| 01 Sep 2026
Effects of atmospheric turbulence on vertical dispersion of carbon dioxide and methane over subtropical paddy fields
Abstract. Paddy fields serve as a critical methane (CH4) source and carbon dioxide (CO2) sink, playing an essential role in terrestrial carbon cycling and greenhouse gas budgets. The effects of atmospheric turbulence on the vertical dispersion of CO2 and CH4 were investigated in this study, using high-frequency observations over subtropical paddy fields. The pronounced diurnal variations in surface concentrations and turbulent fluxes of both gases suggest the dominant control of boundary-layer turbulent diffusion. Focusing on nocturnal stable conditions, mutual information was introduced to characterize relevant nonlinear processes. Turbulence over the paddy fields becomes intermittent when the nighttime wind speed falls below 1.5 m s−1. This weak and intermittent turbulence generally leads to the accumulation of CO2 and CH4 near the surface, whereas short turbulent bursts promote rapid dispersion, showing strong non-stationarity. Under intermittent turbulence, interference from submeso motions renders similarity relationships inapplicable for describing the transport of momentum, heat, as well as passive scalars (CO2 and CH4). This limitation can be effectively mitigated by removing submeso components from original observations. The distinct spectral gap between submeso and turbulent scales indicates the breakdown of energy cascade. As submeso energy is confined to its original scale, the energy supply for small-scale turbulence is limited and turbulence becomes more stochastic. This study reveals the effects of intermittent turbulence on the vertical diffusion of CO2 and CH4 under nocturnal stable conditions, but also highlights the challenges in parameterizing such stable-boundary-layer processes. This provides new insights for understanding carbon cycling over terrestrial ecosystems.
Based on a high-frequency observation system, this paper analyzes the effects of nocturnal intermittent turbulence on the vertical transport of CO₂ and CH₄ in rice paddies. The research questions are clearly defined, the analytical framework is relatively comprehensive, and the main observational conclusions are reasonably persuasive. The research focus of the paper aligns with the requirements of the ACP journal. However, there are still issues that require further clarification regarding data screening, the validation of passive scalar similarity relations, and the explanation of energy cascade mechanisms.
Major comments
The focus of this study is on nighttime conditions with light winds and intermittent turbulence; however, the description of data quality control and the final sample selection process in Section 2.2 is currently insufficient. The authors primarily describe routine preprocessing steps such as spike removal, coordinate rotation, linear detrending, and density correction, while the screening criteria actually used for subsequent analysis mainly include stability and CO₂ and CH₄ flux thresholds. The authors need to further explain the rationale for selecting these thresholds and report the number of raw 30-minute samples, the rejection rates at each screening step, and the final number of samples retained under continuous and intermittent turbulence conditions. In particular, they need to clarify whether the u* and flux thresholds might preferentially exclude data under conditions of weak turbulence and weak transport, as these conditions are precisely the focus of this study. We recommend that the authors conduct necessary sensitivity analyses to demonstrate that results such as those in Figs. 8–9 are not significantly affected by the sample screening criteria.
Major 2. Insufficient direct evidence for the “breakdown–recovery” of the CO₂ and CH₄ similarity relationship
Fig. 8 clearly shows that, under steady-state conditions, continuous turbulence generally conforms to the MOST similarity relationship, whereas intermittent turbulence deviates significantly; after removing the submesoscale components, the reconstructed wind speed and potential temperature values once again approach the similarity relationship. However, Fig. 9 presents results for water vapor, CO₂, and CH₄ only under continuous turbulence conditions; the reconstructed results for intermittent turbulence and after removing the submesoscale components are not provided. Potential temperature exhibits scaling behavior similar to that of passive scalar variables, which provides a physical basis for this inference; however, it does not directly prove that CO₂ and CH₄ also experience the same degree of similarity breakdown under intermittent turbulence conditions and recover after the removal of mesoscale motion. Given that CO₂ and CH₄ are the core subjects of this study and that this conclusion occupies a prominent position in the abstract and conclusions, it is recommended that the authors at least directly present the scaling relationships for CO₂ and CH₄ under the three conditions—continuous, intermittent, and reconstructed—to provide direct empirical support for this core conclusion.
Major 3. Figs. 10–11 are insufficient to support the mechanistic conclusion of “energy cascade breakdown”
Fig. 10 clearly shows that under intermittent turbulence conditions, the low-frequency mesoscale components are enhanced, the high-frequency turbulent components are weakened, and a distinct spectral gap appears between the submesoscale and turbulent scales. These results support the existence of distinct scale separation and the suppression of small-scale turbulent activity under intermittent turbulence conditions; however, the spectral gap itself does not directly prove a “breakdown” in energy cascade from large to small scales, as the energy spectrum primarily reflects the distribution of energy across different scales rather than directly measuring energy transfer between scales. We recommend that the authors refine their description.
The authors further attempt to support this mechanistic explanation using the incremental self-mutual information in Fig. 11, but there is a key issue here that requires clarification. The authors calculated the information required between the current state X(t) and the increment:
ΔX(t) = X(t+τ) − X(t)
rather than the mutual information between the current state X(t) and the future state X(t+τ). If the research objective is to quantify the information retention or “memory” of turbulence over time, the latter has a more direct physical significance. The authors are asked to clarify this point.
Minor comments
Minor 1. It is recommended to supplement information on the station’s surrounding environment and source areas
It is recommended to add a schematic diagram of the land surface or land use around the observation station in Section 2.1, labeling major potential local sources and prevailing wind directions. If conditions permit, a simple EC footprint analysis could be provided to illustrate the range of the land surface represented by the flux observations under major meteorological conditions. This is particularly meaningful for this paper because the authors discuss both local rice paddy CO₂/CH₄ sources and sinks and use wind direction to explain regional CO₂ and PM₂.₅ transport; therefore, a clearer distinction between the local/sub-local footprint and regional advection effects is needed.
Minor 2. Suggest adding simple quantitative statistics on meteorological differences across the three stages
The authors note that nighttime wind speeds in Stage 2 were generally higher than in Stage 1, particularly on multiple occasions between August 13 and 18 when they exceeded 1.5 m/s; however, this difference is not readily apparent from the dense time series in Fig. 1. We recommend reporting the mean or median of nighttime wind speeds across different stages, the degree of dispersion, and the proportion of samples with wind speeds > 1.5 m/s to more directly support the inter-stage differences. Additionally, given that wind direction is used later in the text to discuss regional transport, we suggest including a wind rose diagram or other more intuitive wind direction statistics.
Minor 3. Explanation of the method used to determine the 1.5 m/s transition wind speed in Fig. 4
The HOST relationship in Fig. 4 indicates a wind speed threshold of approximately 1.5 m/s for the transition between strong and weak winds, and this threshold is generally consistent with the intermittent/continuous turbulence classification at LIST = 0.75. This result is reasonable in itself, and the authors compare it in the text with the 2.5 m/s threshold obtained by Zhang et al. (2004) over a desert surface. It is recommended that the authors explicitly state how the 1.5 m/s velocity inflection point was derived—whether through statistical optimization or visual judgment—to enhance the method’s reproducibility.