Heavy Rainfall Increases CO₂ Emissions from Rivers in a Typical Human-Impacted Region

Abstract

Rivers emit substantial amounts of carbon dioxide (CO₂) to the atmosphere, yet its response to heavy rainfall remains unclear with intensive anthropogenic disturbances. To fill the knowledge gap, this study investigated the dynamic variability of CO₂ partial pressure (pCO₂) and CO₂ emissions flux at the Chaohu Lake Basin, a watershed under intensive anthropogenic perturbations, based on field campaigns across diverse river systems during dry season, normal season, and post-rainfall periods. Results demonstrated marked differences in aquatic pCO₂ across river types, with urban rivers (3949 µatm) exhibiting significantly higher levels than non-urban counterparts (1423 µatm). Rainfall events elevated riverine pCO₂, but the effect size varied between river types (urban river versus non-urban river). In non-urban rivers, pCO₂ following heavy rainfall (2461 μatm) was significantly higher (p < 0.05) than those observed during both dry season (1096 μatm) and normal season (712 μatm). In contrast, urban rivers demonstrated only marginal pCO₂ elevation after rainfall (20–30%). Statistical analysis revealed that discharge, total nitrogen (TN), total phosphorus (TP), and ammonia nitrogen (NH₄⁺-N) showed significantly positive correlations with pCO₂, while dissolved oxygen (DO) and pH exhibited significantly negative correlations with pCO₂. Overall, rivers in the Chaohu Lake Basin act as significant sources of atmospheric CO₂, with an annual mean CO₂ emission flux of 297.84 mmol·m⁻²·d⁻¹, and the heavy rainfall events amplify riverine CO₂ emissions (629.91 mmol·m⁻²·d⁻¹), with observed enhancement effects exceeding 300% compared to baseline conditions. To accurately estimate the CO₂ emissions from human-dominated rivers, future research should emphasize the impacts of extreme or heavy rainfall events.

1. Introduction

Inland waters are significant sources of atmospheric greenhouse gases (GHGs) [1,2,3], among which rivers and streams play a central role in the land–atmosphere carbon exchange. Despite covering less than 20% of the total freshwater surface area, they contribute approximately 85% of the carbon dioxide (CO₂) emissions from inland waters [2]. It is estimated that global rivers emit approximately 2.0 Pg C of CO₂ into the atmosphere annually [4], accounting for more than half of the annual carbon sequestration (3.4 Pg) by global terrestrial ecosystems [5]. This amount could be under-estimated because of the high areal flux of urban-dominated rivers [6,7]. In addition, the absence of episodic events, including heavy rainfall, can have a substantial impact on the estimation of riverine CO₂ emissions [8]. To reduce this uncertainty, more research on river carbon emissions conducted in human-dominated areas and under extreme weather events is needed.

The enhanced anthropogenic activities, through urban wastewater discharge and agricultural runoff, have intensified nutrient concentrations in riverine systems, resulting in eutrophication and perturbation of biogeochemical cycles [9]. For example, the input of agricultural fertilizers into riverine systems via surface runoff elevates aqueous nutrient concentrations, thereby potentially enhancing CO₂ production and subsequent emission from aquatic environments [10,11]. Several studies demonstrated that the highest CO₂ emissions were often observed in rivers surrounded by highly urbanized regions [12,13], which should be considered to accurately estimate the CO₂ emissions from human-impacted aquatic ecosystems. Hence, quantifying CO₂ emission flux in urban-agricultural riverine systems is critical for refining total inland water greenhouse gas emission inventories and developing effective mitigation strategies.

In addition to anthropogenic perturbations, episodic hydrometeorological events exert critical control on CO₂ emission patterns. The rainfall events facilitate the transfer of substantial amounts of dissolved organic matter from terrestrial ecosystems to river systems through accelerated surface runoff and enhanced soil leaching processes, consequently promoting microbial CO₂ production in aquatic environments [14]. However, some studies have demonstrated that the combined effects of rainfall dilution and enhanced gas evasion driven by elevated gas transfer velocities can override the stimulatory influence of lateral inputs, resulting in lower river pCO₂ levels during wet seasons compared to dry seasons [15]. And the transient nature of precipitation-induced impacts—typically persisting for 6 to 72 h—introduces significant uncertainties in extrapolating steady-state degassing models [16,17]. To better understand temporal variability and potential “hot-moments” of inland water CO₂ emissions, more process understanding would be required with regard to CO₂ cycling during periods of extreme events like floods for which observations are generally rare [18].

The Chaohu Lake Basin, situated in eastern China, encompasses a network of numerous inflowing rivers and experiences pronounced monsoon influences. The average annual precipitation is more than 1000 mm [19]. Recent years have witnessed accelerated urbanization characterized by expanding urban land use and population growth in municipal areas. Currently, the basin’s population exceeds 10 million inhabitants, with the majority concentrated in the intensively urbanized northern region [13]. This demographic distribution reflects spatial disparities in regional development patterns. The Chaohu Lake Basin rivers, given the presence of multiple waterways traversing highly urbanized areas within the watershed and a climate conducive to intense precipitation events, serve as exemplary representatives of inland waters subject to the combined impacts of heavy rainfall phenomena and anthropogenic activities. Research on these rivers could provide critical references for refining CO₂ budget estimations in fluvial systems across similar regions worldwide.

Therefore, this study selected representative rivers in the Chaohu Lake Basin as research subjects. Field samplings were conducted during dry season, normal season, and post-rainfall conditions to elucidate spatiotemporal variations in pCO₂ and CO₂ emission flux at the water–air interface, with a particular focus on the potential impact of heavy rainfall and its variation among different river types, while simultaneously analyzing physical, chemical, and biological driving factors. This study aims to provide empirical support and theoretical foundations for refining CO₂ budget assessments in fluvial ecosystems within high-intensity anthropogenic activity zones under heavy rainfall scenarios.

2. Materials and Methods

2.1. Study Area and Sampled Riverine Types

Lake Chao, located in Hefei City, Anhui Province, China, is situated in the lower reaches of the Yangtze River. With a surface area of approximately 780 km², it extends 54 km from east to west and 21 km in width from north to south. Characterized by its distinctive bird’s nest-shaped morphology, it ranks as the fifth-largest freshwater lake in China (Figure 1). Lake Chao’s hydrological network comprises over 30 perennial tributaries draining into the lake. Key inflow rivers include the Nanfei River flowing through downtown urban area, the Pai River traversing suburban area [20,21], and the Hangbu, Baishitian, and Zhao Rivers are predominantly located within non-urban zones of the southern basin region. The Nanfei, Hangbu, and Baishitian Rivers collectively contribute over 75% of the total inflow into the lake. Among them, the Nanfei River delivers the highest loads of total nitrogen and total phosphorus, contributing nearly half of the pollutant load with less than one-sixth of the total inflow volume. In contrast, the Hangbu River, accounting for more than half of the total inflow, contributes approximately one-fifth of the pollutant load, serving as a major source of cleaner inflow to Lake Chaohu, which is available from the Chaohu Lake Administration of Anhui Province (https://chglj.hefei.gov.cn/, accessed on 28 January 2026). The catchment is characterized by subtropical monsoon climate, resulting in high rainfall in the wet season (generally from June to September) and much less rainfall in the dry season (generally from December to February) [22].

To comparatively analyze the heterogeneity of greenhouse gas emissions among river systems, we categorized the sampled rivers into two distinct groups: urban rivers (including the Nanfei River and Pai River) and non-urban rivers (comprising the Hangbu River, Baishitian River, and Zhao River). River typology classifications also referenced prior basin-scale studies [13]. Sampling sites for the three non-urban rivers were located at their estuaries to reflect the background water quality and overall emission characteristics of the basin. For the two urban rivers, sampling sites were established not only at the estuaries (downstream) but also along midstream and upstream sections, in order to assess the impact of urban wastewater discharge on water quality and carbon dioxide emissions. The spatial distribution of sampling sites covered both urban and natural landscapes, thereby providing a representative overview of key environmental pressures across the basin.

2.2. Sample Collection and Analysis

This study conducted field sampling at the aforementioned sampling sites in 26 May 2018, 19 January 2019, and 9 May 2019. Specifically, January represented the dry season of the river system, characterized by low discharge, decreased water levels, and reduced flow velocity (in contrast to the wet season with opposing hydrological features, while the normal season exhibits intermediate conditions). May sampling occurred during the normal season, typically displaying moderate hydrological parameters, such as the May 2019 sampling event, which can serve as the initial state. However, a heavy rainfall (16.51 mm/24 h) occurred one day prior to the May 2018 sampling campaign, with cumulative precipitation exceeding 60 mm during the preceding 10-day period (Figure 2). Given that the surface water in the region is primarily derived from precipitation and experiences weak evaporation [23], this rainfall likely altered the river’s hydrological regime (e.g., resulting in increased discharge) and associated environmental conditions. The sampling campaign was designed to account for the pronounced influence of heavy rainfall on pollutant transport. Sampling was therefore conducted within 24 h after rainfall events to reliably capture peak pollution loads associated with runoff, including non-point source pollutants and urban surface discharge.

During sampling, approximately 16 mL of river surface water was injected into 33 mL headspace vials using a syringe for subsequent analysis of the partial pressure of dissolved carbon dioxide (pCO₂) [24,25]. The headspace vials were constructed of brown glass and prepared by first adding 2 g of potassium chloride. The vials were then sealed with butyl rubber septa and secured with aluminum caps. Finally, the vials were evacuated, filled with 5 mL of high-purity nitrogen gas (mole fraction ≥ 99.999%) [25], and the needle puncture sites were sealed with sealing compound to ensure gas-tight integrity. Atmospheric air samples were collected above the water surface and transferred into aluminum foil gas bags for subsequent determination of greenhouse gas concentrations.

This study conducted three field samplings over a one-year period, collecting nine specimens each time, yielding a total of 27 samples. During laboratory analysis of gas samples, the headspace vial is first vigorously shaken for 3 min and then allowed to stand for 30 min to achieve full equilibrium between the gas and liquid phases in the vial [26]. Subsequently, a 5 mL syringe is used to extract the gas from the headspace vial and inject it into a gas chromatograph (7890B, Agilent Technologies, Santa Clara, CA, USA) equipped with a flame ionization detector (FID) for CO₂ detection. The original dissolved concentration of CO₂ in water is determined through the calculation based on CO₂ gas dissolution equilibrium and Henry’s law [27].

In addition, a multi-parameter water quality analyzer (YSI 6000) was used on-site to measure surface water temperature, pH, dissolved oxygen (DO), and other indicators, while water samples were simultaneously collected for the determination of key water environmental parameters. Among these, total nitrogen (TN) was quantified using the alkaline potassium persulfate digestion followed by UV spectrophotometry, and total phosphorus (TP) was measured using the ammonium molybdate spectrophotometric method. Ammonia nitrogen (NH₄⁺-N) and nitrate nitrogen (NO₃⁻-N) were determined using a flow analyzer, while chlorophyll-a (Chl-a) was measured by spectrophotometry following acetone extraction and storage protected from light. Riverine attributes such as width, depth, and mean flow rates at all sampling riverine reaches were obtained by a boat-mounted Acoustic Doppler Current Profiler (ADCP), RiverSurveyor M9 (SonTex, San Diego, CA, USA).

2.3. Data Calculations

The initial dissolved concentration of CO₂ in the water body, denoted as C_w, is calculated as given in Equation (1) [27], while the pCO₂ in the water is computed according to Equation (3).

$$C_w=\left(p_1\times K_e\right)+\left({\displaystyle \frac{V_a}{V_l}}\times {\displaystyle \frac{p_1-p_0}{V_m}}\right)$$

(1)

$$K_e=e^{-58.0931+90.5069\left({\displaystyle \frac{100}{T}}\right)+22.2940\ln \left({\displaystyle \frac{T}{100}}\right)+S\left[0.027766-0.025888\left({\displaystyle \frac{T}{100}}\right)+0.0050578{\left({\displaystyle \frac{T}{100}}\right)}^2\right]}$$

(2)

$$p\left({CO}_2\right)={\displaystyle \frac{C_w}{K_s}}$$

(3)

In the equations, C_w denotes the dissolved CO₂ concentration in the water body (μmol·L⁻¹), p₀ and p₁ represent the partial pressure of CO₂ in high-purity nitrogen and in the gas mixture after equilibration via shaking, respectively (μatm). K_e refers to the solubility of CO₂ gas at equilibrium (mol·L⁻¹·atm⁻¹), and K_s indicates the Henry’s law constant (mol·L⁻¹·atm⁻¹). V_a denotes the gas phase volume in the headspace vial (mL), V_l represents the liquid phase volume in the headspace vial (mL). V_m is the molar volume of CO₂ gas, T indicates the temperature in Kelvin (K), and S refers to the salinity (%).

Based on C_w, the calculation of CO₂ emission flux at the water–air interface refers to the boundary layer model method, with the calculation formula provided in Equation (4) [28].

$$F=k\times \left(C_w-C_{eq}\right)$$

(4)

$$k=0.24\times \left(2841VS+2.02\right){\left(S_c/600\right)}^{3/2}$$

(5)

$$S_c\left({CO}_2\right)=1911.1-118.11\times t+3.4527\times t^2-0.04132\times t^3$$

(6)

In these formulas, F denotes the CO₂ exchange flux (mmol·m⁻²·d⁻¹), C_eq represents the atmospheric CO₂ concentration above the water surface (µmol·L⁻¹), and k is the CO₂ gas exchange rate at the water–air interface (m·d⁻¹) [29]. V represents the stream velocity (m·s⁻¹), S denotes the river slope (dimensionless), and S_c is the Schmidt number, which is determined solely by the water temperature (°C) [28].

2.4. Data Analysis

Prior to statistical analysis, data distribution was assessed using the Shapiro–Wilk test. Variables meeting normality assumptions (p > 0.05) were analyzed using one-way ANOVA with Tukey’s HSD post hoc test. Non-normal variables were log₁₀-transformed prior to ANOVA. Linear regression was performed to assess the relationship between environmental variables and CO₂ to identify trends and quantify the strength and direction of the associations. One-way ANOVA was conducted to determine whether significant differences existed among different groups at the 0.05 significance level. The independent-samples t-test was employed to assess the significance of differences in the metrics between urban and non-urban rivers. All statistical analyses were performed in SPSS (v26) and the figures were created by Origin 2021.

3. Results

3.1. Riverine Physical and Chemical Characteristics

Statistical analysis revealed significant spatiotemporal variations in physicochemical parameters of rivers across the Chaohu Lake Basin (

Table 1. Statistical characteristics of riverine physical and chemical indicators. All of the data were shown as Mean ± SD. Different lowercase letters denote significant temporal differences (p < 0.05) among rivers of the same type.

Time	River Types	Discharge m3 s−1	Tw °C	DO mg L−1	pH	NO3-N mg L−1	NH4-N mg L−1	PO4-P mg L−1	Chl-a μg L−1
	Urban	18.36 ± 9.59 a	23.55 ± 0.45 a	10.06 ± 10.92 a	7.69 ± 0.30 ab	0.51 ± 0.25 b	3.72 ± 2.45 a	0.25 ± 0.21 a	14.95 ± 6.61 a
Heavy Rainfall	Non-urban	122.79 ± 90.52 a	23.97 ± 0.58 a	18.86 ± 1.72 a	7.94 ± 0.03 ab	0.52 ± 0.09 b	0.70 ± 0.11 a	0.04 ± 0.01 a	8.50 ± 6.85 a
	Total	53.17 ± 69.51 a	23.69 ± 0.50 a	12.99 ± 9.73 a	7.78 ± 0.27 ab	0.51 ± 0.21 b	2.72 ± 2.46 a	0.18 ± 0.20 a	12.80 ± 7.03 a
	Urban	10.44 ± 11.16 ab	8.95 ± 1.28 b	7.52 ± 2.71 a	7.50 ± 0.26 b	2.71 ± 0.83 a	3.70 ± 2.03 a	0.14 ± 0.08 a	2.38 ± 0.97 b
Dry Season	Non-urban	6.30 ± 7.57 a	5.50 ± 0.26 b	10.70 ± 0.65 b	7.67 ± 0.48 b	1.89 ± 0.58 a	0.33 ± 0.21 b	0.05 ± 0.04 a	3.76 ± 3.95 a
	Total	9.06 ± 9.82 a	7.80 ± 2.00 b	8.58 ± 2.69 a	7.56 ± 0.33 b	2.44 ± 0.83 a	2.58 ± 2.33 a	0.11 ± 0.08 a	2.84 ± 2.23 b
	Urban	4.19 ± 3.71 b	22.85 ± 0.61 a	7.62 ± 2.23 a	7.87 ± 0.25 a	2.70 ± 1.12 a	4.05 ± 2.06 a	0.18 ± 0.19 a	22.56 ± 8.75 a
Normal Season	Non-urban	7.86 ± 9.03 a	22.86 ± 1.44 a	8.91 ± 2.33 b	8.34 ± 0.19 a	0.87 ± 0.25 b	0.38 ± 0.19 ab	0.01 ± 0.00 a	24.54 ± 18.15 a
	Total	5.41 ± 5.69 a	22.85 ± 0.87 a	8.05 ± 2.21 a	8.03 ± 0.33 a	2.09 ± 1.28 a	2.83 ± 2.45 a	0.12 ± 0.17 a	23.22 ± 11.46 a

). For instance, urban river discharge following the heavy rainfall event reached 18.36 m³ s⁻¹, which was significantly higher (p < 0.05) than the corresponding period in the following year (4.19 m³ s⁻¹). Notably, non-urban rivers exhibited a markedly elevated discharge of 122.79 m³ s⁻¹ under similar post-rainfall conditions. In addition, DO levels in non-urban rivers reached 18.86 mg L⁻¹ following the heavy rainfall, significantly higher (p < 0.05) than those during dry (10.70 mg L⁻¹) and normal water periods (8.91 mg L⁻¹). pH during dry season averaged 7.56, significantly lower (p < 0.05) than the 8.03 observed in normal season. NO₃-N concentrations post-intensive rainfall (0.51 mg L⁻¹) were markedly reduced compared to dry (2.44 mg L⁻¹) and normal water periods (2.09 mg L⁻¹). However, NH₄-N levels following heavy rainfall (2.72 mg L⁻¹) showed no statistically significant differences (p > 0.05) from dry (2.58 mg L⁻¹) or normal water periods (2.83 mg L⁻¹).

On the other hand, distinct differences were observed between urban and non-urban rivers. During the three sampling periods, non-urban rivers exhibited higher DO and pH levels. For instance, the DO values in non-urban rivers measured 18.86, 10.70, and 8.91 mg L⁻¹, respectively, while those in urban rivers were lower at 10.06, 7.52, and 7.62 mg L⁻¹. Conversely, urban rivers demonstrated significantly elevated nutrient concentrations. Specifically, NH₄-N levels in urban rivers exceeded those in non-urban rivers by more than fivefold. For instance, during the dry season, urban river concentrations reached 3.70 mg L⁻¹, compared to merely 0.33 mg L⁻¹ in non-urban counterparts. Similarly, PO₄-P concentrations in urban rivers were measured to be multiple times (ranging up to over tenfold) higher than those observed in non-urban systems.

It is noteworthy that the TN concentrations in urban rivers during the three sampling periods were 6.51, 6.84, and 8.94 mg L⁻¹, respectively (Figure 3), showing no significant temporal variations (p > 0.05). Similarly, non-urban rivers exhibited TN concentrations of 2.50, 2.56, and 2.94 mg L⁻¹ with no statistically significant temporal differences (p > 0.05). However, urban rivers demonstrated TN levels over twice those of non-urban counterparts, revealing pronounced spatial disparities. A comparable spatial differentiation pattern was also observed for TP concentrations between urban and non-urban river systems. Conversely, Chl-a concentrations in the dry season (2.84 μg L⁻¹) were significantly lower (p < 0.05) than post-rainfall (12.80 μg L⁻¹) and the normal season (23.22 μg L⁻¹), while water temperature exhibited analogous seasonal trends. The heavy rainfall event month registered 164 mm of monthly precipitation, representing a twofold increase compared to the corresponding month in the subsequent year (48 mm). Notably, the adjacent two-month (April and June) precipitation remained below 100 mm. All the above comparisons suggest a markedly elevated precipitation level for the month in which the heavy rainfall occurred.

3.2. Variability of pCO₂ Between Urban and Non-Urban Rivers

At the aggregate level, urban and non-urban river waters exhibited statistically significant differences in pCO₂ levels (p < 0.05, Figure 4a). Specifically, urban river waters had a mean pCO₂ of 3949 μatm, which is 2.8-fold higher than that of non-urban river waters (1423 μatm). Notably, urban rivers exhibited substantial internal pCO₂ variability, with values ranging from 1203 μatm to 7498 μatm, whereas non-urban rivers demonstrated smaller intra-system differences.

The analysis of pCO₂ in urban and non-urban rivers during three sampling periods revealed that urban rivers exhibited pCO₂ levels of 4576 μatm, 3726 μatm, and 3545 μatm during post-heavy rainfall, dry season, and normal water periods, respectively (Figure 4b). The heavy rainfall event elevated aquatic pCO₂ by 20–30% compared to normal conditions, though no significant differences were observed among these hydrological phases (p > 0.05). In contrast, non-urban rivers displayed pCO₂ of 2461 μatm, 1096 μatm, and 712 μatm during post-heavy rainfall, dry season, and normal water periods, respectively. The heavy rainfall event resulted in pCO₂ levels that were significantly higher (2.2-fold and 3.5-fold, respectively) compared to those observed during the dry season and normal water periods (p < 0.05). Additionally, during both the dry season and normal water periods, urban rivers consistently exhibited significantly higher pCO₂ levels than their non-urban counterparts (p < 0.05), aligning with the overall trend of urbanization-driven carbon dynamics.

A comparative analysis of pCO₂ in water at the estuaries of five rivers discharging into the lake revealed that urban rivers exhibited considerably higher levels than non-urban rivers (

Table 2. pCO₂ of different rivers during three sampling periods.

	River	Dry Season/ (μatm)	Normal Season/ (μatm)	Heavy Rainfall/ (μatm)	Mean/ (μatm)
	Nanfei (upper)	6260	3749	7118	5709
	Nanfei (middle)	4540	5573	7498	5870
	Nanfei (lower)	3504	3441	4650	3865
Urban Rivers	Pai (upper)	1203	2037	2359	1866
	Pai (middle)	4000	3977	2609	3529
	Pai (lower)	2848	2489	3221	2853
	Mean	3726	3545	4576	3949
	Hangbu	1201	358	2009	1189
Non-urban	Baishitian	1300	114	2582	1332
Rivers	Zhao	785	1664	2793	1747
	Mean	1096	712	2461	1423
All	Mean	2849	2600	3871	3107

). Among the non-urban rivers, the Hangbu River—despite receiving inflow from a basin with minor urban/peri-urban land use—exhibited pCO₂ levels comparable to the other non-urban rivers (e.g., Baishitian and Zhao Rivers), further supporting its classification as a predominantly non-urban system. Specifically, the urban Nanfei and Pai Rivers exhibited pCO₂ values of 3865 μatm and 2853 μatm, respectively, whereas the highest value among non-urban rivers was only 1747 μatm. These findings indicate that urban rivers are associated with elevated aquatic CO₂ concentrations. In addition, a comparison of pCO₂ levels along different reaches of the two urban rivers revealed that the middle and upper sections of the Nanfei River exhibited higher levels, with a mean value exceeding 5500 μatm, whereas the downstream section showed a lower mean value of less than 4000 μatm. This decrease in the downstream reach may be attributed to its greater distance from the urban core. In contrast, the Pai River demonstrated elevated CO₂ levels specifically in its midstream section, likely due to the location of a county town within this segment.

3.3. Influences of Environmental Factors

Figure 5a demonstrates a significantly positive correlation between river discharge and pCO₂ in urban rivers (p < 0.05, R² = 0.22). In contrast, no significant relationship was observed between discharge and pCO₂ in non-urban rivers (p > 0.05). Figure 5b reveals that Chl-a exhibited no statistically significant association with pCO₂ across the entire dataset (p > 0.05), nor when urban and non-urban rivers were analyzed separately. The correlation between water temperature and pCO₂ is not significant (p > 0.05), whether considered as a whole or in parts (Figure 5c). These results suggest that hydrological drivers (e.g., discharge) may play a more pronounced regulatory role in pCO₂ dynamics within urban river systems compared to biological factors like Chl-a.

DO and pH are recognized as critical regulators of aquatic CO₂ dynamics. As illustrated in Figure 6, our results revealed a significant negative correlation between DO and pCO₂ (R² = 0.20, p < 0.05; Figure 6a), while pH demonstrated a much stronger inverse relationship with pCO₂ (R² = 0.35, p < 0.01; Figure 6b). The pH parameter explained a greater proportion of the observed CO₂ variation compared to DO, as evidenced by its higher regression coefficient. These findings collectively indicate that elevated DO and pH levels in water systems correspond to decreased CO₂ concentrations. Notably, the dominant explanatory power of pH suggests its pivotal role in modulating CO₂ evasion processes within the studied aquatic environments.

The generation of CO₂ in fluvial ecosystems constitutes a dynamic and intricately regulated process, wherein nutrient loading frequently emerges as a pivotal driver. The correlations between pCO₂ and nutrients in river water of the Chaohu Lake Basin are illustrated in Figure 7. The results demonstrate that TN, TP, and NH₄⁺-N concentrations exhibit highly significant positive correlations with pCO₂ (p < 0.01), accounting for 61%, 57%, and 61% of the pCO₂ variability, respectively, thereby indicating robust explanatory capacity. The results above indicate that with increasing nutrient load levels, the pCO₂ in river waters of the Chaohu Lake Basin exhibits a significant upward trend.

3.4. CO₂ Emission Fluxes

Based on data such as aqueous CO₂ concentration and combined with a water–air interface diffusion model, the CO₂ exchange flux at the water–air interface of rivers in the Chaohu Lake Basin was estimated. The results indicate that the CO₂ exchange flux at the water–air interface of rivers in the Chaohu Lake Basin is positive (297.84 mmol·m⁻²·d⁻¹), representing a “net source” of CO₂ emissions (

Table 3. CO₂ fluxes at the water–air interface of different rivers during three sampling periods (The positive value represents the source, while the negative value represents the sink).

	River	Dry Season/ (mmol·m−2·d−1)	Normal Season/ (mmol·m−2·d−1)	Heavy Rainfall/ (mmol·m−2·d−1)	Mean/ (mmol·m−2·d−1)
	Nanfei (upper)	374.05	234.05	695.13	434.41
	Nanfei (middle)	316.04	199.45	561.10	358.86
	Nanfei (lower)	179.36	58.72	351.98	196.69
Urban Rivers	Pai (upper)	68.95	52.49	598.93	240.12
	Pai (middle)	64.02	582.44	1175.76	607.41
	Pai (lower)	87.51	77.03	355.85	173.46
	Mean	181.65	200.70	623.13	335.16
	Hangbu	14.10	−3.30	586.66	199.15
Non-urban	Baishitian	39.15	−9.87	308.11	112.46
Rivers	Zhao	11.52	26.80	1035.68	358.00
	Mean	21.59	4.54	643.48	223.20
All	Mean	128.30	135.31	629.91	297.84

). The CO₂ exchange flux was higher in urban rivers (335.16 mmol·m⁻²·d⁻¹) than in non-urban rivers (223.20 mmol·m⁻²·d⁻¹). Specifically, the urban river (Nanfei River) exhibited a gradually decreasing trend in CO₂ flux from upstream to downstream, while the suburban river (Pai River) showed an initial increase followed by a decrease along the flow direction. The flux in the midstream section of Pai River reached 607.41 mmol·m⁻²·d⁻¹, approximately three times higher than that observed in its upstream and downstream sections. These patterns demonstrate significant spatial variations in CO₂ emissions at the water–air interface across river systems within the Chaohu Lake Basin. A comparison of the riverine CO₂ fluxes across the three sampling periods revealed that, overall, the flux after heavy rainfall reached 629.91 mmol·m⁻²·d⁻¹, which was substantially higher than those during the dry season (128.30 mmol·m⁻²·d⁻¹) and the normal season (135.31 mmol·m⁻²·d⁻¹). It is noteworthy that the flux from non-urban rivers after heavy rainfall rose to 643.48 mmol·m⁻²·d⁻¹, a level comparable to that of urban rivers and significantly higher than during regular conditions.

4. Discussion

4.1. Roles of Riverine Physicochemical Factors

In our study, TN, TP, and NH₄⁺-N concentrations exhibit significant positive correlations with pCO₂ (p < 0.01), accounting for 61%, 57%, and 61% of the pCO₂ variability, respectively. Nutrient inputs impact the balance of primary production and respiration in rivers, thus affecting CO₂ consumption and production [10,30]. For example, NH₄⁺ influences the primary productivity of aquatic plants and plankton metabolism, indirectly regulating CO₂ concentration [31]. In this study, urban rivers are characterized by high nutrient loads and low Chl-a concentrations (Table 1). The high nutrient loads indicate a high potential for CO₂ production, while the low Chl-a concentrations reflect a reduced potential for CO₂ consumption. The collective effect of these factors may lead to elevated CO₂ emissions in urban rivers [3]. In addition, our results revealed a significant negative correlation between DO, pH and pCO₂ (Figure 6). pH is an important factor regulating the aquatic carbonate equilibrium, while DO serves as an indicative parameter for biological metabolism in water [32]. From inland rivers in China, there was a significantly negative exponential relationship between pH and fCO₂ and a significantly negative correlation between DO and fCO₂ [33]. In the present study, urban rivers exhibited lower pH and DO levels compared to non-urban rivers, which may explain their higher CO₂ concentrations. Notably, pCO₂ and CO₂ fluxes in urban rivers were very high during both the dry and normal seasons, whereas those in non-urban rivers were very low, particularly in the normal season. For example, during the normal season, pCO₂ and CO₂ fluxes in the Baishitian river were 114 μatm and −9.87 mmol·m⁻²·d⁻¹, respectively. Long-term measurement showed point source organic pollution appeared to be the main driver of pCO₂ in the lower Seine River, downstream of the main outlet of waste water treatment plant effluents, and whatever the period studied [34]. The continuous discharge of domestic wastewater keeps pollutant concentrations in urban rivers consistently high, thereby reducing the amplitude of CO₂ fluctuations over time (Table 1). Although quantitative watershed-scale data are not available, it is well established that urban catchments with high population density generate disproportionately large wastewater loads compared to agricultural catchments [13], which likely explains the persistently elevated nutrient and CO₂ levels observed in urban rivers across all seasons. The reason for higher CO₂ levels in non-urban rivers during dry season may be attributed to a longer water residence time and lower currents, both contributing to the accumulation of gases in the water column [10]. The overall results suggest that neglecting localized pollution impacts on GHG emissions from increasingly urbanized river basins may result in a substantial underestimation of global riverine GHG emissions [35].

4.2. Impact of Heavy Rainfall Events on Riverine CO₂ Emissions

Flood events fundamentally alter riverine carbon cycling, rapidly mobilizing terrestrial carbon stores and disrupting baseflow dynamics [36]. In this study, the heavy rainfall event elevated aquatic pCO₂ by 20–30% compared to normal conditions in urban rivers, and 120–250% in non-urban rivers. Floodplain connectivity and overland flow introduce large quantities of dissolved organic carbon (DOC), dissolved inorganic carbon (DIC), and particulate organic carbon (POC) derived from soils and riparian zones into the river channel. The decomposition of this active carbon rapidly produces CO₂, particularly within anoxic microsites in the water column or resuspended sediments [17,37].

Rainstorm is a key driver of carbon emissions from inland waters [14]. Based on aqueous CO₂ concentration data and combined with a water–air interface diffusion model, following the heavy rainfall event, the riverine CO₂ emission flux in the Chaohu Lake Basin surged to 629.91 mmol·m⁻²·d⁻¹, representing a fivefold increase compared to baseline conditions (Table 3). The diffusion model using Equation (5) captures this, as increased river velocity (V) and slope (S) during floods generate intense turbulence at the water–air interface, dramatically enhancing the k value. In addition, pCO₂ rises substantially after the rainfall event (Table 2), resulting in a large concentration gradient [8]. In this study, the rainfall event resulted in elevated levels of both the k value and dissolved CO₂. Their combined effect led to a higher riverine CO₂ emission flux (Equation (4)).

4.3. A Comprehensive Assessment of Riverine CO₂ Emissions in the Chaohu Lake Basin

The mean riverine CO₂ emission flux at the water–air interface in this study was 297.84 mmol·m⁻²·d⁻¹, with positive flux values indicating that the river acts as a source of atmospheric CO₂. In comparison with other studies, the riverine CO₂ emission flux in this study was 53 times higher than that of Lake Chaohu (5.61 mmol·m⁻²·d⁻¹) [38], 17 times higher than that of rivers in Nanjing city in the areas of the Yangtze River delta of China (17.63 mmol·m⁻²·d⁻¹) [39] and twice that of the CO₂ flux from the source region of the Yellow River (143.9 mmol·m⁻²·d⁻¹) [40], indicating that rivers in the Chaohu Lake Basin represent a significant source of CO₂ emissions. It is noteworthy that CO₂ emissions exhibited significant differences between the two river types. The average flux in urban rivers was 335.16 mmol·m⁻²·d⁻¹, which is 50% higher than that in non-urban rivers. When comparing flux data only under normal conditions (excluding the heavy rainfall event, as non-urban river values becomes exceptionally high following such precipitation), the flux in urban rivers was more than 10 times greater than that in non-urban rivers, indicating that urban rivers are hotspots for CO₂ emissions. This finding is consistent with the conclusions of another study in the same region [13]. Similarly to our study, riverine CO₂ emission rates in the Chongqing metropolitan area were three to six times higher than in rural or forested areas [41]. Furthermore, linear regression analysis revealed that annual CO₂ concentration and flux were significantly correlated with the proportion of built-up area in the mainstem of Liao River [42], and the proportion of built-up area explained 49% of the variation in CO₂ flux. Therefore, increased attention should be given to the carbon emissions from urban rivers within the Lake Chaohu Basin in future studies.

This comprehensive assessment reveals that CO₂ fluxes in urban rivers of the Lake Chaohu Basin were significantly higher than those in non-urban rivers during dry and normal water periods, with values of 181.65 and 200.70 mmol·m⁻²·d⁻¹ versus 21.59 and 4.54 mmol·m⁻²·d⁻¹, respectively. However, following the heavy rainfall event, both river types exhibited a substantial increase in CO₂ emissions, reaching more than 600 mmol·m⁻²·d⁻¹. Notably, non-urban rivers demonstrated a dramatic amplification—by tens to hundreds of times—compared to their baseline fluxes, highlighting the extreme sensitivity of these systems to precipitation events. Therefore, enhancing the monitoring of carbon emissions from non-urban rivers under heavy rainfall conditions is needed. Current carbon cycle models that inadequately account for the pulse-release of carbon during heavy rainfall events likely underestimate emissions from inland waters. This study reveals a critical positive feedback mechanism: climate change-induced intensification of heavy rainfall may amplify fluvial carbon emissions, which in turn accelerates climate change. In the context of global climate change contributing to an increased frequency of extreme precipitation events, future efforts should prioritize the conservation of riparian vegetation and wetlands, as they function as effective “carbon buffer zones” that intercept and sequester terrestrial carbon, thereby mitigating dramatic fluctuations in riverine carbon concentrations following rainfall events. These findings underscore the critical role of hydrological disturbances in regulating riverine CO₂ emissions and emphasize the need to incorporate event-driven dynamics into carbon budgeting models for these human-impacted regions.

4.4. Limitations of the Study

As with most research, the present study also has several limitations that should be acknowledged. Firstly, we acknowledge that topographic differences exist between our studied rivers, with urban systems located on east-oriented slopes and non-urban rivers on north-oriented slopes. While slope aspect and angle may modulate local runoff generation, our study suggests that land use and associated carbon source–sink dynamics override potential orographic effects in this relatively low-relief basin. Future studies with denser topographic sampling could further disentangle these controls. Secondly, the magnitude of the increase in riverine CO₂ emissions attributable to heavy rainfall may subject to uncertainty in our study, primarily because we conducted only one sampling event following the heavy rainfall and compared it with two other sampling datasets. These two datasets were collected several months after the rainfall event and may not adequately represent the pre-rainfall conditions. However, the minimal differences in both pCO₂ and emission flux between the two sampling events form a substantial contrast with the conditions following the heavy rainfall, sufficiently demonstrating that “heavy rainfall enhanced riverine CO₂ emissions to some extent”. To accurately estimate the contribution of a rainfall event to greenhouse gas emissions, it is necessary to measure and consider the initial emission levels beforehand. For example, temporal changes can be captured by collecting GHG emission data at regular intervals before, during, and after rainfall events. This approach ensures a clear understanding of how rainfall influences GHG fluxes.

5. Conclusions

Rivers in the Chaohu Lake Basin overall serve as a net source of greenhouse gas CO₂ emissions (297.84 mmol·m⁻²·d⁻¹), with urban rivers (335.16 mmol·m⁻²·d⁻¹) exhibiting stronger CO₂ emissions compared to non-urban rivers (223.20 mmol·m⁻²·d⁻¹). Notably, CO₂ emissions in non-urban rivers were minimal during the normal season (4.54 mmol·m⁻²·d⁻¹), contrasting with a slightly higher value of 21.59 mmol·m⁻²·d⁻¹ observed in the dry season. Key environmental parameters in water bodies significantly influence riverine pCO₂. Among these, discharge, TN, TP, and NH₄⁺-N showed significantly positive correlations with pCO₂ (p < 0.05), while DO and pH exhibited significantly negative correlations with pCO₂ (p < 0.05). Relatively large increases in riverine CO₂ emissions were observed following the heavy rainfall event, though the extent of enhancement varied across river types: urban rivers exhibited an increase by severalfold, while non-urban rivers even experienced a rise of tens of times. Future research should prioritize monitoring riverine CO₂ emissions under heavy rainfall conditions, particularly in non-urban river systems.