.7℃, and the variation of Reco ranged from 1.1 to 11.4 gCm^{- 2}•d^-1. For every 1℃ increase in soil temperature, the respiratory intensity increased by 0.71 to 0.85 gCm^-2•d^-1. The maximum increase in Reco with soil temperature occurred in 2022, reaching 0.85 gCm^-2•d^-1. In autumn, the increase in Reco per 1℃ rise in soil temperature ranges from 0.13 to 0.22 gCm^-2•d^-1. In spring, the increment of Reco per 1℃ rise in soil temperature ranges from 0.07 to 0.13 gCm^-2•d^-1. In winter, the increment of Reco per 1℃ rise in soil temperature is only 0.05 to 0.08 gCm^-2•d^-1.

In the exponential correlation between Reco and soil temperature, autumn exhibited the highest correlation values in 2020–2022, followed by summer, spring, and winter in descending order (see Table 3). The highest correlation value in autumn (R²=0.85) was observed in 2020, while the highest correlation value in summer (R²=0.69) was recorded in 2022. Q₁₀ is an important parameter that shows the interaction between terrestrial ecosystems and climate systems. The characteristics of Q₁₀ and respiration can help us better understand the interaction between climate change and the carbon cycle of terrestrial ecosystems [34, 35]. The magnitude of Q₁₀ depends on the variation of Reco with soil temperature. When Reco varies more significantly with soil temperature, the Q₁₀ value increases. During the three– year observation, Q₁₀ was highest in summer, ranging from 2.89 to 6.37, followed by autumn with values between 1.67 and 1.91. Both spring and winter showed lower Q₁₀ values than autumn. The Q₁₀ value in summer is 2 to 3 times higher than in other seasons, indicating that NEE, GPP, and Reco of the ecosystem are at high levels in summer, and the temperature sensitivity of crop respiration is also the strongest. In summer and autumn, the Q₁₀ values in 2021 were the highest, 6.37 and 1.91, respectively, and the corresponding annual NEE values were also the highest.

Fig. 9.

Relationship between vegetation respiration (Reco: gCm^-2•d^-1) and soil temperature (Ts:℃) in different seasons. The symbols of square, circle, and triangle represent the corresponding values of the two in different years, and the curve represents the exponential relationship fitting between the two. Figures a-d represent spring, summer, autumn, and winter, respectively.