Next Article in Journal
Internet of Things Based Digital Twin Model Construction and Online Fault-Tolerant Control of Hypersonic Flight Vehicle
Previous Article in Journal
Using YOLOv5, SAHI, and GIS with Drone Mapping to Detect Giant Clams on the Great Barrier Reef
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Drones
Volume 8
Issue 9
10.3390/drones8090459
drones-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles thumb_up
...
Endorse textsms
...
Comment
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Estimating Methane Emissions in Rice Paddies at the Parcel Level Using Drone-Based Time Series Vegetation Indices

by
Yongho Song
1,
Cholho Song
2,
Sol-E Choi
3,
Joon Kim
2,
Moonil Kim
4,
Wonjae Hwang
2,
Minwoo Roh
1,
Sujong Lee
1 and
Woo-Kyun Lee
1,2,*
1
Department of Environmental Science and Ecological Engineering, Korea University, Seoul 02841, Republic of Korea
2
OJEong Resilience Institute (OJERI), Korea University, Seoul 02841, Republic of Korea
3
National Forest Satellite Information & Technology Center, National Institute of Forest Science, Seoul 05203, Republic of Korea
4
Division of ICT-Integrated Environment, Pyeongtaek University, 111 Yongyi-Dong, Pyeongtaek-si 17869, Republic of Korea
*
Author to whom correspondence should be addressed.
Drones 2024, 8(9), 459; https://doi.org/10.3390/drones8090459
Submission received: 7 July 2024 / Revised: 30 August 2024 / Accepted: 31 August 2024 / Published: 4 September 2024
(This article belongs to the Section Drones in Agriculture and Forestry)
Browse Figures
Versions Notes

Abstract

:
This study investigated a method for directly estimating methane emissions from rice paddy fields at the field level using drone-based time-series vegetation indices at a town scale. Drone optical and spectral images were captured approximately 15 times from April to November to acquire time-series vegetation indices and optical orthoimages. An empirical regression model validated in previous international studies was applied to calculate cumulative methane emissions throughout the rice cultivation process. Methane emissions were estimated using the vegetation index and yield data were used as input variables for each growth phase. Methane emissions from rice paddies showed maximum values of 309 kg CH4 ha−1, within a 7% range compared to similar studies, and minimum values of 138 kg CH4 ha−1, with differences ranging from 29% to 58%. The average emissions were calculated at 247 kg CH4/ha, revealing slightly lower average values but individual field values within a similar range. The results suggest that drone-based remote sensing technology is an efficient and cost-effective alternative to traditional field measurements for greenhouse gas emission assessments. However, adjustments and validations according to rice varieties and local cultivation environments are necessary. Overcoming these limitations can help establish sustainable agricultural management practices and achieve local greenhouse gas reduction targets.

1. Introduction

Rice cultivation is one of the most important agricultural industries and is also a significant source of greenhouse gas emissions, accounting for approximately 8% of the total global anthropogenic emissions from 2008 to 2017 [1]. Methane (CH4), emitted from paddy fields where rice is grown, is a potent greenhouse gas and has contributed approximately 0.5 °C to the rise in global temperatures from 2010 to 2019 compared to the period from 1850 to 1900 [2]. In 2021, the agricultural sector in South Korea emitted a total of 21.4 million tons of CO2-equivalent greenhouse gases, marking a 1.3% increase from 2018 and a 1.1% increase from the previous year. Among these, emissions from rice cultivation accounted for the highest proportion (27%) of the total agricultural sector emissions and amounted to 5.7 million tons of CO2eq [3].
Recognizing this global challenge, South Korea joined the Global Methane Pledge, committing to a reduction of at least 30% in methane emissions by 2030 compared with the 2020 levels [4]. With the agricultural sector accounting for 43% of domestic methane emissions, systematic management and reduction of greenhouse gases in agriculture are imperative [5,6]. Methane emissions from domestic paddy fields were calculated using the Tier 2 approach, which involves setting national greenhouse gas emission factors and using formulas that consider yield and growth period parameters [7,8]. Although this method considers various management characteristics and aims for accurate regional-scale estimates incorporating environmental factors, the direct measurement of greenhouse gas emissions using specialized observational equipment is expensive and time-consuming, especially for continuous tracking over large national areas [6,9,10].
Numerous studies have been conducted to estimate methane emissions from rice cultivation, employing either modeling approaches or direct measurements using specialized field equipment in study areas. Most studies have highlighted limitations such as data scarcity, model complexity, and the high costs and time requirements. The primary causes identified include the constraints of real-time monitoring, difficulties in large-area application, and the high cost of equipment [11,12,13].
To overcome these limitations, remote sensing methods that utilize various sensors have emerged as viable alternatives, providing results close to those of direct field measurements. These methods offer advantages in surveying extensive areas and analyzing locations that are difficult for humans to access and are used in a range of applications, including surveying, vegetation monitoring, time-series cover change detection, and crop yield prediction [14,15,16]. Drones offer flexibility in data acquisition timing outside adverse weather conditions, ease of operation, and access to ultra-high-resolution data, thereby making them suitable for various research areas such as crop monitoring, organic matter management investigation, and pest management [8,14,17,18]. With the increasing variety of sensors applicable to drones and the advancement of analytical technologies such as AI, drones are being utilized in various agricultural fields where full-scale surveys were previously impossible [19,20]. These applications include yield prediction, elucidating relationships within plant physiological processes, crop health assessment, and the estimation of greenhouse gas emissions [19,21].
Although research using drones to study methane emissions has been conducted in South Korea [8], these studies have primarily focused on exploring organic matter factors for methane calculation formulas or assessing the status of straw use and winter crop cultivation without directly quantifying methane emissions. The enactment of the Framework Act on Carbon Neutrality and Green Growth to Cope with Climate Crisis (“Carbon Neutrality Framework Act”) in December 2022 necessitates the establishment of basic plans for carbon neutrality and green growth by metropolitan and basic local governments for 2025–2034. In May 2023, the Ministry of Environment distributed revised guidelines for the establishment and progress check of local governments’ carbon neutrality and green growth basic plans, providing emission data and reference materials. However, ambiguities in the definitions and scope present challenges in their application at the local government level [22]. Accurate greenhouse gas emission estimation at the basic local government level is essential, especially as the AFOLU (Agriculture, Forestry, and Other Land Use) sector, among the five greenhouse gas inventory sectors, relies on national-level statistical data as activity data, making accurate local-level greenhouse gas inventory assessment difficult. In particular, the agricultural land category is a land use category with high variability in emissions and uptake due to human agricultural activities, requiring accurate GHG inventory calculations based on the period and fertilization of cultivation [6,7,23].
The purpose of this study is to develop a method for calculating methane emissions from individual rice paddies using drone-based time series vegetation indices and to verify its applicability.

2. Materials and Methods

2.1. Study Area

This study focused on densely populated regions with rice paddies, covering approximately 260 ha in Siu-ri, Namyangju city in Gyeonggi province (Figure 1). The Siu-ri area is situated at altitudes ranging from 107 m to 530 m and predominantly consists of mountainous terrain. Nonetheless, lower elevations near rivers are utilized by villages and agricultural lands. The geographical layout was such that the southern and eastern sections had lower elevations, whereas the northern and western areas had higher elevations. The tributaries flowing into the Bukhangang River in the southeast enhance water accessibility for agriculture. Historical climate data for the Siwoo-ri region, derived from the Yangpyeong Observatory—the closest station provided by the South Korea Meteorological Administration—over ten years from January 2013 to December 2022, revealed an average temperature of 12.4 °C, average monthly rainfall of 103 mm, and an annual precipitation total of 1232 mm.

2.2. Methods

To estimate methane emissions from rice cultivation using drones, the methodology outlined in Figure 2 was implemented. Time-series aerial photography with drones captured both optical and multispectral images. These orthorectified optical images were used to manually interpret temporal changes in ground cover, allowing for the delineation of actual rice-growing fields and the determination of harvest dates. The captured multispectral images were analyzed to calculate vegetation indices, and changes over time were assessed to establish planting dates. Rice growth stages were then classified based on these determined planting and harvest dates. Methane emissions were subsequently quantified applying the empirical regression formula derived by Shi et al. [24]. ESRI ArcGIS Pro 3.1 was utilized for spatial analyses, which included the use of drone time series data and the calculation of vegetation indices, while Microsoft Excel 365 Version 2407 was employed for calculating methane quantities and generating statistical outputs. Statistical analysis of the results was performed using Microsoft Excel 365 Version 2407 and IBM SPSS Statistics Version 26.0.

2.2.1. Drone Photography and Image Acquisition

For the acquisition of multispectral and optical imagery using drones, the P4 Multispectral (P4M) (DJI, Shenzhen, China) was employed (Table 1). This equipment facilitates the collection of spectral data in the Red, Green, Blue, RedEdge, near-infrared (NIR) wavelengths, and optical imagery.
Maintaining a constant altitude during drone flights can result in variations in the area covered by the data owing to changes in ground elevation, potentially affecting the uniformity of the orthoimage quality and causing errors during the image-matching process. To address these issues and ensure both flight stability and image data quality, we utilized the Flight Master System (FMS) Version 1.2.0. This automated flight path program, developed by FMworks, Inc. (Daegu, Republic of Korea), maintains a consistent altitude difference between the drone and the ground using a Digital Elevation Model (DEM) (Figure 3).
In total, fifteen flights were executed, adhering to a schedule of two flights per month with a fortnight interval between flights. However, owing to adverse weather conditions or mechanical issues with the drone, flights scheduled for mid-August and mid-September were missed. The flights were strategically timed for the afternoon hours, when the sun was the highest, intending to minimize shadowed areas as much as possible. All the flights maintained a consistent altitude above ground level (Table 2). All captured images were processed for orthorectification and brightness correction using DJI Terra Version 3.3.4 (DJI, Shenzhen, China) to produce orthophoto images. DJI Terra automatically corrects image deviations according to stored distortion parameter metadata information, including lens correction and reflectance correction [25]. The P4M is equipped with an irradiance sensor, used to normalize the DN of each band. Geometric corrections of the time series images were performed using landmarks and geographical features. Finally, images were captured at low altitudes where atmospheric effects are minimal, making atmospheric correction less important [26,27], so no additional atmospheric correction was performed.

2.2.2. Classification of Paddy Fields and Determination of Local Rice Growing Periods

To delineate the actual rice paddies within this study area, we first segregated areas classified as paddy fields on the LandCoverMap provided by the Ministry of Environment. Subsequently, we selected regions that exhibited a transition from green in the summer to yellow in the fall through visual interpretation of time-series optical imagery. Vegetation indices, which are ratio functions distinct from reflectance, are minimally influenced by weather conditions and are suitable for multi-temporal analysis [18,28,29,30]. Changes in vegetation indices that reflect the cover conditions were used to determine the transplanting and harvesting periods of rice, which are critical for segmenting rice growth stages. These indices show significant changes at certain times and are used as the basis for selection. Various vegetation indices employed in numerous studies for assessing the health and vitality of vegetation and ecological changes in land, such as the Normalized Difference Vegetation Index (NDVI), Normalized Difference Red Edge Index (NDRE), Green Normalized Difference Vegetation Index (GNDVI), and Optimized Soil-Adjusted Vegetation Index (OSAVI), were utilized. The cultivation cycle periods were calculated by analyzing the degree of change in these indices according to the growth stage.
NDVI = NIR − Red NIR + Red
GNDVI = NIR − Green NIR + Green
NDRE = NIR − RE NIR + RE
OSAVI = NIR − Red NIR + Red + 0.16

2.3. Collection of Rice Growth Information

To derive rice growth information for this study area, it is first necessary to identify the rice varieties that are specific to the region. The types of rice seeds distributed in this study area were verified using the seed supply status of regional agricultural cooperatives provided by the Korea Seed and Variety Service (KSVS) via their seed complaint service (www.seednet.go.kr, accessed on 28 August 2024). In Siwoo-ri, it was confirmed that in 2021, 140 kg of Samgwang was supplied. Subsequently, in 2022, the supply included 180 kg of Samgwang and 40 kg of Chamdream. Given its higher distribution rate, Samgwang was selected as the target cultivar for investigation. Information related to rice growth, such as transplanting, heading, and optimal harvesting times, was collected from the Samgwang rice cultivation manual (Table 3) [31].

2.4. Application of Regression Models for Calculating Cumulative Methane Emissions per Unit Area

Shi et al. [24] developed an empirical regression model to calculate the cumulative amount of methane emissions during the rice cultivation period using the rice yield and vegetation index in Nanjing, China, and confirmed its high accuracy (Table 4). The environmental conditions in which the model was formulated have a slightly different climatic condition compared to Siwoo-ri in Namyangju, with an average annual temperature about 3 °C higher and an annual precipitation difference of approximately 200 mm. The annual precipitation and average temperature in the two rice cultivation regions are relatively similar, and both regions have approximately 130 growing degree days (GDD) required for rice cultivation, indicating similar growth cycles and cultivation environments. Rice is known to be affected by variations in transplanting and harvesting times due to differences in temperature and precipitation, which in turn impact GDD [32,33]. Additionally, based on studies showing that vegetation indices during the growth period of different rice varieties exhibit similar trends under similar climate and GDD conditions [34], we assumed that the regression model developed in China could be applied to this study.
The methane emission model was structured as an empirical regression formula that incorporates methane emissions over the rice growth period collected through field observations. Hyperspectral reflectance, measured by a hand-held ASD portable spectrometer (FieldSpec-2), was transformed to match the Landsat 8 band range and used in the model. The P4M used in this study is equipped with a six-camera system that includes Blue, Green, Red, Red Edge, Near-Infrared bands, and an RGB camera (Table 1). The Landsat-8 Operational Land Imager (OLI) includes spectral bands such as Blue (450–515 nm), Green (525–600 nm), Red (630–680 nm), and Near-Infrared (845–885 nm) [35,36]. This indicates that the spectral bands of the P4M are similar to those of the Landsat-8. This model uses vegetation indices and the rice yield of the respective year as variables. The growth period was categorized into the jointing–booting stage (JS), heading stage (HS), grain-filling stage (GS), maturity stage (MS), and entire period (all stages, AS). The average R2 values for each vegetation index model were reported as NDVI-0.61, RVI-0.67, and EVI-0.8 [24]. Based on these findings, this study applied the four most accurate submodels that utilized the EVI as a parameter, considering that it had the highest average R2 value (Table 5). Rice production data were derived from the 2022 municipal-level statistics provided by the Statistics Office of the Republic of Korea, which recorded a yield of 5.59 tons per ha in Namyangju.
The reference model calculated the EVI using a portable ASD FieldSpec-2 (Analytical Spectral devices, Boulder, CO, USA) as a parameter and utilized Digital Number (DN) values in the range of 0–255, in contrast to the DN values of the drone-based spectral imagery in this study, which span from 0–65,535, thereby rendering direct application to the regression formula infeasible. Additionally, using the blue band to calculate the EVI, which serves as an empirical coefficient to reduce atmospheric effects, may have introduced errors in the values [37,38]. To resolve these issues and facilitate the application of the regression model in this study, the EVI was substituted with EVI2. EVI2 omits the Blue band used in EVI, relying solely on the NIR and R bands, and demonstrates values very close to EVI, within a margin of ±0.02 [37]. During the period this study was conducted, EVI2 values ranged from 0.2 to 2.1, which corresponds to a positive range. These values were normalized to a range of 0–1 for application in the methane emission calculation model.
EVI = G NIR − R NIR + C 1 R − C 2 B + L L = 1 , C 1 = 6 , C 2 = 7 . 5 ,   and G = 2 . 5
EVI 2 = 2.5 NIR − R NIR + 2.4 R + 1
The EVI2 values calculated for each growth stage had a resolution of 14.7 to 14.9 cm. To use the pixels directly in the methane quantity calculation and comparison across the models, all images were resampled to a resolution of 15 cm. The methane values calculated per pixel were aggregated by plot and adjusted for emissions per unit area for comparison with similar case studies.

3. Results

3.1. Transplanting and Harvesting Season Detection

Image analysis within this research area resulted in the classification of a total of 34 plots, covering an area of 7 ha. The areas of these plots ranged from a minimum of 0.03 ha to a maximum of 0.76 hectares, exhibiting a relatively small scale of diverse area distributions, with an average area of 0.2 ha.
To determine the actual transplanting dates within this study area, analysis using drone optical imagery and time-series vegetation indices identified a rapid increase in vegetation indices between 27 May 2022 and8 June 2022, pinpointing 30 May 2022 selected as the date (Figure 4).
The harvest period was set to17 October 2022, when vegetation indices began to show a gradual decrease and leveling off trend after the heading stage. Optical imagery confirmed pre- and post-harvest cover changes in most plots (Figure 5). Based on these data, calculating the difference between the transplanting and harvesting periods yielded a cultivation period of 140 days, similar to the growth period stated in the Samgwang rice cultivation manual. However, visual interpretation of the optical imagery revealed that while most paddy fields showed similar cover changes according to the growth stages, not all plots underwent transplanting and harvesting on the same dates, which indicates some variations.
Based on the determined transplanting and harvesting periods, the rice growth stages JS, HS, GS, and MS were derived with reference to the rice cultivation manual. This enabled the separation of EVI2, an input variable used in the methane emission model, by growth stage (Table 6). However, data collection for 15 August 2022, which corresponds to the heading stage, was halted owing to a drone malfunction. An average value, derived from the closest dates of 4 August 2022 and 31 August 2022, was subsequently calculated and applied.

3.2. Estimating Cumulative Methane Emissions within Individual Paddy Fields

3.2.1. Methane Emission Model Calculation Results

In the results derived from the calculations across each model, it was ascertained that methane emission calculations could be successfully performed for individual paddy fields, with observed variations within these fields reflecting the agricultural performance of rice, such as growth conditions (Figure 6).
The thematic maps produced from the analysis occasionally displayed negative values. Optical imagery analysis indicated that these negative values could be attributed to lower EVI2 values resulting from conditions such as reduced growth following transplanting, as depicted in Figure 7a, (Field 6), or soil exposure, as shown in Figure 7b, unlike in fields exhibiting healthy, dark green vegetation in Figure 7a, (Field 21). For the EVI2-HS-GS and EVI2-AS models, which reflect the values during the harvest period, negative values were observed in Fields 33 and 34 (Figure 8b). This occurrence was interpreted as a result of harvesting activities conducted earlier than the selected harvest date, leading to a transition in cover from vegetation to soil and a consequent decrease in EVI2 values (Figure 8).
The EVI2 results were converted from pixel values to calculate the methane emissions for each paddy field using the model (Figure 9). The results from the JS model were higher than those from other models, with methane values in the order of the JS-HS, HS-GS, and AS models. Although a similar trend was observed across most fields where methane emissions were estimated in proportion to area, some fields exhibited lower methane emissions relative to their area than others.
Considering the characteristics of methane emissions from rice growing in the anaerobic conditions of paddy fields, this discrepancy can be attributed to observed lower EVI2 values caused by land exposure due to various agricultural practices within the cultivation area or poor rice health during the cultivation process. These factors are thought to influence the calculation results, as determined through drone imaging. This aligns with the results of previous studies that demonstrated a correlation between methane emissions and vegetation indices during the rice cultivation process [37,38,39].

3.2.2. Methane Emissions Calculation per Unit Area

When calculating the methane emissions per hectare, the average for EVI2-JS was 247 kg CH4/ha, followed by EVI2-JS-HS, EVI2-HS-GS, and EVI2-AS (Table 7). A comparison of the median values revealed that EVI2-JS was higher than that of the other groups. The 75th percentile also shows that EVI2-JS possesses the highest values among the groups, indicating that the top 25% of the data in the EVI2-JS group exhibit higher methane emissions than the other groups, as depicted in Figure 10, suggesting overall higher methane emissions in comparison to other models. Approximately 41–55% of methane is emitted during the early stages of rice growth, decreasing to 3–8% in the later stages [40]. The higher methane emissions observed in the EVI2-JS model support other research findings, thereby suggesting that a higher proportion of methane is emitted during the early growth stages, specifically during the EVI2-JS period [41,42]. Additionally, it was observed that the inclusion of data from the latter stages of growth in the model tended to result in lower methane emissions, which was found to be consistent with the findings of prior research [43,44].

3.2.3. Comparison of Field-Measured Methane Flux Result

To verify the validity of the results of this study, we obtained and compared 2019 measurement data collected using the closed chamber method [45] at the Korea University Farm (KUF) in Namyangju (37°35′01″ N, 127°14′16″ E), located approximately 4 km from this study area within the same administrative region. The methane flux used for comparison was measured with a device installed to maintain a constant flow rate of air between the chamber and the detector, using a CH4 sensor (Axetris, Kaegiswil, Switzerland) [46]. The experiment was conducted from May to September 2019, coinciding with the transplanting and harvesting period of this study. The soil was collected from three different rice fields near the KU farm in Gyeonggi Province, and opaque acrylic cylinders were installed under continuous flooding conditions. The CO2 equivalent values (34) used in this paper were back calculated to be 14, 95, and 194 kg CH4/ha. These differences were found to be influenced by variations in electron acceptors that regulate the rate of redox processes in paddy soils.
A statistical analysis was performed to evaluate the accuracy by comparing the highest measured value of 194 kg/CH4 with the results from each model (Table 8). First, a Shapiro–Wilk test was conducted to assess normality. The results indicated that the EVI2-JS and AS models met the normality assumption, whereas the EVI2-JSHS and HSGS models did not. Consequently, a t-test was performed for the JS and AS models. Conversely, the Mann–Whitney U test was used for the JSHS and HSGS models. The analysis revealed significant differences between the measured values and the JS, AS, and JSHS models. In contrast, the HSGS model, with a p-value of 0.076, showed the closest similarity to the measured values as it was greater than 0.05.
All three methane flux values were found to be lower when compared to the model values from this study. The highest value of 194 kg CH4/ha was similar to the lower quartile of the regression model values, specifically in the order of JS-HS 99%, HS-GS 97%, and JS 95%. The reference model used in this study derives the regression equation for methane calculation using yield per unit area and vegetation index as parameters, resulting in specific coefficients and constants for these variables [23]. The discrepancy between the measured flux values and this study results is considered to be due to the difference in yield applied in the reference model (10.27 t/ha) and the average yield of Namyangju in this study area (5.59 t/ha).

3.2.4. Comparison of Methane Emission Calculation Results with Prior Research

A comparison was made with similar studies that calculate methane emissions during rice cultivation to validate study results. However, because of the absence of precise prior research cases matching the target area and period, the results from the DNDC (DeNitrification-DeComposition) model were further analyzed. The DNDC model is a representative biogeochemical model used in the agricultural sector for predicting greenhouse gas emissions and has been utilized in various studies to identify and forecast greenhouse gases generated during crop cultivation processes [47,48,49,50]. Based on Hwang et al. [47], the DNDC model was run for Siwoo-ri at a resolution of 1 km. Through this, methane emissions of 327–329 kg CH4/ha were identified.
In the National greenhouse gas Inventory Report (NIR) [3], the 2006 IPCC Guidelines were utilized to assess the rice cultivation in the agriculture sector. Utilizing national-specific emission factors and comprehensive national statistics, the methodology for estimating methane emissions was applied [3,5]. As a result, the emissions attributed to the rice cultivation sector were quantified as 5698 thousand metric tons of CO2 equivalent for the year 2020. Applying a methane-to-carbon global warming potential factor of 34 to the cultivation area of 778 thousand ha, the methane emissions were estimated to be 349 kg CH4/ha (Table 9).
Kim et al. [51] conducted a study over three years, measuring methane emissions using chambers installed for different rice varieties in the northern, central, and southern regions of South Korea. Among these, the experiment in the Hwaseong-si of Gyeonggi-do, located in the northern region, calculated methane emissions by multiplying the growth period by the daily methane emission values for the Samgwang variety of rice. The results showed methane emissions of 334 kg CH4/ha, 270 kg CH4/ha, and 290 kg CH4/ha in the first, second, and third years, respectively, with an average of 298 kg CH4/ha.
Jang et al. [53] utilized drones to classify areas of rice straw fertilizer return and winter crop cultivation using the 2019 IPCC Refinement’s calculation method to estimate methane emissions. In Anseong, Yeoju, Yongin, Incheon, Pyeongtaek, and Hwaseong in the Gyeonggi Province, methane emissions ranged from a minimum of 195 kg CH4/ha to a maximum of 300 kg CH4/ha, with an average of 235 kg CH4/ha.
Choi et al. [52] utilized the 2010 National Census of Agriculture, Forestry, and Fisheries data from Statistics Korea and a water management correction factor calculation program to estimate methane emissions from paddy fields across all administrative districts in the country using the IPCC estimation formula. The values obtained for the municipalities in Gyeonggi Province showed methane emissions ranging from a minimum of 490 kg CH4/ha to a maximum of 560 kg CH4/ha, with an average of 520 kg CH4/ha. The provincial-level average value for Gyeonggi Province was calculated to be 452 kg CH4/ha.
The results derived from this study, particularly from the EVI2-JS model, indicated methane emissions ranging from a minimum of 138 kg CH4/ha to a maximum of 309 kg CH4/ha, with an average emission rate of 247 kg CH4/ha (Table 7). Although these results tend to be somewhat lower than those of previous studies, they are similar to the range reported by Jang et al. [53]. This similarity suggests that predicting methane emissions through spectral imagery corresponding to the JS stage is feasible and highlights its potential as a significant indicator for accurately measuring methane emissions during rice cultivation.

4. Discussion

4.1. Result of Discussion

Drone photography from April to November collected and analyzed time-series spectral and optical imagery, enabling the identification of field-specific growth cycles and temporal changes in vigor throughout the rice cultivation process.
Comparative analysis of the temporal changes in NDVI, NDRE, GNDVI, and OSAVI was conducted to detect the critical phenological stages of rice cultivation. The sharp increase in vegetation indices observed in April and May, following a decrease due to initial flooding of the fields, was designated as the onset of the planting phase. The peak of these indices, followed by a subsequent decline, indicated the heading phase. The period after heading, when the vegetation indices decreased to reach soil-level values, was identified as the harvesting phase. These stages were corroborated by visual inspections from orthorectified optical images and are consistent with the growth stages documented in the Samkwang rice cultivation manual. Furthermore, the patterns of vegetation index fluctuations conformed to changes reported in rice growth stages by related studies, thereby supporting the validity of the methodological approach [54,55,56,57,58]. Methane emissions from rice paddies are influenced by the growth stages of rice. As the rice grows, methane emissions increase. After the heading stage, methane emissions tend to decrease due to the reduction in photosynthesis and the decrease in organic matter necessary for methane production [42,59,60]. In this study, we applied an empirical regression model to calculate methane emissions using rice yield and vegetation indices for each growth stage as parameters. The parameters used in this model include growth stages that distinguish rice development, vegetation index values reflecting growth vigor, and rice yield as the final output of growth, thus incorporating the influence of crop growth on methane emission calculations.
The utilization of high-resolution EVI2 indices via drones has proven capable of replacing traditional EVI and has demonstrated the feasibility of precise, field-level estimation of methane emissions based on the crop condition within the fields. Furthermore, the methane emissions calculated per field aligned with the ranges found in prior studies, indirectly validating the approach and results of this research.
To validate these study results, we compared them with the methane flux results of Hwang et al. [46], which were conducted at close range and in similar regions and environmental conditions. As a result, while some methane values were within a similar range to the methane values predicted by the regression model, some very low values were observed depending on the soil type. These results suggest that methane emissions can vary depending on the characteristics of paddy soil and indicate that future development of independent regression equations for calculating methane emissions in rice farming should include variables such as water management and soil properties. Furthermore, it was confirmed that the coefficients and constants of the methane measurement model are determined by the magnitude of variables such as yield and vegetation index, emphasizing the need to develop regression equations tailored to the characteristics of the rice cultivation area.
When methane emissions per unit area calculated were compared with results from the DNDC model, NIR [3], and the studies by Kim et al. [51] and Jang et al. [53], the maximum differences observed ranged from −3% to 7%, while the minimum differences varied from 29% to 58%. The maximum values, falling within a 7% range, confirmed the similarity of the emission calculations. However, significant differences were noted in the minimum values. These discrepancies are attributed to the use of nationally accredited baseline emission factors derived under continuous flooding conditions, which were used by previous research that employed field measurement equipment in limited areas.
For result comparison, similar regional studies on methane emissions from rice paddies, excluding the DNDC model, have used the 2006 IPCC GL methane estimation formula to calculate methane emissions. This formula calculates annual methane emissions by multiplying the daily methane emission per unit area by the rice cultivation period and the cultivated area [61]. Within this framework, correction factors for water management practices, organic matter management practices, soil, and rice varieties, which influence methane emissions, are applied. However, due to the lack of statistical data on pre-cultivation water management, soil, and rice varieties in the domestic context, these factors are not applied. Available correction factors include the basic emission factor, water management correction factor during the rice cultivation period, and organic matter correction factor. All these factors are calculated based on samples from rice cultivation areas. However, there is a shortage of correction factors for various conditions that reduce methane emissions, such as changes in the growth environment leading to decreased vigor or the exposure of soil within the plots eliminating anaerobic soil conditions [3,51,53].
Additionally, these studies often relied on regional averages that did not adequately reflect variations in water regimes or soil conditions. According to related studies [62,63,64,65,66], methane emissions increase with more vigorous rice growth under the anaerobic soil conditions typical in continuously flooded fields. The findings of this study are consistent with these results, suggesting that fields managed under conditions close to continuous flooding tend to produce methane emissions per unit area that are comparable to the highest observed values.
Jang et al. [53], which had the most similar values to this study, differed from other studies by using drone imagery to estimate the extent of organic matter usage on a plot-by-plot basis. They distinguished between fields with rice straw incorporation into the soil and those without and applied this differentiation regionally to adjust the correction factors for calculating methane emissions. Within this study, they compared the traditional emission calculation method with the method using differentiated correction factors, finding that methane emissions could nearly double even within the same area of the same region using the traditional method [53]. These results indicate that significant regional variations in methane emissions can occur even within the statistical zones currently used, depending on local environmental conditions and cultivation practices.
Choi et al. [52], such as other comparative papers, used the methodology proposed in the 2006 IPCC Guidelines, applying national baseline emission factors and national statistics, but found that the organic matter application correction factors used and the number of tillage days varied by rice variety, exacerbated regional disparities. Furthermore, comparisons with field measurements of methane flux revealed significant regional variations, with discrepancies in some areas suggesting that the impact factors for organic matter usage may be set too high and may need to be reevaluated [52]. These differences highlight the limitations of Tier 2 level sampling used for methane calculation and suggest a need for more precise reflection of environmental changes occurring during the rice cultivation process.

4.2. Limitation and Future Study

During the process of acquiring images using drones, errors in altitude or positional accuracy can occur due to inaccuracies in the positioning sensors [21]. Additionally, unexpected wind influences, weather conditions, and shadows from ground objects can potentially cause errors in drone imaging [67]. Positional inaccuracies and spatial errors in the images can be mitigated by using sensors with high spatial and spectral resolution and by employing Real-Time Kinematic (RTK) GPS systems to enhance data accuracy. Furthermore, real-time monitoring of weather conditions and sun positions during drone flights can further improve data quality through calibration and validation.
The reference model applied for methane calculation using time-series vegetation indices was conducted in Nanjing, China. This introduces a limitation, as specific environmental differences between this study area and Nanjing were not reflected in the regression model. These environmental differences influence the rice growth process, affecting the calculation of methane emissions from paddy fields during rice cultivation. Additionally, the sensors and measurement methods used in the reference model’s calculation process differ from those used in this study’s time-series vegetation index acquisition using drones, which could lead to discrepancies in vegetation index values between the two methods. Therefore, future research needs to accumulate data on methane emission estimates under various conditions linked to the local rice cultivation environment and practices. Furthermore, standardized measurement methods should be applied to minimize the potential for errors between different sensors.
In order to adapt the drone-based methane emission estimation method for rice cultivation to practical national statistical levels, it is necessary to make additional adjustments for various factors that influence emission calculations. These factors include the environmental conditions, rice varieties, cultivation methods, and water management practices within domestic rice-growing regions. For this purpose, long-term data collection on the environmental characteristics and yields of individual fields during the regional rice cultivation process is essential, along with the analysis of methane emission variability. In addition, the regional characteristics of the model will need to be reflected through comparison with field measurements of different environments and rice vigor conditions.
To validate the reliability of this study’s drone-based methane emission estimation method, a comparison with data from specialized field observation equipment is necessary. However, the cost, timing, and availability of methane detection equipment posed limitations to conducting this research. In South Korea, flux tower data for methane emissions are provided only in specific areas, and the installation sites are not near this study region, making the use of these data limited [66]. To address this, this study indirectly validated the results by comparing them with studies that performed field measurements and national statistical methods in paddy fields in the Gyeonggi Province. However, there is a limitation in that the cumulative methane emissions related to the methane flux measurement data from the days matching the drone shooting dates and locations were not collected. For the future development of methane emission models using drones, it is essential to secure field methane measurement data under various environmental conditions and vitality levels of rice growth. Additionally, it is necessary to investigate various control groups regarding cultivation characteristics per plot, such as rice yield, water management practices, and fertilizer usage, and to accumulate time-series vegetation index data for the same period. Through this, the development of region-specific independent methane emission models should be achieved.
In this study, due to the lack of field validation at the same time and place, we used results from different regions. Using results from other regions without validation to estimate rice paddy methane emissions has limitations, as it does not account for differences in local rice cultivation environments, even if they appear similar. Specifically, our study could not reflect the differences in methane emissions from rice paddies due to soil properties, even if they have similar weather conditions and rice growth periods.
To overcome this, first standards for measurement sensors and survey methods need to be established. It has been confirmed that sensors used in remote sensing surveys can produce different results even within the same spectral range, depending on the target, scope, and type of sensor used. These findings indicate that, in developing regression equations using vegetation indices, different outcomes may arise even when measuring spectral values for the same rice crop. To minimize such errors, it is essential to establish standardized survey methods. Secondly, sufficient field observation data must be collected to account for the various cultivation environments and practices. The amount of methane emitted from rice paddies is influenced by various factors such as weather, soil, water management, and growth stages. It is essential to obtain many field measurement samples of methane emissions under different conditions and utilize vegetation indices that can reflect changes in these variables. This approach will enable the development of methane emission estimation models tailored to each specific environmental condition. Lastly, conducting research that involves drone imaging at the same time as field observations can potentially overcome these limitations. To achieve this, field measurements of methane using specialized observation equipment should be carried out simultaneously with drone imaging at the target sites. This simultaneous approach allows for the calibration and validation of methane emission models, thereby ensuring the accuracy and validity of the results.
As climate change progresses, changes in regional growth temperatures and weather conditions, such as rainfall, will affect rice growth and methane emissions. By anticipating changes in transplanting and harvesting times based on these factors and selecting rice varieties suitable for changing growth environments, it will be possible to implement smart agricultural planning to adapt to global warming.

4.3. Implication

When comparing the drone-based method conducted in this study with previous studies on estimating methane emissions from rice paddies, the estimation techniques can be broadly categorized into three approaches: calculations using estimation equations that reflect agricultural practices and national methane emission factors presented by the IPCC guidelines, field measurement using observation equipment, and flux towers. The calculation method using estimation equations has the advantage of applying country-specific methane emission factors and separating coefficients based on agricultural practices and fertilization through statistical surveys, providing reliability and scalability. However, it has the disadvantage of not being able to finely account for the limitations of statistical surveys and variations in local environments and agricultural practices [3]. The field measurement equipment and flux tower observation method ensure accuracy and reliability, but they have limitations in terms of equipment use complexity, cost-effectiveness, and regional scalability. The drone-based estimation method offers advantages in scalability and operational ease but has limitations in accuracy and reliability. However, it is believed that these limitations can be overcome in the future as the collection of methane observation data linked with drone data accumulates and systematic validation is carried out.
This study has demonstrated that advanced remote sensing technology using drones can be a highly useful tool for measuring and managing greenhouse gas emissions in the agricultural sector. Initially, it was possible to overcome the limitations of satellite and aerial photography in distinguishing small-scale cultivated rice fields. This capability is expected to address the shortcomings of current methodologies that use rice yield samples and administrative area statistics for rice production, thereby contributing to the production of more accurate statistical data. Furthermore, the use of time-series vegetation indices enabled the identification of rice growth stages and health, which not only facilitates the development of more precise methane calculation models but also enhances the accuracy of greenhouse gas emission estimates. If the methods and regression models from this study are directly applied to other regions, there may be differences in methane emission results due to variations in local environmental conditions, and errors may occur; however, a consistent trend is expected to be identifiable. These findings suggest that, with the development of correction factors based on rice cultivation environments and the application of region-specific methane regression models, it would be possible to apply this approach to other regions. Additionally if advancements in satellite sensors and image correction technologies enable the acquisition of high-resolution, high-quality time-series imagery and regional methane emission regression models can be developed and applied using this data, it would be possible to expand this research to estimate methane emissions from rice paddies at the national scale using satellite imagery. Lastly, it is anticipated that these findings will provide an objective basis for future research and policy development aimed at sustainable agricultural practices and strategies for reducing greenhouse gas emissions in response to global warming challenges.

5. Conclusions

This study introduces a method for directly calculating methane emissions from paddy fields on a per-field basis using drone time-series vegetation indices, thereby deriving methane emissions per unit area and examining the potential of drone-based remote sensing technology for greenhouse gas emission measurements.
Through time-series changes in vegetation indices using drones, this study classified the growth stages and periods of rice from transplanting at the end of May to harvest. Methane emissions from rice paddies ranged from 309 kg CH4 ha−1 to 138 kg CH4 ha−1, with an average of 247 kg CH4 ha−1. Within the paddy fields, healthier rice growth resulted in higher methane emissions, while poor growth or increased soil exposure led to reduced methane emissions. The results were found to be within a range similar to those of previous studies conducted in Gyeonggi Province.
These findings validate the effectiveness of drone-based remote sensing technology for methane emission measurements, thereby offering a cost- and time-efficient alternative to traditional field measurement methods. This research approach could be particularly useful for estimating greenhouse gas emissions in regions with limited manpower, providing crucial information for local governments to set and manage greenhouse gas reduction targets. Furthermore, the calculation of methane emissions at the field level using drones can serve as essential foundational data for the formulation and implementation of policies aimed at reducing greenhouse gas emissions by local authorities. This is anticipated to contribute to advancing sustainable agricultural management and refining Tier 3-level greenhouse gas inventory calculations.

Author Contributions

Conceptualization, Y.S. and W.-K.L.; methodology, Y.S., C.S. and M.K.; software, Y.S.; validation, C.S., S.-E.C., M.K., W.H. and W.-K.L.; formal analysis, Y.S.; investigation, Y.S., J.K., M.R. and S.L.; resources, Y.S., J.K., M.R. and S.L.; data curation, Y.S.; writing—original draft preparation, Y.S.; writing—review and editing, Y.S., C.S. and S.-E.C.; supervision, W.-K.L.; project administration, W.-K.L.; funding acquisition, W.-K.L. All authors have read and agreed to the published version of the manuscript.

Funding

This research was supported by the National Research Foundation (NRF) of Korea Grant funded by the Korea government Ministry of Science and ICT (MSIT) (NRF-2021K1A3A1A78097879) and the OJEong Resilience Institute (OJERI) at Korea University as the Core Research Institute of the Basic Science Research Program funded by the Ministry of Education (NRF-2021R1A6A1A10045235) and Technology Development Project for Creation and Management of Ecosystem based Carbon Sinks (RS-2023-00218243) through KEITI, Ministry of Environment.

Data Availability Statement

Data will be made available on request by email.

Conflicts of Interest

The authors declare no conflicts of interest.

References

  1. Saunois, M.; Stavert, A.R.; Poulter, B.; Bousquet, P.; Canadell, J.G.; Jackson, R.B.; Raymond, P.A.; Dlugokencky, E.J.; Houweling, S.; Patra, P.K.; et al. The Global Methane Budget 2000–2017. Earth Syst. Sci. Data 2020, 12, 1561–1623. [Google Scholar] [CrossRef]
  2. Intergovernmental Panel on Climate Change (IPCC). Climate Change 2021: The Physical Science Basis; IPCC: Geneva, Switzerland, 2021. [Google Scholar]
  3. Greenhouse Gas Inventory and Research Center (GIR). National Greenhouse Gas Inventory Report 2021; Greenhouse Gas Inventory and Research Center: Sejong, Republic of Korea, 2021. [Google Scholar]
  4. European Commission and United States of America. The Participants in the Global Methane Pledge; European Commission and United States of America: Paris, France, 2021. [Google Scholar]
  5. Greenhouse Gas Inventory and Research Center (GIR). National Greenhouse Gas Inventory Report 2019; Greenhouse Gas Inventory and Research Center: Sejong, Republic of Korea, 2019. [Google Scholar]
  6. Gwon, H.S.; Choi, E.J.; Lee, S.I.; Lee, H.S.; Lee, J.M.; Kang, S.S. Research Review of Methane Emissions from Korean Rice Paddies. J. Clim. Chang. Res. 2022, 13, 117–134. [Google Scholar] [CrossRef]
  7. Jin, J.-H.; Jeong, H.-C.; Lee, S.-I.; Lee, H.-S.; Park, H.-R.; Yu, Y.-S.; Lee, J.-M.; Lee, Y.-H.; Gwon, H.-S. Life Cycle Assessment of Greenhouse Gas Emission of Rice Cultivation under Minimum Tillage in the Gimje, South Korea. Korean J. Soil Sci. Fertil. 2023, 56, 300–312. [Google Scholar] [CrossRef]
  8. Park, J.; Jang, S.; Kim, H.; Hong, R.; Song, I. Application of Drone Images to Investigate Biomass Management Practices and Estimation of CH4 Emissions from Paddy Fields. J. Korean Soc. Agric. Eng. 2020, 62, 39–49. [Google Scholar] [CrossRef]
  9. Gwon, H.S.; Khan, M.I.; Yoon, Y.E.; Lee, Y.B.; Kim, P.J.; Hwang, H.Y. Unexpected Higher Decomposition of Soil Organic Matter during Cold Fallow Season in Temperate Rice Paddy. Soil Tillage Res. 2019, 192, 250–257. [Google Scholar] [CrossRef]
  10. Nikolaisen, M.; Cornulier, T.; Hillier, J.; Smith, P.; Albanito, F.; Nayak, D. Methane Emissions from Rice Paddies Globally: A Quantitative Statistical Review of Controlling Variables and Modelling of Emission Factors. J. Clean. Prod. 2023, 409, 137245. [Google Scholar] [CrossRef]
  11. Zhang, W.; Sun, W.; Li, T. Uncertainties in the National Inventory of Methane Emissions from Rice Cultivation: Field Measurements and Modeling Approaches. Biogeosciences 2017, 14, 163–176. [Google Scholar] [CrossRef]
  12. Ouyang, Z.; Jackson, R.B.; McNicol, G.; Fluet-Chouinard, E.; Runkle, B.R.K.; Papale, D.; Knox, S.H.; Cooley, S.; Delwiche, K.B.; Feron, S.; et al. Paddy Rice Methane Emissions across Monsoon Asia. Remote Sens. Environ. 2023, 284, 113335. [Google Scholar] [CrossRef]
  13. Qi, Z.; Guan, S.; Zhang, Z.; Du, S.; Li, S.; Xu, D. Effect and Mechanism of Root Characteristics of Different Rice Varieties on Methane Emissions. Agronomy 2024, 14, 595. [Google Scholar] [CrossRef]
  14. Na, S.; Hong, S.; Ahn, H.; Park, C.; So, K.; Lee, K. Detrending Crop Yield Data for Improving MODIS NDVI and Meteorological Data Based Rice Yield Estimation Model. Korean J. Remote Sens. 2021, 37, 199–209. [Google Scholar]
  15. Kim, Y.; Hong, S. Estimation of Rice Grain Protein Contents Using Ground Optical Remote Sensors. Korean J. Remote Sens. 2008, 24, 551–558. [Google Scholar]
  16. Lee, K.; Hong, S.-Y.; Kim, S.-G.; Park, C.-W.; Ahn, H.-Y.; Na, S.; So, K.-H. Estimation of Rice Cultivation Area by Threshold Method Using Sentinel-1 Imagery in South Korea. Korean J. Soil Sci. Fertil. 2020, 53, 345–354. [Google Scholar] [CrossRef]
  17. Ryu, J.; Ahn, H.; Na, S.; Lee, B.; Lee, K. Evaluation of NDVI Retrieved from Sentinel-2 and Landsat-8 Satellites Using Drone Imagery under Rice Disease. Korean J. Remote Sens. 2022, 38, 1231–1244. [Google Scholar] [CrossRef]
  18. Ahn, H.; Ryu, J.; Na, S.; Lee, B.; Kim, M.; Lee, K. Comparative Analysis of Pre-Processing Method for Standardization of Multi-Spectral Drone Images. Korean J. Remote Sens. 2022, 38, 1219–1230. [Google Scholar]
  19. Guo, Y.; Xiao, Y.; Hao, F.; Zhang, X.; Chen, J.; de Beurs, K.; He, Y.; Fu, Y.H. Comparison of Different Machine Learning Algorithms for Predicting Maize Grain Yield Using UAV-Based Hyperspectral Images. Int. J. Appl. Earth Obs. Geoinf. 2023, 124, 103528. [Google Scholar] [CrossRef]
  20. Guo, Y.; Chen, S.; Li, X.; Cunha, M.; Jayavelu, S.; Cammarano, D.; Fu, Y.H. Machine Learning-Based Approaches for Predicting SPAD Values of Maize Using Multi-Spectral Images. Remote Sens. 2022, 14, 1337. [Google Scholar] [CrossRef]
  21. Sun, Z.; Wang, X.; Wang, Z.; Yang, L.; Xie, Y.; Huang, Y. UAVs as Remote Sensing Platforms in Plant Ecology: Review of Applications and Challenges. J. Plant Ecol. 2021, 14, 1003–1023. [Google Scholar] [CrossRef]
  22. Greenhouse Gas Inventory and Research Center (GIR). Climate Change and Carbon Neutrality; Greenhouse Gas Inventory and Research Center: Sejong, Republic of Korea, 2023. [Google Scholar]
  23. Choi, E.; Kim, G.; Lee, S.; Lee, J.; Yoon, J.; Gwon, H. Estimation of Greenhouse Gases Emission from the Crop Cultivation Session of Agricultural Sector According to 2019 Refinement to the 2006 IPCC Guidelines. J. Clim. Chang. Res. 2020, 11, 703–712. [Google Scholar] [CrossRef]
  24. Shi, Y.; Lou, Y.; Zhang, Z.; Ma, L.; Ojara, M.A. Estimation of Methane Emissions Based on Crop Yield and Remote Sensing Data in a Paddy Field. Greenh. Gases Sci. Technol. 2020, 10, 196–207. [Google Scholar] [CrossRef]
  25. Ge, W.; Li, X.; Jing, L.; Han, J.; Wang, F. Monitoring Canopy-Scale Autumn Leaf Phenology at Fine-Scale Using Unmanned Aerial Vehicle (UAV) Photography. Agric. For. Meteorol. 2023, 332, 109372. [Google Scholar] [CrossRef]
  26. Jakob, S.; Zimmermann, R.; Gloaguen, R. The Need for Accurate Geometric and Radiometric Corrections of Drone-Borne Hyperspectral Data for Mineral Exploration: MEPHySTo-A Toolbox for Pre-Processing Drone-Borne Hyperspectral Data. Remote Sens. 2017, 9, 88. [Google Scholar] [CrossRef]
  27. Jeon, E.I.; Kim, K.; Cho, S.; Kim, S. A Comparative Study of Absolute Radiometric Correction Methods for Drone-Borne Hyperspectral Imagery. Korean J. Remote Sens. 2019, 35, 203–215. [Google Scholar] [CrossRef]
  28. Deng, L.; Mao, Z.; Li, X.; Hu, Z.; Duan, F.; Yan, Y. UAV-Based Multispectral Remote Sensing for Precision Agriculture: A Comparison between Different Cameras. ISPRS J. Photogramm. Remote Sens. 2018, 146, 124–136. [Google Scholar] [CrossRef]
  29. Cao, S.; Danielson, B.; Clare, S.; Koenig, S.; Campos-Vargas, C.; Sanchez-Azofeifa, A. Radiometric Calibration Assessments for UAS-Borne Multispectral Cameras: Laboratory and Field Protocols. ISPRS J. Photogramm. Remote Sens. 2019, 149, 132–145. [Google Scholar] [CrossRef]
  30. Fawcett, D.; Panigada, C.; Tagliabue, G.; Boschetti, M.; Celesti, M.; Evdokimov, A.; Biriukova, K.; Colombo, R.; Miglietta, F.; Rascher, U.; et al. Multi-Scale Evaluation of Drone-Based Multispectral Surface Reflectance and Vegetation Indices in Operational Conditions. Remote Sens. 2020, 12, 514. [Google Scholar] [CrossRef]
  31. Chungcheongnam-do Agricultural Technology Institute. Manual for Growing “Samgwangbyeo”; Chungcheongnam-do Agricultural Technology Institute: Cheonan-si, Republic of Korea, 2012. [Google Scholar]
  32. Hu, X.; Huang, Y.; Sun, W.; Yu, L. Shifts in Cultivar and Planting Date Have Regulated Rice Growth Duration under Climate Warming in China since the Early 1980s. Agric. For. Meteorol. 2017, 247, 34–41. [Google Scholar] [CrossRef]
  33. Ryu, J.H.; Jeong, H.; Cho, J. Performances of Vegetation Indices on Paddy Rice at Elevated Air Temperature, Heat Stress, and Herbicide Damage. Remote Sens. 2020, 12, 2654. [Google Scholar] [CrossRef]
  34. Zhang, K.; Ge, X.; Shen, P.; Li, W.; Liu, X.; Cao, Q.; Zhu, Y.; Cao, W.; Tian, Y. Predicting Rice Grain Yield Based on Dynamic Changes in Vegetation Indexes during Early to Mid-Growth Stages. Remote Sens. 2019, 11, 387. [Google Scholar] [CrossRef]
  35. Roy, D.P.; Wulder, M.A.; Loveland, T.R.; Woodcock, C.E.; Allen, R.G.; Anderson, M.C.; Helder, D.; Irons, J.R.; Johnson, D.M.; Kennedy, R.; et al. Landsat-8: Science and Product Vision for Terrestrial Global Change Research. Remote Sens. Environ. 2014, 145, 154–172. [Google Scholar] [CrossRef]
  36. Omari, K.; Abuelgasim, A.; Alhebsi, K. Aerosol Optical Depth Retrieval over the City of Abu Dhabi, United Arab Emirates (UAE) Using Landsat-8 OLI Images. Atmos. Pollut. Res. 2019, 10, 1075–1083. [Google Scholar] [CrossRef]
  37. Yan, K.; Gao, S.; Chi, H.; Qi, J.; Song, W.; Tong, Y.; Mu, X.; Yan, G. Evaluation of the Vegetation-Index-Based Dimidiate Pixel Model for Fractional Vegetation Cover Estimation. IEEE Trans. Geosci. Remote Sens. 2022, 60, 1–14. [Google Scholar] [CrossRef]
  38. Jiang, Z.; Huete, A.R.; Didan, K.; Miura, T. Development of a Two-Band Enhanced Vegetation Index without a Blue Band. Remote Sens. Environ. 2008, 112, 3833–3845. [Google Scholar] [CrossRef]
  39. Zhen, Z.; Yunsheng, L.; Rui, L.; Li, M.; Jun, L.; Amina, L. The Potential of Vegetation Indices: Estimating Methane Emissions in Paddy Fields under Nighttime Warming and Water Saving Conditions. Greenh. Gases Sci. Technol. 2020, 10, 545–556. [Google Scholar] [CrossRef]
  40. Hayashida, S.; Ono, A.; Yoshizaki, S.; Frankenberg, C.; Takeuchi, W.; Yan, X. Methane Concentrations over Monsoon Asia as Observed by SCIAMACHY: Signals of Methane Emission from Rice Cultivation. Remote Sens. Environ. 2013, 139, 246–256. [Google Scholar] [CrossRef]
  41. Schijtz, H.; Seiler, W.; Conrad, R. Processes Involved in Formation and Emission of Methane in Rice Paddies. Biogeochemistry 1989, 7, 33–53. [Google Scholar]
  42. Alberto, M.C.R.; Wassmann, R.; Buresh, R.J.; Quilty, J.R.; Correa, T.Q.; Sandro, J.M.; Centeno, C.A.R. Measuring Methane Flux from Irrigated Rice Fields by Eddy Covariance Method Using Open-Path Gas Analyzer. Field Crops Res. 2014, 160, 12–21. [Google Scholar] [CrossRef]
  43. Li, D.; Liu, M.; Cheng, Y.; Wang, D.; Qin, J.; Jiao, J.; Li, H.; Hu, F. Methane Emissions from Double-Rice Cropping System under Conventional and No Tillage in Southeast China. Soil Tillage Res. 2011, 113, 77–81. [Google Scholar] [CrossRef]
  44. Inubushi, K.; Muramatsu, Y.; Umebayashi, M. Influence of Percolation on Methane Emission from Flooded Paddy Soil. Jpn. J. Soil Sci. Plant Nutr. 1992, 63, 184–189. [Google Scholar] [CrossRef]
  45. Meijide, A.; Manca, G.; Goded, I.; Magliulo, V.; Di Tommasi, P.; Seufert, G.; Cescatti, A. Seasonal Trends and Environmental Controls of Methane Emissions in a Rice Paddy Field in Northern Italy. Biogeosciences 2011, 8, 3809–3821. [Google Scholar] [CrossRef]
  46. Hwang, W.; Park, M.; Cho, K.; Hyun, S. Drainage Practice of Rice Paddies as a Sustainable Agronomic Management for Mitigating the Emission of Two Carbon-Based Greenhouse Gases (CO2 and CH4): Field Pilot Study in South Korea. Sustainability 2024, 16, 2802. [Google Scholar] [CrossRef]
  47. Hwang, W.; Kim, C.; Cho, K.; Hyun, S. Characteristics of Greenhouse Gas Emissions from Rice Paddy Fields in South Korea under Climate Change Scenario RCP-8.5 Using the DNDC Model. Pedosphere 2021, 31, 332–341. [Google Scholar] [CrossRef]
  48. Fumoto, T.; Kobayashi, K.; Li, C.; Yagi, K.; Hasegawa, T. Revising a Process-Based Biogeochemistry Model (DNDC) to Simulate Methane Emission from Rice Paddy Fields under Various Residue Management and Fertilizer Regimes. Glob. Chang. Biol. 2008, 14, 382–402. [Google Scholar] [CrossRef]
  49. Min, H.; Kim, M.; Kim, J.; Hwang, W. DNDC Modeling for Greenhouse Gases Emission in Rice Paddy of South Korea and the Effect of Flooding Management Change and RCP 8.5 Scenario. Ecol. Resilient Infrastruct. 2018, 5, 189–198. [Google Scholar] [CrossRef]
  50. Shin, M.H.; Jang, J.R.; Won, C.H.; Jung, Y.H.; Lee, S.I.; Lim, K.; Choi, J.D. Simulation of GHG Emission from Paddy Field Using DNDC Model. J. Korean Soc. Agric. Eng. 2014, 56, 47–57. [Google Scholar] [CrossRef]
  51. Kim, G.-Y.; Jeong, H.-C.; Ju, O.-J.; Kim, H.-K.; Park, J.-H.; Gwon, H.-S.; Kim, P.-J. Establishment of Baseline Emission Factor of Methane in Korean Rice Paddy Soil. Korean J. Environ. Agric. 2013, 32, 359–365. [Google Scholar] [CrossRef]
  52. Choi, S.W.; Kim, J.; Kang, M.S.; Kang, N.G. Estimation and Mapping of Methane Emissions from Rice Paddies in Korea: Analysis of Regional Differences and Characteristics. Korean J. Agric. For. Meteorol. 2018, 20, 88–100. [Google Scholar] [CrossRef]
  53. Jang, S.J.; Park, J.S.; Hong, R.G.; Hong, J.P.; Kwon, C.L.; Song, I.H. Estimation of Paddy CH4 Emissions through Drone-Image-Based Identification of Paddy Rice Straw Application & Winter Crop Cultivation. J. Korean Soc. Rural Plan. 2021, 27, 21–33. [Google Scholar] [CrossRef]
  54. Zhang, G.; Xiao, X.; Dong, J.; Kou, W.; Jin, C.; Qin, Y.; Zhou, Y.; Wang, J.; Menarguez, M.A.; Biradar, C. Mapping Paddy Rice Planting Areas through Time Series Analysis of MODIS Land Surface Temperature and Vegetation Index Data. ISPRS J. Photogramm. Remote Sens. 2015, 106, 157–171. [Google Scholar] [CrossRef]
  55. Qiu, B.; Li, W.; Tang, Z.; Chen, C.; Qi, W. Mapping Paddy Rice Areas Based on Vegetation Phenology and Surface Moisture Conditions. Ecol. Indic. 2015, 56, 79–86. [Google Scholar] [CrossRef]
  56. Yeom, J.M.; Jeong, S.; Deo, R.C.; Ko, J. Mapping Rice Area and Yield in Northeastern Asia by Incorporating a Crop Model with Dense Vegetation Index Profiles from a Geostationary Satellite. GIScience Remote Sens. 2021, 58, 1–27. [Google Scholar] [CrossRef]
  57. Lee, K.; Park, C.-W.; So, K.-H.; Na, S. Selection of Optimal Vegetation Indices and Regression Model for Estimation of Rice Growth Using UAV Aerial Images. Korean J. Soil Sci. Fertil. 2017, 50, 409–421. [Google Scholar] [CrossRef]
  58. Li, B.; Peng, S.; Shen, R.; Yang, Z.L.; Yan, X.; Li, X.; Li, R.; Li, C.; Zhang, G. Development of a New Index for Automated Mapping of Ratoon Rice Areas Using Time-Series Normalized Difference Vegetation Index Imagery. Pedosphere 2022, 32, 576–587. [Google Scholar] [CrossRef]
  59. Miyata, A.; Leuning, R.; Denmead, O.T.; Kim, J.; Harazono, Y. Carbon Dioxide and Methane Fluxes from an Intermittently Flooded Paddy Field; Elsevier: Amsterdam, The Netherlands, 2000; Volume 102. [Google Scholar]
  60. Singh, S.; Kashyap, A.K.; Singh, J.S. Methane Flux in Relation to Growth and Phenology of a High Yielding Rice Variety as Affected by Fertilization; Springer: Berlin/Heidelberg, Germany, 1998; Volume 201. [Google Scholar]
  61. Buendia, C.; Kranjc, K.; Fukuda, J.; Osako, N.S.; Shermanau, Y.; Federici, P. 2019 Refinement to the 2006 IPCC Guidelines for National Greenhouse Gas Inventories Volume 1 General Guidance and Reporting Task Force on National Greenhouse Gas Inventories; IPCC: Geneva, Switzerland, 2019. [Google Scholar]
  62. Yang, S.S.; Chang, H.L. Diurnal Variation of Methane Emission from Paddy Fields at Different Growth Stages of Rice Cultivation in Taiwan. Agric. Ecosyst. Environ. 1999, 76, 75–84. [Google Scholar] [CrossRef]
  63. Neue, H.-U. Methane Emission from Rice Fields. Bioscience 1993, 43, 466–474. [Google Scholar] [CrossRef]
  64. Yan, X.; Yagi, K.; Akiyama, H.; Akimoto, H. Statistical Analysis of the Major Variables Controlling Methane Emission from Rice Fields. Glob. Chang. Biol. 2005, 11, 1131–1141. [Google Scholar] [CrossRef]
  65. Wang, Z.; Gu, D.; Beebout, S.S.; Zhang, H.; Liu, L.; Yang, J.; Zhang, J. Effect of Irrigation Regime on Grain Yield, Water Productivity, and Methane Emissions in Dry Direct-Seeded Rice Grown in Raised Beds with Wheat Straw Incorporation. Crop J. 2018, 6, 495–508. [Google Scholar] [CrossRef]
  66. Kwun, S.K.; Shin, Y.K.; Eom, K. Estimation of Methane Emission from Rice Cultivation in Korea. J. Environ. Sci. Health—Part A Toxic/Hazard. Subst. Environ. Eng. 2003, 38, 2549–2563. [Google Scholar] [CrossRef]
  67. Guo, Y.; Fu, Y.H.; Chen, S.; Robin Bryant, C.; Li, X.; Senthilnath, J.; Sun, H.; Wang, S.; Wu, Z.; de Beurs, K. Integrating Spectral and Textural Information for Identifying the Tasseling Date of Summer Maize Using UAV Based RGB Images. Int. J. Appl. Earth Obs. Geoinf. 2021, 102, 102435. [Google Scholar] [CrossRef]
Drones 08 00459 g001
Figure 1. Map of the Siu-ri in Gyeonggi-do, South Korea, showing the location of this study area. Red areas indicate the rice paddies in the study field.
Figure 1. Map of the Siu-ri in Gyeonggi-do, South Korea, showing the location of this study area. Red areas indicate the rice paddies in the study field.
Drones 08 00459 g001
Drones 08 00459 g002
Figure 2. Flowchart of overall methodology.
Figure 2. Flowchart of overall methodology.
Drones 08 00459 g002
Drones 08 00459 g003
Figure 3. DEM equidistant flight method used by FMS.
Figure 3. DEM equidistant flight method used by FMS.
Drones 08 00459 g003
Drones 08 00459 g004
Figure 4. Temporal dynamics of parcel level average vegetation index. (a) NDVI, (b) GNDVI, (c) NDRE, (d) OSAVI. # is the drone flight number in Table 2.
Figure 4. Temporal dynamics of parcel level average vegetation index. (a) NDVI, (b) GNDVI, (c) NDRE, (d) OSAVI. # is the drone flight number in Table 2.
Drones 08 00459 g004
Drones 08 00459 g005aDrones 08 00459 g005b
Figure 5. Utilizing temporal orthoimage for determining harvest day. The red areas indicate the paddy field boundaries, and the numbers represent the indices of the paddy fields. (a) presents images taken shortly before harvest, showing that most areas are about to be harvested, except for fields 33 and 34, which have already been harvested. (b) displays images taken after the harvest, indicating that harvesting has been completed in most areas, except for a few.
Figure 5. Utilizing temporal orthoimage for determining harvest day. The red areas indicate the paddy field boundaries, and the numbers represent the indices of the paddy fields. (a) presents images taken shortly before harvest, showing that most areas are about to be harvested, except for fields 33 and 34, which have already been harvested. (b) displays images taken after the harvest, indicating that harvesting has been completed in most areas, except for a few.
Drones 08 00459 g005aDrones 08 00459 g005b
Drones 08 00459 g006
Figure 6. Thematic map of methane emissions by rice paddy field (a) EVI2-JS, (b) EVI2-JS-HS, (c) EVI2-HS-GS, (d) EVI2-AS (Unit: g CH4/pixel).
Figure 6. Thematic map of methane emissions by rice paddy field (a) EVI2-JS, (b) EVI2-JS-HS, (c) EVI2-HS-GS, (d) EVI2-AS (Unit: g CH4/pixel).
Drones 08 00459 g006
Drones 08 00459 g007
Figure 7. Identification of abnormal methane emission regions using drone optical imaging-based methane emission model results.
Figure 7. Identification of abnormal methane emission regions using drone optical imaging-based methane emission model results.
Drones 08 00459 g007
Drones 08 00459 g008
Figure 8. Time series EVI2 value comparison by paddy field. (a) Fields showing similar trends were indicated with different colors using points and lines. Fields 33 and 34 showed a rapid decline in EVI2 values after the heading stage compared to other plots. (b) Optical image verification indicated that early harvesting was conducted.
Figure 8. Time series EVI2 value comparison by paddy field. (a) Fields showing similar trends were indicated with different colors using points and lines. Fields 33 and 34 showed a rapid decline in EVI2 values after the heading stage compared to other plots. (b) Optical image verification indicated that early harvesting was conducted.
Drones 08 00459 g008
Drones 08 00459 g009
Figure 9. Methane emission estimation by rice paddy field area.
Figure 9. Methane emission estimation by rice paddy field area.
Drones 08 00459 g009
Drones 08 00459 g010
Figure 10. Box plot for methane emissions per unit area by models.
Figure 10. Box plot for methane emissions per unit area by models.
Drones 08 00459 g010
Table 1. Multispectral camera specification.
Table 1. Multispectral camera specification.
Drone & SensorP4 Multispectral
Spectral BandsBlue: 450 nm ± 16 nm, Green: 560 nm ± 16 nm, Red: 650 nm ± 16 nm, Red Edge: 730 nm ± 16 nm, Near-Infrared: 840 nm ± 26 nm
FOV62.7°
GSD(* H/18.9) cm/pixel
** Hover Accuracy RangeVertical: ±0.5 m, Horizontal: ±1.5 m
* H: Aircraft altitude relative to the area mapped (unit: meter); ** RTK disabled and GNSS positioning.
Table 2. Drone flight specification.
Table 2. Drone flight specification.
No.DateWeatherTimeAltitude
(m)
StartEnd
16 April 2022Sunny14:2716:30210
218 April 2022Sunny14:0817:13210
34 May 2022Sunny14:2216:51210
417 May 2022Sunny12:4315:05210
527 May 2022Partly Cloudy12:1314:36210
68 June 2022Partly Cloudy13:4216:04210
716 June 2022Cloudy13:3116:26210
81 July 2022Partly Cloudy12:5215:03210
915 July 2022Sunny13:4016:06210
104 August 2022Partly Cloudy12:5516:58210
1131 August 2022Cloudy12:0213:50210
1229 September 2022Sunny14:0716:11210
1317 October 2022Sunny13:4216:14210
142 November 2022Sunny12:5814:57210
1516 November 2022Sunny13:5715:49210
Table 3. Rice growth manual information.
Table 3. Rice growth manual information.
CategoryProperties
CultivarsSamgwangbyeo (Mid-late Maturation)
Growing regionsPlans of the central region, south-mid region
Growing periodAbout 130 days
Transplanting dateThe latter half of May (27 May–5 June)
Heading dateThe middle of August (Around 16 August)
Harvesting seasonThe early part of October (Around 10 October)
Table 4. Reference study site environmental information.
Table 4. Reference study site environmental information.
CategoryProperties
Experimental siteNanjing University, Nanjing, Jiangsu Province, Eastern China (32°12′ N, 118°43′ E)
Climate statusAverage annual precipitation: 1000 mm
Average annual temperature: 15.6 °C
CultivarsNanjing 46
Growing season 133 d (Transplant: 17 June 2017/Harvest: 28 October 2017)
Table 5. Various models for estimating paddy cumulative methane emissions [24].
Table 5. Various models for estimating paddy cumulative methane emissions [24].
ModelX2EquationR2
1EVI-JS CCE = − 3.84 X 1 + 744.18 X 2 − 358.12 0.89
2EVI-JS-HS CCE = − 1.1 X 1 + 574.2 X 2 − 259.5 0.80
3EVI-HS-GS CCE = − 0.27 X 1 + 646.47 X 2 − 304.96 0.83
4EVI-AS CCE = − 2.57 X 1 + 748.96 X 2 − 356.49 0.88
X1 = rice yield, X2 = phenological stage average vegetation index.
Table 6. Separating rice growing seasons and drone EVI2 data.
Table 6. Separating rice growing seasons and drone EVI2 data.
Transplanting DayJSHSGSMS
Nanjing 467 May14 July–
20 August		21 August–
3 September 		4 September–
6 October		7 October–
28 October
Samgwang-byeo31 May15 July–
5 August		About
16 August		17 August–
5 October		About
10 October
Drone EVI231 May15 July–
4 August		15 August31 August–
29 September		17 October
Table 7. Methane emissions per unit area by rice field methane calculation models (Unit: kgCH4/ha).
Table 7. Methane emissions per unit area by rice field methane calculation models (Unit: kgCH4/ha).
EVI2-JSEVI2-JS-HSEVI2-HS-GSEVI2-AS
Maximum309262254213
Average247223205175
Minimum138136129101
Upper quartile284247229198
Median252227206173
Lower quartile205192189152
Confidence
intervals 95%		±17.68±12.94±13.61±11.95
Table 8. Methane emissions per unit area by rice field methane calculation models.
Table 8. Methane emissions per unit area by rice field methane calculation models.
Statistical MethodEVI2-JSEVI2-JS-HSEVI2-HS-GSEVI2-AS
Shapiro–Wilk testW0.9520.9170.9360.95
p-value0.1410.0130.0470.124
Independent Samples t-testT-statistic5.75--−3.08
p-value2.50 × 10−7--0.003
Mann–Whitney U testU-statistic-850714-
p-value-0.000370.076-
Table 9. Comparison of methane emission results per unit area.
Table 9. Comparison of methane emission results per unit area.
NameData YearCH4 Emissions
per ha
(kg CH4 ha−1)		RegionLocation
Model EVI2-JS2022138–309Gyeonggi-doNamyangju
(Siu)
DNDC model2022327–329Gyeonggi-do
NIR *
[3]		2020215National-
Kim et al.
[51]		2010–2012270–334Gyeonggi-doHwaseong
Choi et al.
[52]		2010490–560Gyeonggi-doUijeongbu, Uiwang, Icheon,
Yeoju, Yangju, Yongin,
Gimpo, Osan, Gwacheon
Jang et al.
[53]		2018, 2019195–300Gyeonggi-doAnseong, Yeoju, Yongin,
Icheon, Pyeongtaek, Hwaseong
Hwang et al.
[46]		201914, 95, 194Gyeonggi-doNamyangju
(Wabu)
* NIR; National Greenhouse Gas Inventory Report of Korea.
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Song, Y.; Song, C.; Choi, S.-E.; Kim, J.; Kim, M.; Hwang, W.; Roh, M.; Lee, S.; Lee, W.-K. Estimating Methane Emissions in Rice Paddies at the Parcel Level Using Drone-Based Time Series Vegetation Indices. Drones 2024, 8, 459. https://doi.org/10.3390/drones8090459

AMA Style

Song Y, Song C, Choi S-E, Kim J, Kim M, Hwang W, Roh M, Lee S, Lee W-K. Estimating Methane Emissions in Rice Paddies at the Parcel Level Using Drone-Based Time Series Vegetation Indices. Drones. 2024; 8(9):459. https://doi.org/10.3390/drones8090459

Chicago/Turabian Style

Song, Yongho, Cholho Song, Sol-E Choi, Joon Kim, Moonil Kim, Wonjae Hwang, Minwoo Roh, Sujong Lee, and Woo-Kyun Lee. 2024. "Estimating Methane Emissions in Rice Paddies at the Parcel Level Using Drone-Based Time Series Vegetation Indices" Drones 8, no. 9: 459. https://doi.org/10.3390/drones8090459

Article Metrics

Back to TopTop