Search

• Multi-objective optimization of rice production via Pareto fronts computed with NSGA-III. • Coupled with process model DNDC to calculate yield, water use, CH 4 and N 2 O emissions. • Increased rice yield is achievable with reduced irrigation and greenhouse gas emissions. Sustainable rice farming practices are urgently needed to meet increasing food demand, cope with water scarcity, and mitigate climate change. Traditional farming methods that prioritize a single objective, such as increasing crop yield, reducing irrigation water consumption, or minimizing greenhouse gas (GHG) emissions, have proven to be insufficient. Simultaneously optimizing multiple competing rice farming objectives remains less explored. The present study aimed to increase current rice yield, reduce irrigation water consumption, and tackle the dilemma to reduce GHG (CH 4 and N 2 O) emissions at once. Using a heuristic and holistic method, we optimized farm management, focusing on irrigation regimes, sowing window, fertilization rate, tillage depth, and their interactions. We calibrated and validated the process-based DeNitrification–DeComposition (DNDC) model with 5 years of eddy covariance observations in a rice paddy site. The DNDC model was integrated with a Pareto-based non-dominated sorting genetic algorithm (NSGA-III) to solve the multi-objective optimization problem. Results show that current farm management practices have failed to achieve the potential yield and imposed high environmental costs in the form of water use (604–810 mm/yr) and GHG emissions (CH 4 : 186–220 kg C/ha/yr; N 2 O: 0.3–1.6 kg C/ha/yr). By contrast, our new approach suggested that the optimized management could maintain or even increase current crop yield to its potential (∼10 t/ha), while reducing irrigation demand and GHG emissions by >50 %. Our results suggest that earlier sowing, combined with improved irrigation practices, may play a crucial role in maximizing crop yield and potentially providing environmental benefits. We found that the optimal proportion of non-flooded days was around 54 % of the growing season length, and the optimal stage for non-flooding days was the vegetative stage. Our approach could further be applied to evaluate the environmental sustainability of farming systems under various climate and local conditions, and to guide policymaking and farming practices. [Display omitted] [ABSTRACT FROM AUTHOR]

Ground reference data are an essential prerequisite for supervised crop mapping. The lack of a low-cost and efficient ground referencing method results in pervasively limited reference data and hinders crop classification. In this study, we apply a convolutional neural network (CNN) model to explore the efficacy of automatic ground truthing via Google Street View (GSV) images in two distinct farming regions: Illinois and the Central Valley in California. We demonstrate the feasibility and reliability of our new ground referencing technique by performing pixel-based crop mapping at the state level using the cloud-based Google Earth Engine platform. The mapping results are evaluated using the United States Department of Agriculture (USDA) crop data layer (CDL) products. From ~ 130,000 GSV images, the CNN model identified ~ 9,400 target crop images. These images are well classified into crop types, including alfalfa, almond, corn, cotton, grape, rice, soybean, and pistachio. The overall GSV image classification accuracy is 92% for the Central Valley and 97% for Illinois. Subsequently, we shifted the image geographical coordinates 2–3 times in a certain direction to produce 31,829 crop reference points: 17,358 in Illinois, and 14,471 in the Central Valley. Evaluation of the mapping results with CDL products revealed satisfactory coherence. GSV-derived mapping results capture the general pattern of crop type distributions for 2011–2019. The overall agreement between CDL products and our mapping results is indicated by R2 values of 0.44–0.99 for the Central Valley and 0.81–0.98 for Illinois. To show the applicational value of the proposed method in other countries, we further mapped rice paddy (2014–2018) in South Korea which yielded fairly well outcomes (R2 = 0.91). These results indicate that GSV images used with a deep learning model offer an efficient and cost-effective alternative method for ground referencing, in many regions of the world. [ABSTRACT FROM AUTHOR]

• Image fusion NDVI products were evaluated against in-situ spectral data. • Considerable differences between fusion methods were observed • Biases in Landsat and MODIS were propagated to the fusion products. • Spatial interpolation process led to biases in heterogeneous land covers. • Image fusion products agreed better in time than in space. Satellite image fusion methods that improve spatial and temporal resolution have significant potential to advance understanding of ecosystem dynamics in space and time. However, systematic evaluations of image fusion methods against in situ spectral data are lacking. Here, we used a suite of in situ spectral data collected at 60 elementary sampling units (10 × 10 m) covering 15 Landsat pixel (30 × 30 m) plots and one Moderate Resolution Imaging Spectroradiometer (MODIS) pixel (250 × 250 m) throughout the entire growing season in a heterogeneous rice paddy landscape to evaluate four state-of-the-art image fusion NDVI products. They include the Enhanced Spatial and Temporal Adaptive Reflectance Fusion Model (ESTARFM), Flexible Spatiotemporal DAta Fusion (FSDAF), SaTellite dAta IntegRation (STAIR), and the CubeSat Enabled Spatio-Temporal Enhancement Method (CESTEM); the former three blended Landsat and MODIS data, whereas the latter combined CubeSats, Landsat, and MODIS observations. All fusion products showed strong linear relationships against in situ data when combining all spatial and temporal observations (R2: 0.73 to 0.93) although there were partly negative biases (–1% to –9%). These biases resulted from forcing data to image fusion algorithms, such as Landsat (–4%) and MODIS (–7%). Performance difference between fusion methods were considerably larger for spatial than for temporal variation. Furthermore, Landsat NDVI explained only 17–22% of spatial variation against in situ spectral data, which can be translated into weak performance of image fusion products to predict spatial variability in NDVI. Image fusion products that relied on spatial interpolation showed large biases (–15% to –30%) for a vegetation plot surrounded by mixed land cover plots. Our results highlight key sources of uncertainty and will be instrumental in improving satellite image fusion methods to monitor land surface phenology in space and time. Image, graphical abstract [ABSTRACT FROM AUTHOR]

• Non-growing season ecosystem respiration negated the CO 2 sink during the growing season. • Carbon allocation to rice components varied substantially over the seasons. • The ratio of net to gross primary productivity varied seasonally and interannually. • Mid-season drainage substantially reduced CH 4 emission. • The proportion of ebullitive CH 4 flux was 10 to 17% of the total during the cultivation period. Rice is a major crop that feeds more than half of the global population. Rice paddies play a complex role in the carbon cycle by emitting CH 4 to the atmosphere while either sequestering or releasing CO 2 to the atmosphere. Here, we present 3.5 years of eddy covariance measurements of CO 2 and CH 4 fluxes over an intermittently irrigated, single crop rice paddy in South Korea in tandem with carbon stock measurements for leaves, stems, grains, and roots. The rice paddy acted as a slight sink or neutral in CO 2 (−47 ± 51 g C m −2 y −1; mean ± 95% confidence interval) with 941 ± 130 g C m −2 y −1 and 894 ± 108 g C m −2 y −1 as the gross primary production (GPP) and ecosystem respiration (Reco), respectively. The mean annual net primary productivity (NPP) was 666 ± 31 g C m −2 y −1, which was partitioned to leaves (9%), stems (30%), grains (51%), and roots (10%), for a harvest index of 57%. The paddy emitted 20.6 ± 1.5 g C m −2 y −1 in the form of CH 4 , of which between 82 and 96% was emitted during the cultivation period (~132 days). The CH 4 flux showed a mid-season decline caused by mid-season drainage, which produced aerobic soil conditions that inhibited CH 4 emissions. The proportion of estimated ebullitive flux to total CH 4 flux ranged from 10 to 17% during the cultivation period. GPP, Reco, and CH 4 fluxes varied by 12%, 11%, and 7%, respectively, over the 3-year period. When considering the harvest and the global warming potential of CH 4 , the paddy shifted from a greenhouse gas sink to a strong source (2,741 ± 468 g CO 2 eq m −2 y −1). These findings will improve our understanding of carbon budgets and cycles, and help develop greenhouse gas and water management strategies for rice paddies. [ABSTRACT FROM AUTHOR]

Statistical gross primary productivity (GPP) estimation from remote sensing observations has mostly been attempted on the basis of multispectral observations. To make full use of the information contained in vegetation spectra, however, hyperspectral observations should be used in combination with appropriate multivariate methods. Nevertheless, only very few previous studies attempted to estimate GPP directly from hyperspectral observations and did so on the basis of reflectance, with only a limited number of temporally discontinuous observations. In this study, we used long-term, continuous, half-hourly hyperspectral observations covering the visible and near-infrared spectral range to estimate GPP directly from upwelling irradiance using partial least square (PLS) regression in a rice paddy. To gain a better understanding of processes underlying the PLS estimation, we used extensive complementary field observations to run process-based simulations using the SCOPE model. We then applied PLS regression to the simulated hyperspectral data in the same way as for the observations and disentangled contributions related to relevant physiological processes, namely sun-induced chlorophyll fluorescence (SIF) and xanthophyll cycle-related spectral changes (XC). We found that upwelling hyperspectral irradiance in the visible and near-infrared spectral range predicted GPP better than reflectance. Furthermore, PLS-based GPP estimates outperformed both far-red SIF and widely used vegetation index-based methods. However, the most relevant information for the observation-based PLS-models was not clearly related to XC or SIF as the near-infrared spectral range showed comparable performance. Also, the simple average of upwelling irradiance over the 850–900 nm range outperformed the other non-multivariate approaches, including far-red SIF. These results held for the evaluation in terms of the seasonal variation of GPP, while there was apparently a small contribution of SIF and XC for the diurnal variation. The simulation-based analysis showed that SIF and XC contributed useful information to both GPP and photosynthetic light use efficiency (LUE) estimates at both seasonal and diurnal time scales. The strongest unique contribution from either SIF or XC, however, was to the diurnal variation of GPP and XC showed considerably better performance than SIF. We did not find improvements when combining the spectral regions of XC (500–570 nm) and SIF (650–800 nm) to estimate GPP. While SIF showed improvements when combined with the remaining spectral information excluding XC, this was not the case for XC. Our approach combines the strengths of process-based modeling with multivariate statistical analysis to improve our understanding of the usable information content in vegetation spectra and is highly relevant for further developing suitable methods for GPP estimation at large scales. • Hyperspectral GPP estimation outperforms SIF and other multispectral methods. • Upwelling radiation outperforms reflectance for multivariate GPP estimation. • Simulated xanthophyll cycle-related spectral changes very promising for GPP. • No improvements when combining SIF and xanthophyll cycle-related spectral changes. • Clear discrepancies between main mechanisms for observed and simulated results. [ABSTRACT FROM AUTHOR]

Rice is a major crop that feeds more than half of global population. Rice paddy playscomplex roles in carbon cycle by emitting CH4 to the atmosphere while CO2 is sequesteredfrom or released to the atmosphere. Here, we present 3.5 years of eddy covariancemeasurements of CH4 and CO2 fluxes over an intermittently-irrigated rice paddy in Korea intandem with carbon stock measurements in leaf, stem, grain, and root. The interannualvariation of CO2 fluxes was little although the timings of transplantation and harvest weredifferent for each year. Rice paddy acted as a slight sink or neutral in CO2 (-47.3 ± 50.9 g Cm−2 y−1; mean ± 95% confidence interval) with 942.2 ± 129.4 g C m−2 y−1 and894.9 ± 106.5 g C m−2 y−1 for gross primary production (GPP) and ecosystemrespiration (Reco), respectively. Carbon stock in rice showed large interannualvariation (760.2 ± 183.7 g C m−2) with 70 ± 19.8 g C m−2, 232.1 ± 83.3 g Cm−2, 378.2 ± 75.5 g C m−2, and 79.8 ± 12.8 g C m−2 for leaf, stem, grain, androot, respectively. The rice paddy emitted 21.7 ± 2.5 g C m−2 y−1 of CH4. Whenconsidering the global warming potential of CH4 (197.3 g CO2-Ceq m−2 y−1),the paddy shift from carbon sink to source (150 g CO2-Ceq m−2 y−1). The CH4flux showed a bimodal seasonal pattern caused by the mid-season drainage whichturned the soils aerobic conditions so prohibited CH4 emissions. The irrigationpractices, which are different every year, resulted in the significant interannualvariation of CH4 emission (± 7%). These findings will be useful to understand carbonbudgets and cycles, and develop carbon and water managements in rice paddy. [ABSTRACT FROM AUTHOR]

Sun-induced chlorophyll fluorescence (SIF) has recently become an important topic in large scale photosynthesis research. This is mainly related to the selectivity of SIF for vegetation on the one hand and the close mechanistic relationships to photosynthesis on the other hand. The latter, however, is already nonlinear at the leaf level and further complicated by the canopy scattering of far-red SIF. This scattering process, which is strongly affected by the structure of the canopy, results in only a fraction of SIF photons emitted at the leaf-level to escape the canopy. The challenge for canopy-level SIF research is therefore to disentangle the contributions related to canopy structure and leaf physiology. While it is widely assumed that the latter has the stronger relationship to photosynthesis, this has not been tested experimentally at the canopy scale so far. We used a recently developed approach to estimate the canopy escape probability of far-red SIF to analyze the contributions of canopy structure and leaf physiology separately. For this, we conducted a combined analysis for three in-situ datasets from a rice paddy, a wheat and a corn field that all had fPAR, PAR, far-red SIF, GPP and canopy reflectance observations covering most of the growing season. Our analysis focused on half-hourly observations both at the level of fluxes involving the absorbed radiation as well as at the level of light use efficiencies, which excludes the radiation component.We found that canopy structure effects dominated the seasonal component of the SIF escape probability and that this was the cause of a positive correlation of the latter to photosynthetic light use efficiency. The estimated total SIF yield of all leaves in the canopy, in contrast, was dominated by diurnal variations and was almost entirely uncorrelated to photosynthetic light use efficiency. Furthermore, our findings confirm the strong relationships between NIRv to both SIF and GPP at short time scales and the site level that were previously only reported at large spatial and long temporal scales. Thus, our findings bridge gaps between spatio-temoral scales and challenge the widely held assumption that the leaf physiology component in SIF provides most of the additional information for photosynthesis estimation that goes beyond APAR. We think that our approach has the potential to unify previous findings into a coherent picture that could enable the community to more easily identify the advantages and limitations of different methods of estimating GPP from remotely sensed observations. [ABSTRACT FROM AUTHOR]

• Energy/carbon fluxes at two heights were compared before changing the measurement height. • Random errors could explain substantial portions of flux differences. • Systematic errors also existed, primarily caused by footprint differences. • Annual differences were equivalent to 31–115% of mean year-to-year variations. • A correction method was proposed for reducing flux differences. During long-term flux monitoring, we occasionally encounter situations where the existing measurement setup must be changed. Considering that one of the main purposes of long-term flux monitoring is identifying long-term variability and trends, which are relatively small, it is important to understand or correct for flux differences caused by the measurement setting changes. In this study, before changing the flux measurement height of the CRK (Cheorwon Rice paddy, Korea) site from 10 m to 5 m, we installed another eddy covariance system at 5 m and compared the sensible and H 2 O/CO 2 /CH 4 fluxes at both heights from April 2020 to April 2021. Although random errors could explain substantial portions of the flux differences between the two heights, it was confirmed that systematic errors also existed because the means of the flux differences were not zero, and the distributions of the flux differences were also significantly skewed. Further analysis showed that the primary cause of these systematic errors was the footprint differences rather than the difference in turbulent transport between the two heights. Based on the results, we proposed a strategy for securing the continuity of the flux time-series, which is necessary to understand the long-term variability using all data before and after a measurement change, and discussed the necessity of such a comparative observation. [ABSTRACT FROM AUTHOR]