Medium-Term Monitoring of Greenhouse Gases above Rice-Wheat Rotation System Based on Mid-Infrared Laser Heterodyne Radiometer

Abstract

The rice-wheat rotation system is a major agricultural practice in China as well as an important source of greenhouse gas (GHG) emissions. In this study, the developed mid-infrared laser heterodyne radiometer (MIR-LHR) was used for the remote sensing of atmospheric CH₄ and N₂O concentrations above the rice-wheat rotation system. From April 2019 to May 2022, the atmospheric column concentrations of CH₄ and N₂O above the rice-wheat rotation system were continuously observed in Hefei, China. The peak values of the N₂O column concentration appeared 7~10 days after wheat seasonal fertilization, with additional peaks during the drainage period of rice cultivation. During the three-year rice-wheat crop rotation cycle, a consistent trend was observed in the CH₄ column concentrations, which increased during the rice-growing season and subsequently decreased during the wheat-growing season. The data reveal different seasonal patterns and the impact of agricultural activities on their emissions. During the observation period, the fluctuations in the CH₄ and N₂O column concentrations associated with the rice-wheat rotation system were about 40 ppbv and 6 ppbv, respectively. The MIR-LHR developed for this study shows great potential for analyzing fluctuations in atmospheric column concentrations caused by GHG emissions in the rice-wheat rotation system.

1. Introduction

The sustainability and uncertainty of global warming have a profound impact on the sustainable development of human society. The continuous increase in the temperature and atmospheric carbon dioxide (CO₂) concentration is one of the most indicative characteristics of global warming. Addressing the environmental risks posed by climate warming is a major global challenge [1]. In October 2018, the United Nations Intergovernmental Panel on Climate Change (IPCC) issued a “Special Report on Global Warming of 1.5 °C”. In the report, scientists emphasized the need for countries to take concrete actions to mitigate climate warming and limit the temperature increase to below 1.5 °C [2,3]. Missing the 1.5 °C target could lead to further warming, with a 2.0 °C increase by 2040 [4]. The long-term excessive anthropogenic emissions of GHGs, exceeding natural capacity, triggering the GHG effect and near-surface temperature rise, are currently recognized as the primary cause of global warming [5]. The farmland ecosystem is the major source of GHG emissions, with agricultural emissions accounting for 10–12% of global anthropogenic emissions. At the same time, farmland soils are also significant sinks for GHGs [6,7,8].

The farmland ecosystem plays an important role in the study of GHG and global change. Global research on GHGs in farmland ecosystems began in the 1970s, and since then, the production and emission processes of CO₂, methane (CH₄), and nitrous oxide (N₂O) have been extensively studied. Studies have shown that 70–90% of atmospheric CH₄ and N₂O originate from surface biological sources. The anaerobic decomposition in global rice fields contributes about 6% to the total atmospheric CH₄ annually. It is generally recognized that the increase in N₂O emissions from farmland is one of the main reasons for the continuous increase in the atmospheric N₂O concentration in recent decades. It is estimated that N₂O emissions resulting from fertilization in farmland ecosystems account for about 13~24% of the total atmospheric N₂O emissions [9,10,11]. Global paddy fields contribute approximately 5–19% of the total annual CH₄ emissions, while N₂O emissions from these fields account for 6–10% of the global warming potential [12]. In southeast China, rice-wheat rotation is the predominant cropping system. This system is critical to national food security, as it accounts for nearly 30% of China’s total grain production. The middle and lower reaches of the Yangtze River, benefiting from favorable water and temperature conditions, form one of China’s key grain-producing regions and represent a transitional zone between humid and semi-humid regions [13,14]. Winter wheat is sown during the cold season, maximizing land use through crop rotation. More than half of China’s total rice-growing area adopts the rice-wheat rotation system, which includes both irrigated paddy fields and rain-fed wheat fields. The heterogeneity of these landscapes poses challenges in accurately quantifying GHG emissions across the entire rotation cycle [15,16].

GHG emissions in agriculture are usually measured using static chamber gas chromatography, which has long measurement intervals and lacks real-time continuous monitoring capabilities [17]. The eddy covariance technique measures emission fluxes by analyzing the covariance of the temperature, gas concentration, and vertical wind speed, offering broad applicability and high reliability [18]. Gu et al. developed a sensor using this technique to measure CO₂ and H₂O fluxes at the Jiangdu Agricultural Monitoring Station in Jiangsu Province [19].

However, emission flux data collected from local field areas cannot be reliably extrapolated to larger regional or global scales. The high-precision observation of regional atmospheric GHG concentrations offers a more effective approach for calculating the emission of large-scale rice fields and understanding their spatial and temporal distribution [20]. Recently, the laser heterodyne radiometer (LHR) system has made significant progress, which is largely due to the evolution of cost-effective laser technology, waveguides, and fiber optics, as well as the diffusion of their respective applications [21,22]. Weidmann et al. and Xue et al. developed mid-infrared (MIR) LHRs using an inter-band cascade laser (ICL) as the Local Oscillator (LO) for the measurements of atmospheric column abundances of CH₄, N₂O, and H₂O [23,24,25,26].

This study aims to make up for the gaps in the long-term monitoring and analysis of GHG emissions from rice-wheat rotation systems by employing the most advanced remote sensing technology. The MIR-LHR offers a novel approach for the accurate and continuous measurement of atmospheric CH₄ and N₂O concentrations. The developed MIR-LHR has been used for continuous observation, in conjunction with local agricultural activities, to analyze the concentrations of CH₄ and N₂O in the atmosphere during the rice-wheat rotations. The insights gained from this research are anticipated to advance the integration of the MIR-LHR with agricultural flux measurement devices, thereby improving our understanding of how agricultural fluxes impact atmospheric GHG column concentrations. These findings will aid policymakers, agricultural practitioners, and researchers in formulating strategies to reduce the environmental footprint of agricultural activities and promote sustainable agricultural practices.

2. Materials and Methods

2.1. Experimental Details

The MIR-LHR developed in this work is schematically shown in Figure 1. A distributed feedback inter-band cascade laser (DFB-ICL) operating at ambient temperature is served as the LO. The ICL laser can be continuously tuned to detect N₂O in the atmosphere, with a spectral range from 2535 cm⁻¹ to 2543 cm⁻¹ and an output power of more than 15 mW. The unique ICL, with an adjustable spectral range from 2828 cm⁻¹ to 2837 cm⁻¹, was designated to monitor atmospheric CH₄. The system is equipped with a high-precision solar tracker for capturing sunlight that encapsulates atmospheric absorption signatures. The photoelectric tracking mechanism is coupled with a proportional-integral-derivative (PID) control system to track the solar motion, which ensures the tracking accuracy of 7 arc seconds.

The solar radiation captured by the solar tracker is modulated via a mechanical chopper, with the modulation frequency and duty cycle being tuned by adjusting the driving current. During the study period, the driving current was set at a constant 50%. A dichroic beam splitter was utilized to superimpose the modulated solar radiation onto the LO beam emanating from the ICL. The mixing beams traversed an optical band-pass filter with a full width at half maximum (FWHM) of 100 nm, and then converged onto the photomixer D1. The ICL beam deflected by the beam splitter was allocated for frequency calibration purposes.

Subsequently, an accurate relative frequency measurement was performed using a germanium etalon with a free spectral range of about 0.0246 cm⁻¹ and a photomixer D2. The intermediate frequency (IF) signal derived from the alternating current (AC) output of the photomixer underwent band-pass filtering and was rectified to a direct current (DC) signal by a Schottky barrier diode. The IF signal was subsequently amplified by a low-noise preamplifier with a gain of 60 dB. The amplified IF signal was demodulated by a lock-in amplifier (LIA), which was synchronized with the modulation frequency of the mechanical chopper. The demodulated signal was then digitized by a DAQ card, operating at a sampling rate of 500 kHz. The DC output from the photomixer, which was used to monitor the ICL power, was also captured by the DAQ system.

The MIR-LHR was deployed in a field located in the suburbs of Hefei (31.9° N 117.16° E) for atmospheric column measurements of N₂O and CH₄. The crop rotation system in Hefei is rice-wheat rotation, with the rice season from April to October and winter wheat season from November to April of the following year. The observation period is from April 2019 to May 2022, which basically includes three rice-wheat rotation cycles.

2.2. Data Acquisition and Inversion Methods

In a single LO frequency scan, the LOs were initially set to temperatures of 10 °C and 14.5 °C. The laser controllers (LDC-3724, ILX Lightwave, Bozeman, MT, USA) adjusted the current in steps of 0.1 mA over the range of 34–44 mA to measure N₂O and CH₄ absorption at 2831.92 cm⁻¹ and 2538.34 cm⁻¹, respectively. The raw MIR-LHR signals were preprocessed before data retrieval, involving intensity normalization and wavenumber calibration. Intensity normalization involved subtracting the signal offset to align with the pre-laser emission mean value and dividing by the DC signal to correct for variations caused by the LO. The preprocessed MIR-LHR measurement results of the CH₄ and N₂O detections are shown in Figure 2a (red curve) and Figure 2b (blue curve).

The data retrieval process employed the optimal estimation method (OEM), originally proposed by Rodgers and later applied to LHR data by Weidmann [27]. For clarity, a brief overview is provided. The atmospheric transmission spectrum of solar radiance is derived using a radiative transfer forward model (F), developed from the reference forward model (RFM, version 4.34), a rapid line-by-line radiative transfer model [28]. The link between the processed LHR measurements and atmospheric state vectors is defined as follows:

$$y=F\left(x\right)+\epsilon$$

(1)

where y represents the measurement vector, x is the state vector used in the forward model, and ε denotes the error vector. The OEM data retrieval employs a Levenberg-Marquardt (LM) iterative approach, utilizing Bayesian statistics with Gaussian probability density functions to minimize the cost function (χ²):

$${\chi }^2={\left(y-F\right)}^T{S}_{\epsilon }^{-1}\left(y-F\right)+{\left(x_i-x_a\right)}^T{S}_a^{-1}\left(x_i-x_a\right)$$

(2)

where x_a is an a priori state vector with an a priori covariance matrix S_a, and S_ε is the error covariance matrix. The iterative state vector x_i+1 is calculated using the following equation:

$$x_{i+1}=x_i+{\left[\left(1+\gamma \right)S_a^{-1}+K_i^T{S}_{\epsilon }^{-1}{K}_i\right]}^{-1}\times \left[K_i^T{S}_{\epsilon }^{-1}\left(y-F_i\right)-S_a^{-1}\left(x_i-x_a\right)\right]$$

(3)

where K represents the Jacobian matrix, while γ is the Levenberg-Marquardt (LM) parameter. Figure 3 illustrates a schematic overview of the LHR retrieval process, which integrates a radiative transfer forward model and an inversion algorithm. In the forward modeling phase, the atmospheric transmission spectrum, denoted as F(x_n), is computed in conjunction with its respective weighting function, represented by K. The computation integrates the atmospheric state parameters, including temperature T, pressure P, and volume mixing ratios (VMRs), alongside the instrumental line shape (ILS), the a priori state vector, and the solar zenith angle. Subsequently, within the inversion phase, the forward model is invoked iteratively to optimize the cost function, as articulated in Equation (2), adhering to the Levenberg-Marquardt (LM) algorithm, as detailed in Equation (3). The iterative refinement culminates in the derivation of the target gas’s vertical profile.

2.3. Experimental Design and Yield Measurement

The field experiment in an annual rice-wheat rotation cropping system was performed in the suburbs of Hefei (31.9° N/117.166° E, 40 m above sea level), China. The climate is mild with moderate rainfall, featuring an annual average temperature of 15.7 °C and an annual average precipitation of approximately 1000 mm. Hefei has an annual frost-free period of 228 days and receives approximately 2000 h of sunshine. In 2019, statistical data indicated that the city’s grain planting area spanned roughly 363,000 hectares, with a total crop yield of 560,000 tons, of which rice accounted for approximately 400,000 tons.

Table 1. Dates of cultivation and management during the rice-wheat rotation in 2019–2022.

Crop	Date (Month/Day)			Agricultural Activity
	2019–2020	2020–2021	2021–2022
Rice season	4/27	4/20	4/23	sowing
	6/2	5/25	5/22	transplanting
	7/23	7/15	7/20	drainage
	8/2	7/25	7/30	flooding
	8/10	8/1	8/3	fertilizing (N-fertilizer)
	10/7	9/30	10/4	harvesting
Wheat season	10/22	10/20	10/24	sowing
	3/10	3/5	3/7	fertilizing (N-fertilizer)
	4/15	4/10	4/11	harvesting

shows the dates of major agricultural activities during rice-wheat rotation in 2019–2022. Although the timing of agricultural activities each year is affected by the weather conditions, the local annual field management activities and measures for rice and wheat are generally consistent.

The field was managed using traditional practices, which involved flooding the rice paddies during the growing season. After harvest, the fields were left uncultivated until the next planting. Continuous irrigation was maintained throughout the rice growth period, with the field being drained for 7–10 days prior to fertilization and rehydrated afterward. The irrigation water depth was kept at approximately 6–7 cm to ensure optimal rice growth conditions. Each observed plot followed local standard management practices, and the wheat season land was plowed to a depth of 15 cm before planting.

The meteorological conditions used in the experiments are shown in Figure 4. Spring (April to May) and autumn (October to November) saw average temperatures ranging from 15 °C to 20 °C, with summer (June to September) experiencing higher temperatures, averaging above 30 °C, and winter (December to March) being relatively cold, with average lows around 0 °C. In terms of precipitation, the summer season was the most rainfall-intensive, particularly in July, where monthly rainfall exceeded 200 mm, while winter months were comparatively dry, with less than 50 mm of rainfall on average. The raw spectral data of atmospheric N₂O and CH₄ were detected by the MIR-LHR using sunlight as the signal source. The raw data were collected during clear and cloudless conditions to minimize the potential confounding effects of solar radiation scattering and cloud absorption on the atmospheric column measurements. Therefore, signals were not collected during periods of heavy rainfall or cloudy weather.

3. Results

3.1. Inversion Results of MIR-LHR

The pressure and temperature profiles used for the data retrieval were obtained from the China Meteorological Data Service Center (CMCC) and the Europe Center for Medium-Range Weather Forecast (ECMWF). For the methane profiles, the data were interpolated from the typical mid-latitude daytime profiles. And an a priori profile of nitrous oxide was obtained from the Earth Observing System (EOS). Note that the atmosphere is separated into 45 layers from the surface to 70 km. Figure 5 shows the retrieved vertical concentration profiles of (a) CH₄ and (b) N₂O, respectively. Based on the retrieved CH₄ and N₂O profiles, the total column abundance of CH₄ was found to be ~1.906 ppm (or ~3.862 × 10²⁰ molecules/cm²) and the total column abundance of N₂O was found to be ~338 ppb (or ~6.834 × 10¹⁹ molecules/cm²), respectively. The total retrieval errors were calculated to be ~1% (CH₄) and ~0.8% (N₂O), respectively.

3.2. Atmospheric N₂O

Figure 6 shows the seasonal fluctuations in the N₂O column concentrations above the rice-wheat rotation system observed from 2019 to 2022. The data, captured during clear midday hours, are presented in two parts: (a) depicts the column concentrations of N₂O and (b) shows the associated residuals. Over the entire rice-wheat season, there was no obvious reduction in N₂O column concentrations. The data reveal seasonal variations in the atmospheric N₂O column concentrations within the rice-wheat rotation system, with peak concentrations reaching up to 8 ppb. By incorporating a linear fit to the data, the residual plot more clearly demonstrates the changes in atmospheric N₂O column concentrations.

The monthly average of atmospheric N₂O column concentrations measured from April 2019 to May 2022 is shown in Figure 7. The data reveal periodic fluctuations with distinct peaks and troughs. These fluctuations indicate strong seasonal variations and may be influenced by agricultural activities such as fertilization, which is known to increase N₂O emissions. The data show that the peak occurs around the middle of each year, indicating that agricultural activities were consistently carried out during this period. The error bar represents the standard deviation or measurement uncertainty. Although most of the data points show relatively consistent measurements, there are several periods with larger error margins, indicating that the concentration of N₂O varies greatly in these periods. In the measured data, the average column concentrations over each 12-month period were 335.7 ppbv, 336.8 ppbv, and 337.6 ppbv, respectively, indicating an overall upward trend, with an average annual increase of 0.7 ppbv.

Figure 8 presents the MIR-LHR N₂O retrieval results during the rice-wheat rotation period from 2020 to 2021. During the single cycle of the rice-wheat rotation, two rounds of nitrogen fertilization were implemented within the agricultural system. In the rice season, nitrogen fertilizer was applied on August 1st to support a crucial stage of rice growth and development. Subsequently, during the wheat season, another application of nitrogen fertilizer was conducted on March 5th. The water management strategy for the rice season followed a dual-level paddy field irrigation system, alternating between flooded and dry conditions. The drying phase was scheduled specifically from July 25th to August 1st. Outside of this period, the field remained flooded until a week before harvest, at which point the field was drained.

The results showed that the peak period of N₂O emissions over rice-wheat rotation farmland mainly occurred within 7–10 days after fertilization during the rice-wheat rotations, as depicted in the green area of Figure 8. At the same time, the peak of N₂O emission appeared obviously in the drainage period of the rice season, as shown in the purple area of Figure 8.

3.3. Atmospheric CH₄

As shown in Figure 9, the CH₄ column concentration above the rice-wheat rotation system in 2021–2022 was simultaneously monitored. During the single rice-wheat rotation cycle, the significant CH₄ emission only occurred in the rice-growing season, while there was no obvious CH4 emission or absorption in the winter wheat-growing season. The data indicated that significant CH₄ emissions mainly occurred during the paddy rice season under flooding conditions. In contrast, CH₄ emission and absorption in the rice season during the drying period and the wet irrigation period are negligible, as well as not being significant during the winter wheat season. The application of N-fertilizer significantly affected CH₄ emissions, indicating that agricultural practices can significantly affect the emission dynamics of the potent GHG.

4. Discussion

Previous studies on greenhouse gas emissions from rice-wheat rotation systems have focused on the emission fluxes of crops, without considering their impact on the columnar concentrations of the lower atmospheric layer. There is a certain correlation between the emission fluxes of crops and atmospheric column concentrations. However, this relationship is not well defined. The MIR-LHR, as a ground-based observational instrument, can be utilized to monitor the trends in greenhouse gas concentration changes above croplands.

As shown in Figure 6, the findings from this study, conducted from 2019 to 2022 with consistent water management practices in paddy fields, indicate that N₂O emissions were negligible in continuously flooded rice paddies. The minimal N₂O emissions observed occurred during the dewatering period prior to harvest, suggesting that water management practices play a crucial role in determining the N₂O emission levels. Notably, significant N₂O emissions were detected within 7–10 days after fertilization, inherently linked to the components of the applied fertilizer. Although background emissions of N₂O were not accounted for in this study, the data obtained still demonstrate the capability of the developed MIR-LHR for the continuous monitoring of emission trends above the rice-wheat rotation system.

The N₂O emissions during the drainage period of rice cultivation cannot be ignored, although most of the emissions occurred after fertilization in the rice-wheat rotation system. The alternating soil moisture conditions during the drainage period facilitate nitrification and denitrification processes. Fluctuations in the soil moisture and aerobic conditions resulting from paddy field drying can significantly increase N₂O production in rice fields [29,30,31,32]. The report emphasizes that the entire rice-wheat rotation cycle, including the drying stage, must be considered when assessing and managing N₂O emissions in agricultural practices.

As shown in Figure 9, CH₄ emissions from conventionally transplanted rice fields were primarily concentrated within 1 month after transplantation and peaked 2–3 weeks post-transplantation. This phenomenon is primarily due to the decomposition of residual crops in the previous season and the application of organic materials in the current season. During the drainage period, CH₄ emission flux decreased sharply, but it still remained a high emission flux after rehydration. The seasonal variation in CH₄ emissions is primarily controlled by water management, which is consistent with the extreme anaerobic conditions required for CH₄ production. As shown in the light purple area of Figure 9, the CH₄ emissions from rice paddies decreased sharply during drainage. Subsequently, the CH₄ emissions gradually increased during rice flooding. Under flooded conditions, the paddy fields produced an anaerobic environment, which was a crucial condition for CH₄ production. CH₄ was generated by the anaerobic decomposition of organic matter via methanogenic archaea under anoxic conditions. When the paddy field was drained, the redox potential of the soil increased, which enhanced the oxidation conditions and inhibited the activity of methanogenic archaea, thereby reducing the CH₄ production. After drainage, the oxygen level in the soil increased, which enhanced the activity of methanotrophic bacteria. These bacteria consume methane through the oxidation process [33,34]. As a result, CH₄ emissions into the atmosphere were reduced, leading to a decrease in the soil methane levels. During the observation period, the fluctuation in CH₄ emissions associated with the rice-wheat rotation system was approximately 40 ppbv, indicating that methane emissions were significantly reduced after farmland drainage.

Although our research highlights the potential of the MIR-LHR for the continuous detection of N₂O and CH₄ above the rice-wheat rotation system, there are certain limitations to the application of this new technology and the interpretation of some underlying mechanisms. The limitations are reflected in the fact that while the atmospheric column concentrations detected by the MIR-LHR can to some extent indicate the impact of agricultural activities, an analysis combined with flux measurements is not included. This means that the direct link between the observed column concentrations and the actual emissions from the agricultural practices, which would be provided by flux data, is not established in this study. Therefore, further research that integrates column concentration data with flux measurements is necessary to fully understand the dynamics of greenhouse gas emissions in the context of rice-wheat rotation systems.

Additionally, this study was conducted over a period of only three years, which is a relatively short timescale to fully assess the impact of agricultural management practices on atmospheric column concentrations. The brevity of the study period limits the ability to draw long-term conclusions about the effectiveness of various management strategies on greenhouse gas emissions. To obtain a more comprehensive understanding of the long-term effects of agricultural practices on N₂O and CH₄ concentrations, extended monitoring periods and a broader range of environmental conditions should be considered in future research.

5. Conclusions

In this study, we employed the innovative MIR-LHR to conduct a comprehensive analysis of CH₄ and N₂O emissions from a rice-wheat rotation system. Through long-term monitoring from April 2019 to May 2022, we detailed the atmospheric column concentrations of these gases, revealing distinct seasonal patterns and the influence of agricultural activities on their emissions. We retrieved the total column abundance of atmospheric CH₄ and N₂O based on the transmission spectra of the MIR-LHR, with errors of 2 ppbv and 9 ppbv, respectively. Our findings demonstrate a clear seasonal pattern for N₂O emissions, peaking during rice and wheat fertilization events. We observed an increase in the CH₄ levels during the rice season, contrasted with a significant decrease during the drainage period, underscoring the critical role of water management in regulating GHG emissions. The insights gained from our research provide a scientific foundation and practical approaches for developing sustainable agricultural practices aimed at minimizing greenhouse gas emissions. Future research should explore integrating MIR-LHR technology with other monitoring systems and examine additional variables such as the soil moisture and temperature to enhance our understanding of GHG dynamics in different agroecosystems.