Assessing the Impact of Climate Change on Methane Emissions from Rice Production Systems in Southern India

Abstract

The impact of climate change on methane (CH₄) emissions from rice production systems in the Coimbatore region (Tamil Nadu, India) was studied by leveraging field experiments across two main treatments and four sub-treatments in a split-plot design. Utilizing the closed-chamber method for gas collection and gas chromatography analysis, this study identified significant differences in CH₄ emissions between conventional cultivation methods and the system of rice intensification (henceforth SRI). Over two growing seasons, conventional cultivation methods reported higher CH₄ emissions (range: from 36.9 to 59.3 kg CH₄ ha⁻¹ season⁻¹) compared with SRI (range: from 2.2 to 12.8 kg CH₄ ha⁻¹ season⁻¹). Experimental data were subsequently used to guide parametrization and validation of the DeNitrification–DeComposition (DNDC) model. The validation of the model showed good agreement between the measured and modeled data, as denoted by the statistical tests performed, which included CRM (0.09), D-index (0.99), RMSE (7.16), EF (0.96), and R² (0.92). The validated model was then used to develop future CH₄ emissions projections under various shared socio-economic pathways (henceforth SSPs) for the mid- (2021–2050) and late (2051–2080) century. The analysis revealed a potential increase in CH₄ emissions for the simulated scenarios, which was dependent on specific soil and irrigation management practices. Conventional cultivation produced the highest CH₄ emissions, but it was shown that they could be reduced if the current practice was replaced by minimal flooding or through irrigation with alternating wetting and drying cycles. Emissions were predicted to rise until SSP 370, with a marginal increase in SSP 585 thereafter. The findings of this work underscored an urgency to develop climate-smart location-specific mitigation strategies focused on simultaneously improving current water and nutrient management practices. The use of methanotrophs to reduce CH₄ production from rice systems should be considered in future work. This research also highlighted the critical interaction that exists between agricultural practices and climate change, and emphasized the need to implement adaptive crop management strategies that can sustain productivity and mitigate the environmental impacts of rice-based systems in southern India.

1. Introduction

The Intergovernmental Panel on Climate Change (IPCC)’s Sixth Assessment Report [1] highlighted the progressive increase in greenhouse gas (GHG) emissions observed since the 1750s. Such increases in emissions have been primarily attributed to human activities, with CH₄ levels reaching ~1900 ppb in 2019. There has also been a significant rise in global surface temperatures, which increased from 0.8 °C between 1850 and 1900 to 1.3 °C between 2010 and 2019. These trends will likely continue in the near-term [1]. Rice (Oryza sativa L.) is a staple food crop for over three billion people worldwide and is cultivated in puddled conditions in tropical and sub-tropical regions. Rice cropping contributes approximately 11% of global CH₄ emissions from anthropogenic sources [2,3]. Such emissions are projected to increase in response to higher demand for rice to support a growing population [4,5]. Studies have shown that GHG emissions from rice cultivation can be similar to, and often higher than, the emissions reported for wheat (Triticum aestivum L.) or maize (Zea mays L.) production, including those from intensively managed systems [6]. For rice-based systems, high emissions rates are, in part, due to the anaerobic decomposition of soil organic matter (SOM) in flooded fields, where the redox potential can range between −150 and −300 mV [7].

Methane production in rice fields is influenced by inherent soil characteristics (importantly, SOM content and soil pH), water and nutrient management, and the soil biogeochemical processes directly affected by management (e.g., irrigation practices) [8]. Methane is released to the atmosphere through various pathways that include plant aerenchyma, ebullition, and diffusion in flooded field conditions. Adoption of dry seeding methods has been shown to reduce CH₄ emissions [9]. The nitrogen (N) fertilizer type and placement in the soil, and the application rate can also play a critical role in influencing emissions, but these effects are dependent on the soil type and the specific environmental conditions when fertilizer is applied. For example, the application of ammonium sulfate [(NH₄)₂SO₄] can mitigate CH₄ emissions, as it affects the soils’ redox potential. Previous studies [10,11] reported a 43% reduction in CH₄ emissions following application of 100 kg ha⁻¹ of (NH₄)₂SO₄ compared with other (straight) sources of nitrogen. Such reductions were attributed to changes in the redox potential, together with decreased production of root exudates (this being a primary source of food for methanogenic soil microorganisms), and increased competition between sulfate-reducing bacteria and methanogens [10,11].

The spatiotemporal variability of CH₄ emissions from rice fields poses a significant challenge for biophysical simulation models. Recent advances in mechanistic/process-based modeling approaches have led to the development of geochemical process-based simulation models, such as DeNitrification–DeComposition (DNDC). This model can be used to provide estimates of CH₄, nitrous oxide (N₂O), CO₂, and ammonia (NH₃) emissions, and nitrate (NO₃) leaching losses from the (sub-)field through to the farm and catchment scales, and it has been employed for biophysical simulation of rice, upland crops, and wetland systems [12]. The data required for validation of the DNDC model can be sourced both from pot culture and field experiments. This model has been widely utilized for assessing GHG emissions under different irrigation water management regimes, and fertilizer or organic manure applications. Despite extensive research on methane emissions from agricultural systems, and their contribution to climate change, studies examining the impact of climate change scenarios on methane emissions in rice ecosystems remain scarce. This is particularly true for the farming systems of southern India, where several methods of rice cultivation have been adopted, including continuous flooding, direct seeding, alternating wetting and drying cycles, and the system of rice intensification. Therefore, the aim of the work reported in this article was to fill the knowledge gap identified above by quantifying methane emissions under different rice cultivation methods that included a range of irrigation management practices combined with application of synthetic fertilizer and organic amendments. The DNDC model, validated with data derived from field experiments, was employed to provide estimates of methane emissions under projected (future) climate scenarios.

2. Materials and Methods

2.1. Field Experiments

Field experiments were conducted over two consecutive rice growing seasons to determine the impact of different cultivation methods and fertilizer treatments on CH₄ emissions. The experiments were established at the Tamil Nadu Agricultural University at Coimbatore (India) on a wetland field (11°00′06.6960″ N, 076°55′34.5000″ E, elevation: 427 m above sea level) during the Rabi season of 2015–2016 and the Kharif season of 2016–2017. The rice variety CO(R)51 (fine grain; 105–110 days from planting to harvest) was sown on 26 August 2015 and 13 July 2016. The soil type at the experimental site was a clay loam with an alkaline pH_1:5 (8.54) and EC_1:5 of 0.13 dS m⁻¹. A full description of the soil type at the site is given in the Supplementary File attached to this article.

The study utilized a split-plot design with two main treatments, namely (1) the conventional method, and (2) the system of rice intensification (henceforth SRI). These treatments were further divided into four sub-treatments according to the type of soil amendment applied as follows: (1) urea (46% N), (2) ammonium sulfate (21% N, 24% SO₃), (3) urea + vermicompost, and (4) ammonium sulfate + vermicompost. Each sub-treatment had three replications (n = 3) to ensure sufficient statistical reliability. The average nutrient composition of vermicompost was 2.13 ± 0.596% N, 1.68 ± 0.405% P, and 1.63 ± 0.748% K. In the conventional method, standing water to a depth of 5 cm was maintained throughout the cropping season until the crop reached physiological maturity. The SRI method is a management system in which young rice seedlings are planted singly in a square grid pattern, and the soil is kept moist but well-drained throughout the entire rice growing period. Under the SRI method, irrigation was applied, on average, once every 7 days, which enabled the soil to remain near the drained upper limit throughout the season. The total nutrient load delivered to both the conventional and SRI methods was 150 kg ha⁻¹ N (whether as urea or ammonium sulfate, depending upon the sub-treatment), and basal applications of P and K at rates of 50 kg ha⁻¹ each. The vermicompost was applied at a rate of 5 Mg ha⁻¹, based on standard (local) agronomic advice. During the 2015–2016 season, the total N rate was applied in split applications as follows: 25% as basal application, 25% at 31 days after transplanting (DAT), 25% at 44 DAT, and the remaining 25% at 66 DAT. During the 2016–2017 season, the total N rate was applied as follows: 25% as basal application, 25% at 30 DAT, 25% at 59 DAT, and 25% at 78 DAT. The amendments were incorporated into the soil following basal application and top-dressed (no soil incorporation) in the subsequent in-crop applications.

Key field operations for crop establishment and crop management involved transplanting, gap filling, cono-weeding, topdressing of fertilizer, and irrigation. Cono-weeding was performed only in SRI, with all other cultural practices (except for water) being the same in both the conventional and SRI cultivation methods. The Supplementary File provides detailed records of the field operations performed at the experimental site. In addition to the agronomic trial described above, gas samples were collected both from the conventional and SRI methods for all sub-treatments over the two growing seasons. Samples were analyzed using gas chromatography, which enabled CH₄ and N₂O fluxes to be estimated as a function of the imposed treatment. Measurements of CH₄ and N₂O aimed to capture a broader spectrum of GHG emission scenarios across a range of rice cultivation practices that are common in southern India, thus representing realistic farming conditions under the current climate.

2.2. Sample Collection and Analysis

Methane emissions were quantified using a custom-built acrylic chamber (dimensions: 60 cm × 30 cm × 100 cm). To assess the net change in CH₄ concentration, which represented the balance between CH₄ emissions and absorption, measurements were taken by withdrawing a sample from the chamber at time zero (immediately after the chamber’s closure) and subsequently at regular intervals of 1 h. The gas chromatograph used to analyze the sample was equipped with a flame ionization detector (FID). Sampling of CH₄ was conducted at key phenological phases of the rice crop development that included active tillering, flowering, milking, and grain-filling stages. This approach allowed for a comprehensive analysis of CH₄ emissions across different stages of crop development, providing insights into the temporal dynamics of CH₄ release in rice cultivation systems. A Shimadzu GC-2014 gas chromatograph instrument equipped with a flame ionization detector (FID) was utilized. Nitrogen, thoroughly purified, was used as the carrier gas. Gas samples were introduced into the analyzer via a 1.0 mL fixed loop on the sampling valve. The samples were then injected into the column system, where the analyzer, upon activation, triggered a valve to begin cycling the samples according to a pre-set schedule. The gas chromatograph was carefully calibrated with different concentrations of CH₄, ranging from 1 ppm to 5 ppm, both before and after each measurement. This calibration resulted in an average retention time for CH₄ of between 4 and 4.17 min (Chemtron^® Science Laboratories Pvt. Ltd., Mumbai, India). The primary standard curves displayed linearity across all the concentration ranges tested. The oven temperature was maintained at 100 °C, while the FID was operated at 200 °C to detect CH₄. The minimum detectable limit for CH₄ was 1 ppm, and the flux rate, which was determined by measuring the peak area, was expressed as mg m² day⁻¹ [13].

2.3. Estimation of Methane Emissions

Methane emissions from all treatments were calculated from experimental data collected at the sites using Equation (1) (after [14])

$$F={\displaystyle \frac{V}{A}}\times {\displaystyle \frac{∆C}{∆t}}$$

(1)

where F is the CH₄, N₂O or CO₂ emission rate (mg m⁻² h⁻¹), V is the volume of chamber above the soil (m³), A is the cross-sectional area of the chamber (m²), ΔC is the difference in concentration between time zero and time t (mg m⁻³), and Δt is the time interval between two consecutive sampling events (h).

Greenhouse gas emissions, expressed as kg ha⁻¹ over the entire season, were estimated as follows

$$GHG={\displaystyle \frac{F\times 24\times \mathrm{10,000}\times D}{\mathrm{1,000,000}}}$$

(2)

where F is the methane emission rate (mg m⁻² h⁻¹), 24 is used to convert from hours to days, 10,000 is used to convert from m² to ha, D is the duration of the crop (days), and 1,000,000 is used to convert from mg to kg.

2.4. Model Description and Validation

The DeNitrification–DeComposition (DNDC) model is a process-based tool that simulates soil conditions, crop yield, and the impact of management practices on soil organic carbon (SOC) dynamics and gas emissions (N₂O, NO, CH₄, NH₃) in agricultural ecosystems. Operating on a daily time step, the model has two main components: one for predicting soil temperature, soil moisture, soil pH and redox potential, and substrate concentrations, and the other for estimating emissions of gases from the soil–plant system. The model integrates physical, chemical, and biological processes with empirical data and has been widely utilized across a wide range of agro-ecosystems. The DNDC model requires data on soil properties (pH, SOC, texture), land use (cropping), climate, and management practices (fertilizer use, tillage, irrigation water). The model can track the crop’s biomass and decomposition rates to simulate SOC dynamics and predict N₂O emissions by simulating nitrification and denitrification processes [12]. Given that the model has been calibrated in different soil and climatic conditions, only one validation of the model was performed as part of our study. The validation of the DNDC model was conducted using data obtained from the field experiments described earlier. Additionally, measured CH₄ emission data collected from on-farm field trials (the conventional method with continuous flooding and the SRI method) were employed to further validate the DNDC model. This dual approach to validating model-derived data, which combined experimental plot data with on-farm trial results, ensured that a robust evaluation of the model could be performed across a range of rice cultivation scenarios.

2.5. Statistical Analyses

Modeled CH₄ emissions were compared with measured values from experimental work, and the model’s quality was determined using the following tests: Residual Mean Square Error (RMSE), coefficient of determination (R²), D-Index, coefficient of residual mass (CRM), and model efficiency (EF) [15,16]. The RMSE was employed to measure the average deviation between model simulated outputs and observed values, with lower RMSE values indicating greater model accuracy. The R² value provided a measure of how well the observed outcomes were replicated by the model, based on the proportion of total variation of the outcomes explained by the model. The D-Index ranges from 0 to 1, and it served to complement the model evaluation, where values that approximated 1 indicated better model performance. A CRM value of zero means perfect agreement (100%) between measured and modeled data. The EF (which accounts for negative values) gauged the relative deviation between observed and predicted values against the variability of the observed data, with a maximum value of 1 indicating flawless model predictions. Together, these statistical tests provided a robust framework for interpreting the modeled data and assessing the accuracy of the DNDC model.

2.6. Climate Projection Data

For the projection of future climate scenarios, data corresponding to the Shared Socio-economic Pathways (SSPs) for Coimbatore (11°01′00″ N, 76°57′20″ E), Thanjavur (10°47′13.2″ N, 79°08′16.1″ E), and Tirunelveli (8°42′49″ N, 77°45′24″ E) were acquired from the Coupled Model Inter-comparison Project Phase 6 (CMIP6) through the Earth System Grid Federation (ESGF) portal (https://esgf-node.llnl.gov/search/cmip6/, accessed 22 November 2023). The selected datasets were available at a daily timescale, with a spatial resolution of 0.25°. To align these data with historical (1951–2014) and projected (2015–2100) climate scenarios, bias correction was performed utilizing the Empirical Quantile Mapping (EQM) method, which ensured that the required accuracy and relevance for the SSP 126, SSP 245, SSP 370, and SSP 585 scenarios were achieved [17]. Each SSP scenario represented varying degrees of radiative forcing and emissions trajectories. SSP 126 is a low-emission and mitigation scenario with a 2.6 Wm⁻² radiative forcing, embodying optimistic environmental sustainability efforts. Conversely, SSP 585 represents a high-emission scenario with an 8.5 Wm⁻² radiative forcing, indicating continued high emissions and limited mitigation efforts. The intermediate scenarios, SSP 245 (4.5 Wm⁻² radiative forcing) and SSP 370 (7.0 Wm⁻² radiative forcing), span the spectrum between low- and high-emission outcomes, and provide a wider range of predicted future climate conditions. The climate data, once converted into a format compatible with the DNDC model, were utilized to analyze the impacts of climate change on CH₄ emissions in rice production systems under varying projected future climate scenarios. This approach allowed for a detailed examination of how different climate change trajectories could potentially drive CH₄ production and release into the atmosphere, according to specific crop, soil and water management practices commonly used in rice cultivation in southern India.

2.7. Carbon Dioxide Concentrations

To assess carbon dioxide (CO₂) concentrations over the historical period, recorded data from the Mauna Loa Observatory, Hawai’i (19°28′46″ N, 155°36′10″ W, elevation: 4169 m above sea level) were utilized, providing a reliable baseline of atmospheric CO₂ levels. For future projections, an average increase in CO₂ concentration during historical periods was calculated to establish a trajectory for rising CO₂ levels. Subsequently, CO₂ increments were set at 2, 3, and 4 ppm per year for the mid-century forecasts. For projections towards the end of the century, the increased rates were adjusted to 3, 4.5, and 6 ppm of CO₂ per year, respectively, to reflect an anticipated acceleration of the growth in CO₂ concentration due to various socio-economic pathways and environmental feedback mechanisms. These projections were categorized according to the mid- and late-century timelines to be able to capture the evolving impact of CO₂ on climate dynamics and agricultural ecosystems. The specified increments served as parameters for modeling the potential influence of rising CO₂ levels on CH₄ emissions in rice cultivation, underpinning the analysis with a nuanced understanding of atmospheric conditions.

2.8. Climate Change Mitigation Studies

After the DNDC model was validated and the climate change impact analysis for rice-based systems was conducted, mitigation options were considered. For this, the options considered were incorporated into the DNDC and their effectiveness, in terms of potential CH₄ emission reductions, was determined. These mitigation strategies included minimal flooding (maximum of 10 irrigations) and alternating wetting and drying (AWD) cycles. Such strategies were applied to selected locations (Coimbatore, Thanjavur, and Tirunelveli), and their effectiveness in reducing CH₄ emissions was then assessed. These locations have distinct climatic and soil conditions, which allowed the analysis to explore likely effects across different environments. The results derived from the analyses provided valuable insights into how management practices could help mitigate (or otherwise increase) GHG emissions from rice production under a range of climatic conditions.

3. Results and Discussion

3.1. Effects of Cultivation Methods and Soil Amendments on Methane Emissions

Seasonal variations in CH₄ emissions were evident from the field experiments, as shown in

Table 1. Effect of different cultivation methods and soil amendments on methane (CH₄) emissions from rice production systems. CM, conventional method; SRI, system of rice intensification.

Treatment (Sub-Treatment)	CH4 (kg ha−1 season−1)
Season	2015–2016	2016–2017
CM (urea)	49.4	59.3
CM (ammonium sulfate)	43.8	51.8
CM (urea + vermicompost)	42.9	45.1
CM (ammonium sulfate + vermicompost)	36.9	42.3
SRI (urea)	6.9	12.8
SRI (ammonium sulfate)	4.9	10.1
SRI (urea + vermicompost)	3.7	7.2
SRI (ammonium sulfate + vermicompost)	2.2	4.8

. Emissions were significantly higher in the second (range: from 4.8 to 59.3 kg CH4 ha⁻¹) compared with the first season (range: from 2.2 to 49.4 kg CH₄ ha⁻¹). Analysis by cultivation methods over two years revealed that CH₄ emissions were the highest with the conventional method and were the lowest with the SRI method (range: from 2.2 to 12.8 kg CH₄ ha⁻¹ per season). The relatively higher CH₄ production observed with the conventional method was attributed to the continuous flooding regime, which enhanced methanogenesis. By contrast, with the SRI method, saturated soil conditions were avoided through better control of the amount of water applied to the crop and by adjusting the frequency of irrigation. Such conditions are known to reduce the risk of CH₄ formation in the soil [18]. The presence of oxygen boosts methanotrophs’ activity, converting CH₄ into CO₂ + H₂O, while continuous flooding leads to anaerobic conditions in soil that encourage methanogenesis [19]. Further examination of the impact of N fertilizer on CH₄ emissions indicated that the soil treated with urea exhibited the highest emissions, followed by the ammonium sulfate treatment. The application of vermicompost generally reduced CH₄ emissions, making the combined use of SRI with ammonium sulfate and vermicompost the treatment with the lowest emissions. These observations were consistent with previous studies (e.g., [8,20,21]), in which CH₄ emissions under the conventional method (permanent flooding) were up to 60% higher than under SRI.

Incorporation of organic manures into the soil during the final puddling in rice production has been shown to reduce CH₄ emissions (e.g., [22] in China and [23] in India). These findings underscore the importance of such practices in mitigating CH₄ emissions from rice fields. Targeted interventions, such as the SRI method used here, is an effective management strategy for reducing emissions, as shown by our study, in which water was supplied to the crop at 7-day intervals. This irrigation frequency allowed the soil water content to remain near the drained upper limit and to avoid anaerobic conditions developing due to prolonged soil saturation. Implementation of the SRI method also reduced irrigation water use, which concurrently increased water productivity.

3.2. Validation of the DNDC Model

The validation of DNDC model was performed with experimental data derived from the field trial by comparing observed with modeled data. The DNDC-simulated CH₄ emissions showed good agreement with the measured data, which was observed for all treatments. Measured CH₄ emissions ranged from 2.2 to 84.2 kg ha⁻¹ season⁻¹ compared with 3.3 to 80 kg ha⁻¹ season⁻¹ as derived from the model simulations. The statistical analyses for the observed and modeled data are shown in

Table 2. Statistical estimates for the comparison of observed and simulated parameters.

Statistical Estimate	CH4 Emissions (as kg C ha −1)
Residual Mean sSquare Error (RMSE)	7.16
Coefficient of residual mass (CRM)	0.09
D-Index	0.99
Model efficiency (EF)	0.96
Coefficient of determination (R2)	0.92

, and these analyses confirmed the reliability of the model. Figure 1 shows the correlation between the observed and modeled CH₄ emission data (R² = 0.92; SE = 2.42). The estimated error was also within an acceptable range (RMSE = 7.16 kg ha⁻¹; D-Index = 0.99; CRM = 0.09; model efficiency = 0.96).

3.3. Climate Change Impact Analysis and Mitigation Strategies

The DNDC model, once validated, was employed to estimate the effect of climate change on CH₄ emissions in rice production systems, revealing both temporal and spatial variations in such emissions. The shared socio-economic pathway (SSP) scenarios from the IPCC enabled the assessment of climate-related impacts for the two contrasting rice cultivation methods used in southern India. A steady increase in CH₄ emissions was observed under the SSP 126 and SSP 370 scenarios, with a slight uptick also noted under the SSP 585 scenario. This trend was consistent across all three locations (Coimbatore, Thanjavur, and Tirunelveli). The increase in CO₂ concentrations found at all these locations may contribute significantly to the increased CH₄ output. As CO₂ levels rise, they can enhance plant growth and biomass production, which, if returned to the soil as stubble, will provide a source of organic matter for CH₄-producing microbes in (flooded) soil. This interaction between rising CO₂ concentrations and increased CH₄ emissions underscores the complex dynamics of GHG emissions in rice production systems. There is, therefore, a need for integrated approaches to simultaneously managing both CO₂ and CH₄ emissions in rice-based cropping systems for their effective mitigation.

Under the conventional method, the highest baseline emission of CH₄ was observed at Tirunelveli (75.8 kg ha⁻¹ season⁻¹), followed by Thanjavur (74.6 kg ha⁻¹ season⁻¹) and Coimbatore (42.1 kg ha⁻¹ season⁻¹). In contrast, the lowest baseline CH₄ emission was achieved with the SRI method byalternating wetting and drying cycles, which was observed at all three locations (Tirunelveli: 9.5 kg ha⁻¹ season⁻¹; Thanjavur: 9.1 kg ha⁻¹ season⁻¹; Coimbatore 6.5 kg ha⁻¹ season⁻¹). The future climate scenario analysis indicated an increase in CH₄ emissions at all three locations. For SSP 126 (near mid-century), the model showed that Thanjavur would have the highest (seasonal) CH₄ emissions (~115 kg ha⁻¹), followed by Tirunelveli (~98 kg ha⁻¹) and Coimbatore (~96 kg ha⁻¹), under the conventional method (Figure 2). The projected scenarios of CO₂ increase (that is, 2.5, 3, and 4 ppm) had little effect on the resultant increase in CH₄ emissions, as these increased by less than 3 kg CH₄ ha⁻¹ per season when the atmospheric CO₂ concentration was concurrently increased. With the SRI method, future emissions are expected to increase at all three locations, being highest at Coimbatore (~52% increase on average, depending upon the concurrent increase in CO₂ concentration), followed by Thanjavur and Tirunelveli (~34% and ~12% increases, respectively) relative to the baseline emission levels. An increase in emissions with SRI is also expected by near mid-century, following a similar trend to the conventional method. On average, and depending on the concurrent CO₂ increase, CH₄ emissions are expected to be the highest at Thanjavur (~119 kg CH₄ ha⁻¹ season⁻¹), followed by Tirunelveli (~104 kg CH₄ ha⁻¹ season⁻¹) and Coimbatore (~101 kg CH₄ ha⁻¹ season⁻¹) under the conventional method for the SSP 126 late mid-century analysis (Figure 3). A similar trend in seasonal CH₄ emissions for the SSP 126 late mid-century is expected with the SRI method.

For the near mid-century under the SSP 245 scenario (Figure 2), CH₄ emissions with the conventional method were projected to increase by an average of ~112% at Coimbatore, followed by Thanjavur (~45%) and Tirunelveli (~17%), depending upon the assumed increase in atmospheric CO₂ levels, whereas with the SRI method, the expected increase in CH₄ emissions was higher at Thanjavur, followed by Tirunelveli and Coimbatore. For the late mid-century under the SSP 245 scenario (Figure 3), CH₄ emissions were projected to be the highest at Thanjavur (~118 kg CH₄ ha⁻¹ season⁻¹), followed by Coimbatore (104 kg CH₄ ha⁻¹ season⁻¹) and Tirunelveli (99 kg CH₄ ha⁻¹ season⁻¹) with the conventional method (average across all CO₂ scenarios). A similar trend was observed with SRI; however, CH₄ emissions were found to be considerably lower for all CO₂ scenarios. For the near mid-century under the SSP 370 scenario (Figure 2), the analysis showed that CH₄ emissions with the conventional method will be highest at Thanjavur (~118 kg CH₄ ha⁻¹ season⁻¹), followed by Tirunelveli (~101 kg CH₄ ha⁻¹ season⁻¹) and Coimbatore (~97 kg CH₄ ha⁻¹ season⁻¹) on average across all assumed CO₂ levels. With the SRI method, CH₄ emissions were found to be lower, but the increases were still large (69–73% at Coimbatore, 47–50% at Thanjavur, and 25–28% at Tirunelveli, depending on the assumed CO₂ increase) relative to the baseline emission levels. With the SRI method, CH₄ emissions are also expected to increase by near mid-century but to a lesser extent than with the conventional method. Figure 3 (SSP 370 late mid-century scenario) shows that with the conventional method, emissions were found to be highest at Thanjavur (~126 kg CH₄ ha⁻¹ season⁻¹), followed by Tirunelveli (~112 kg CH₄ ha⁻¹ season⁻¹) and Coimbatore (~110 kg CH₄ ha⁻¹ season⁻¹), depending on the assumed CO₂ levels.

Under the SSP 370 late mid-century scenario (Figure 3), similar trends in CH₄ emissions were found for both cultivation methods, but there were differences between locations (Thanjavur >> Tirunelveli > Coimbatore). These trends highlighted significant regional variations in the projected CH₄ emissions from rice production systems and emphasized the need for tailored mitigation strategies for each region. Under the SSP 585 near mid-century scenario (Figure 2), CH₄ emissions with the conventional method were projected to increase significantly from the baseline at all locations. The simulations showed increases in CH₄ emissions of ~115% for Coimbatore, ~46% for Thanjavur, and a little less than 20% for Tirunelveli. For the SRI method, the increase in emissions was greatest at Thanjavur, followed by Tirunelveli and Coimbatore. These simulated results suggested that while alternative irrigation methods can contribute to reduce CH₄ emissions, adjustments to specific edapho-climatic conditions may be needed to improve their efficiency.

Future increases in CH₄ emissions with the SRI method will vary by region. Projections for Thanjavur showed that emissions may be the highest, followed by Tirunelveli and Coimbatore. These simulated results suggested that while the SRI method may offer advantages in terms of potential for emission reductions, regional differences will continue to be significant. The effectiveness of SRI is therefore likely to be influenced by the specific conditions of the region in which the practice is implemented. During the late mid-century under the SSP 585 scenario (Figure 3), the simulations showed that the highest CH₄ emissions are expected to be in Thanjavur with projected levels of 126, 128, and 129 kg ha⁻¹ season⁻¹ assuming increases in CO₂ of 2.5, 3.0, and 4.0 ppm, respectively. Coimbatore will follow, with CH₄ emissions reaching 109, 110, and 111 kg ha⁻¹ season⁻¹ at the same levels of increase in CO₂. Simulations for Tirunelveli showed the lowest emissions among all three regions, with projected levels of 107, 108 and 110 kg CH₄ ha⁻¹ season⁻¹ at increases in CO₂ of 2.5, 3.0, and 4.0 ppm, respectively, when rice is cultivated using the conventional method. Although a similar trend is expected with the SRI, CH₄ emissions will be significantly lower compared with the conventional method. This reduction highlights the potential benefits of adopting the SRI method over traditional practices under future climate scenarios, emphasizing the critical role of water management practices in reducing such emissions. The SSP 585 scenario further highlighted the correlation between increased CO₂ concentrations and CH₄ emissions, with increases across all locations and cultivation methods. This observation suggested that while adjustments in cultivation methods can mitigate emissions, overarching climatic factors (e.g., CO₂ levels) will play a crucial role in determining CH₄ output.

The observed increase in CH₄ production under conventional methods can be attributed to the anoxic conditions fostered by continuous flooding, facilitating anaerobic digestion of organic substrates, in agreement with earlier work (e.g., [24,25,26]). Similarly, other studies have reported [27] increased CH₄ emissions during wetter compared with drier seasons. Increased CO₂ levels under the various SSP scenarios can boost biomass production and the increased release of root exudates, which, in turn, can encourage methanogens’ activity under flooded soil conditions. However, in low-emission scenarios (e.g., SSP 126), reduced temperature and CO₂ levels can restrict the production of root exudates, thus curtailing methanogens’ activity and the release of CH₄. These findings underscore the complex interplay among cultivation methods, irrigation water management, and climatic conditions in influencing CH₄ emissions from rice-based systems. There is a requirement for the implementation of integrated approaches that are able to combine proven agronomic practices with broader climate change mitigation strategies to effectively reduce greenhouse gas emissions from irrigated agriculture.

4. Conclusions

The experimental studies and climate change impact analyses conducted as part of this work have elucidated the complex dynamics of methane emissions within rice-based cropping systems. It is evident that the conventional cultivation method, characterized by continuous flooding, is a significant contributor to methane emissions from rice production systems. Generally, the use of urea resulted in higher methane emissions compared with ammonium sulfate. The use of vermicompost with either urea or ammonium sulfate yielded lower methane emissions across all cultivation methods compared with synthetic fertilizers alone, suggesting potential advantages in their combined application.

The good agreement encountered between the measured and modeled data, indicated that the DNDC model simulations of methane emissions are reliable for all climate scenarios and cultivation methods. Under future climate scenarios (SSP 126 to SSP 370), methane emissions will likely increase, both in the near and late mid-century, with a relatively smaller rise under the SSP 585 scenario. These projections highlighted a direct correlation between atmospheric CO₂ concentrations and methane emissions, emphasizing the importance of specific edapho-climatic conditions, fertilizer management (rate and source), seasonal weather, and irrigation water management in determining emission levels. It is recommended that location-specific research be conducted to further refine GHG emissions models and modeling approaches to help develop mitigation strategies tailored to regional conditions.

5. Future Research and Development Opportunities

The development of user-friendly, region-specific GHG emission models is crucial for the accurate estimation of methane emissions. Current models present challenges in terms of their input requirements and applicability to specific cropping systems or environmental conditions. There is a need for simpler tools that can accommodate agricultural practices and edapho-climatic conditions outside those in which the models were initially developed, such as the incorporation of dry seeding for rice cropping and livestock components. Effort should be spent on refining the available models and modeling approaches to enable more precise quantification of methane emissions for a wider range of crops/cropping systems through their rotation cycle and by considering climate scenarios.