To enhance the quality of rice-CH4 emissions accounting by reflecting spatial heterogeneity, we conducted on-site measurements at various paddy fields, by the closed chamber (CC) approach, to help reduce the accounting uncertainty of rice-CH4 emissions. The relevance of CC and eddy covariance (EC) methods for measuring methane fluxes in a flooded rice paddy field was assessed27. The CC method is capable of capturing the heterogeneity within the field with its mobile facility and direct link to the measurement point. The advantage of EC technique lies in its wider area in scope and high temporal resolution in monitoring daily and seasonal changes. The result from the CC method was generally higher than that of the EC method and tended to overestimate due to the inclusion of optimally grown rice plants at high temperatures for flux measurements, and the EC method aggregated different sources and masked the individual processes behind the fluxes at each point27.
Our on-site rice CH4 emission measurements were conducted in the rice paddies at 17 sites across 9 counties of Taiwan (Supplementary Table S1) following the CC method28. In the rice paddy, an ultraportable gas analyzer (LGR915-0001, Los Gatos Research, San Jose, CA, USA) connected with an in situ closed chamber attached onto a stainless-steel frame was inserted to a 10 cm depth of soil. A whole-plant chamber was used for CH4 measurement in the rice paddies. During the rice seedling period (transplantation and first fertilization), a semicircular transparent acrylic chamber 30 cm in diameter and 16 cm in height was placed over the rice seedling, and the air volume was 10.6 L over a 0.071 m2 surface area. When the plant became mature (active-tillering fertilization, flowering fertilization, and ripening), a cuboid transparent chamber with dimensions of 25 × 25 × 85 cm (length × width × height) was placed over the rice paddy to create a space with 59.38 L of air over a 0.062 m2. The exchange of GHG concentration between the soil and the air in the chamber was monitored by the gas analyzer and recorded on a data logger. The logging frequency was every 20 s for a short incubation time of 10 min because increased humidity in the chamber can affect stomatal activity or inhibit the pressurized-ventilation system of plants.
The CH4 measurements were carried out in triplicate at each rice paddy during every field visit. Triplicate measurements were randomly chosen, and the distance between each replicate was at least 5 m to prevent possible disturbances. The CH4 flux was calculated and adapted from the ideal gas law (Eq. 1).
Flux= (S × V × tc × M)/(RT × 1000 × A) (Eq. 1)
where Flux: CH4 flux (mg m− 2 h− 1), S: slope of the linear regression line between CH4 concentrations (ppm) and recorded frequency (20 seconds), V: chamber volume (L), tc: time unit transformation constant: 180 = 1 hour × (60 min/hour) × [(60 s/min)/20 s], M: molecular weight (g/mol): CH4 = 16, R: ideal gas constant = 0.082 (L atm K− 1 mol− 1), T: absolute temperature (K), A: the area of the bottom part of the chamber (m2), 1000: weight unit transformation constant: 1 mg = 1000 µg.
The cumulative CH4 emissions during the whole cultivation period were determined to compare the CH4 emissions between different seasons and among different sites. At the beginning of the experiment, in addition to light conditions, we used a black cloth to cover the chamber and then measured the CH4 emissions to evaluate such emissions in the dark. We also compared the CH4 emissions between daytime and nighttime. Based on the preliminary results, there was no significant difference between light and dark environments and day and night for CH4 emissions. Thus, we only measured CH4 fluxes under light conditions, and CH4 emissions were represented on a daily basis by multiplying the CH4 flux (mg m− 2 h− 1) (Eq. 1) by 24 hours. Then, the cumulative CH4 emissions in each cultivated season were the summation of the GHG flux on a daily basis multiplied by the number of days of the corresponding cultivation periods from transplanting to harvesting.
Emissions accounting matters in the assessment of mitigation potential, which closely relates to the magnitude of emissions. Estimation models are likely to mis-estimate emissions if they do not fully capture the pertinent driving factors that are specific to the cultivation sites. The existing estimation models for rice CH4 emissions may greatly underestimate the magnitude of emissions (for example, ref. 29). These estimation models did not consider site-specific conditions such as soil texture, planting method, cultivar type, and growing season, although they did factor in soil organic carbon (SOC), pH, water level management during the crop season and preseason, and organic amendment application. For most rice-growing countries, the models produced much lower estimates than the measured emissions29.
FAOSTAT, taking the IPCC Tier-1 method with default emissions factors, is prone to underestimate rice CH4 emissions. Global rice and aggregate agricultural CH4 emissions were re-estimated by incorporating updated information from recent literature on rice farming systems globally and rice CH4 emission measurements4, 30. Ref. 30 concluded (1) an estimate of 34 million tons CH4 emitted per year globally from rice cultivation constitutes 35% more than the FAOSTAT estimation (25 million tons) and other prior estimates; and (2) an estimate of 156 million tons CH4 emissions from all agricultural activities globally represents a 6% higher level than that estimated by the Tier-1 IPCC approach in the FAOSTAT (147 million tons CH4). These re-estimations revealed that large underestimation in both the magnitude and share of rice CH4 emissions may mask the opportunity for mitigation via food systems.
To effectively reduce GHG emissions, climate policy should give higher priority to emission sources that have larger mitigation potential. Large emission sources would certainly receive more attention in this regard. Proper estimation or measurement of the magnitude and mitigation potential of emissions helps pave the foundation for planning and strategizing of GHG mitigation efforts.
To estimate the mitigation potential of rice CH4 emissions via spatial reallocation of cultivation areas across Taiwan, we set up a linear programming model with the objective function of rice CH4 minimization and the constraints that (1) aggregate output of paddy rice remains the same as in the base case and (2) cultivation areas do not exceed the historically attained levels in the subregion and cropping season. To calibrate the linear programming model, we need the following parameters and statistics for the subregions and cropping seasons: (1) per-hectare output, (2) per-hectare CH4 emissions, (3) rice cultivation areas, and (4) historically maximum areas of cultivation. We selected 2017 to conduct this ex-post analysis of cultivation area reallocation, as 2017 was a year free of current and lagged effects of weather anomalies such as droughts and typhoons.
To obtain the full data dimension as required by our linear programming, that is, per-hectare CH4 emission factors for all 18 rice-growing Taiwanese counties with each having two cropping seasons (except in Yilan), we extrapolated (Supplementary 1) based on the parameters we measured on-site from the 17 sites of the 9 counties (Fig. 1). The per-hectare CH4 emissions measurement information by county and by cropping season is next coupled with the county- and cropping-season-specific per-hectare rice yield of all 18 rice-growing counties, which is constantly surveyed by the agriculture ministry for administering market stabilization. Based on the resulting county- and cropping-season-specific parameters of CH4 emissions per ton of rice output, our linear programming model redistributes rice cultivation areas of the year 2017 among the two cropping seasons in the 18 rice-growing subregions (counties or cities).
The linear programming model for CH4-minimizing relocation of rice cultivation areas is set up as follows.
Minimize RE_TOT = Sum{r,REG, s,SSN, RE(r,s)}
Subject to:
RQ_TOT > = Sum{r,REG, s,SSN, RQ(r,s)}, and
RA(r,s) < = RA_MAX(r,s);
where
REG: the set of 18 rice-growing subregions (counties or cities) in Taiwan;
SSN: the set of spring and summer cropping seasons;
RE_TOT: sum of rice CH4 emissions (kilotons CO2-e) from all 18 rice-growing subregions in Taiwan;
RE(r,s): rice CH4 emissions (kilotons CO2-e) by subregion (r) and cropping season (s);
RQ_TOT: sum of rice output (tons, unhusked) from all 18 rice-growing subregions in Taiwan in the base year (2017);
RQ(r,s): rice output (tons, unhusked) by subregion and cropping season;
RA(r,s): rice cultivation area (hectare) by subregion and cropping season;
RA_MAX(r,s): maximum rice cultivation area (hectare) by subregion and cropping season between 2010 and 2020.
The model reallocates the rice cultivated area of 2017 based on the subregion-specific parameter of CH4 emissions per ton of rice output (i.e., emission intensity), which is calculated by dividing the CH4 emissions factor (i.e., CH4 emissions per hectare, of our measurements) by the crop yield (rice output per hectare, surveyed by the agriculture ministry). As the model processes to minimize national aggregate CH4 emissions from rice cultivation, subregions (counties or cities) with relatively lower emission intensities would be preferred to those with higher emission intensities and therefore receive priority in filling the cultivation area up to its maximum attainable levels (i.e., the upper bounds). We specified the upper bounds of available rice paddies for each subregion based on the harvested area data over the past decade (2010–2020). The historically maximum harvested area in a subregion/cropping season suggested its capacity for rice cultivation, and this was mainly shaped by the availability of irrigation water and climatic conditions.