Uncertainty in phosphorus fluxes and budgets across the U.S. long-term agroecosystem research network

Phosphorus (P) budgets can be useful tools for understanding nutrient cycling and quantifying the effectiveness of nutrient management planning and policies; however, uncertainties in agricultural nutrient budgets are not often quantitatively assessed. The objective of this study was to evaluate uncertainty in P fluxes (fertilizer/manure application, atmospheric deposition, irrigation, crop removal, surface runoff, leachate) and the propagation of these uncertainties to annual P budgets. Data from 56 cropping systems in the P-FLUX database, which spans diverse rotations and landscapes across the U.S. and Canada, were evaluated. Results showed that across cropping systems, average annual P budget was 22.4 kg P ha^-1 (range = -32.7 to 340.6 kg P ha^-1 ), with an average uncertainty of 13.1 kg P ha^-1 (range = 1.0 to 87.1 kg P ha^-1 ). Fertilizer/manure application and crop removal were the largest P fluxes across cropping systems and, as a result, accounted for the largest fraction of uncertainty in annual budgets (61 and 37%, respectively). Remaining fluxes individually accounted for <2% of the budget uncertainty. Uncertainties were large enough that determining whether P was increasing, decreasing, or not changing was inconclusive in 39% of the budgets evaluated. Findings indicate that more careful and/or direct measurements of inputs, outputs, and stocks are needed. Recommendations for minimizing uncertainty in P budgets based on the results of the study were developed. Quantifying, communicating, and constraining uncertainty in budgets among production systems and multiple geographies is critical for engaging stakeholders, developing local and national strategies for P reduction, and informing policy. This article is protected by copyright. All rights reserved.

Multiyear methane and nitrous oxide emissions in different irrigation management under long-term continuous rice rotation in Arkansas

Rice paddies are one of the major sources of anthropogenic methane (CH₄ ) emissions. The alternate wetting and drying (AWD) irrigation management has been shown to reduce CH₄ emissions and total global warming potential (GWP) (CH₄ and nitrous oxide [N₂ O]). However, there is limited information about utilizing AWD management to reduce greenhouse gas (GHG) emissions from commercial-scale continuous rice fields. This study was conducted for five consecutive growing seasons (2015-2019) on a pair of adjacent fields in a commercial farm in Arkansas under long-term continuous rice rotation irrigated with either continuously flooded (CF) or AWD conditions. The cumulative CH₄ emissions in the growing season across the two fields and 5 years ranged from 41 to 123 kg CH₄ -C ha^-1 for CF and 1 to 73 kg CH₄ -C ha^-1 for AWD. On average, AWD reduced CH₄ emissions by 73% relative to CH₄ emissions in CF fields. Compared to N₂ O emissions, CH₄ emissions dominated the GWP with an average contribution of 91% in both irrigation treatments. There was no significant variation in grain yield (7.3-11.9 Mg ha^-1 ) or growing season N₂ O emissions (-0.02 to 0.51 kg N₂ O-N ha^-1 ) between the irrigation treatments. The yield-scaled GWP was 368 and 173 kg CO₂ eq. Mg^-1 season^-1 for CF and AWD, respectively, showing the feasibility of AWD on a commercial farm to reduce the total GHG emissions while sustaining grain yield. Seasonal variations of GHG emissions observed within fields showed total GHG emissions were predominantly influenced by weather (precipitation) and crop and irrigation management. The influence of air temperature and floodwater heights on GHG emissions had high degree of variability among years and fields. These findings demonstrate that the use of multiyear GHG emission datasets could better capture variability of GHG emissions associated with rice production and could improve field verification of GHG emission models and scaling factors for commercial rice farms.

Detecting Intra-Field Variation in Rice Yield With Unmanned Aerial Vehicle Imagery and Deep Learning

Unmanned aerial vehicles (UAVs) equipped with multispectral sensors offer high spatial and temporal resolution imagery for monitoring crop stress at early stages of development. Analysis of UAV-derived data with advanced machine learning models could improve real-time management in agricultural systems, but guidance for this integration is currently limited. Here we compare two deep learning-based strategies for early warning detection of crop stress, using multitemporal imagery throughout the growing season to predict field-scale yield in irrigated rice in eastern Arkansas. Both deep learning strategies showed improvements upon traditional statistical learning approaches including linear regression and gradient boosted decision trees. First, we explicitly accounted for variation across developmental stages using a 3D convolutional neural network (CNN) architecture that captures both spatial and temporal dimensions of UAV images from multiple time points throughout one growing season. 3D-CNNs achieved low prediction error on the test set, with a Root Mean Squared Error (RMSE) of 8.8% of the mean yield. For the second strategy, a 2D-CNN, we considered only spatial relationships among pixels for image features acquired during a single flyover. 2D-CNNs trained on images from a single day were most accurate when images were taken during booting stage or later, with RMSE ranging from 7.4 to 8.2% of the mean yield. A primary benefit of convolutional autoencoder-like models (based on analyses of prediction maps and feature importance) is the spatial denoising effect that corrects yield predictions for individual pixels based on the values of vegetation index and thermal features for nearby pixels. Our results highlight the promise of convolutional autoencoders for UAV-based yield prediction in rice.

Substantial hysteresis in emergent temperature sensitivity of global wetland CH4 emissions

Wetland methane (CH4) emissions ([Formula: see text]) are important in global carbon budgets and climate change assessments. Currently, [Formula: see text] projections rely on prescribed static temperature sensitivity that varies among biogeochemical models. Meta-analyses have proposed a consistent [Formula: see text] temperature dependence across spatial scales for use in models; however, site-level studies demonstrate that [Formula: see text] are often controlled by factors beyond temperature. Here, we evaluate the relationship between [Formula: see text] and temperature using observations from the FLUXNET-CH4 database. Measurements collected across the globe show substantial seasonal hysteresis between [Formula: see text] and temperature, suggesting larger [Formula: see text] sensitivity to temperature later in the frost-free season (about 77% of site-years). Results derived from a machine-learning model and several regression models highlight the importance of representing the large spatial and temporal variability within site-years and ecosystem types. Mechanistic advancements in biogeochemical model parameterization and detailed measurements in factors modulating CH4 production are thus needed to improve global CH4 budget assessments.

Rice Inundation Assessment Using Polarimetric UAVSAR Data

Irrigated rice requires intense water management under typical agronomic practices. Cost effective tools to improve the efficiency and assessment of water use is a key need for industry and resource managers to scale ecosystem services. In this research we advance model-based decomposition and machine learning to map inundated rice using time-series polarimetric, L-band Uninhabited Aerial Vehicle Synthetic Aperture Radar (UAVSAR) observations. Simultaneous ground truth observations recorded water depth inundation during the 2019 crop season using instrumented fields across the study site in Arkansas, USA. A three-component model-based decomposition generated metrics representing surface-, double bounce-, and volume-scattering along with a shape factor, randomness factor, and the Radar Vegetation Index (RVI). These physically meaningful metrics characterized crop inundation status independent of growth stage including under dense canopy cover. Machine learning (ML) comparisons employed Random Forest (RF) using the UAVSAR derived parameters to identify cropland inundation status across the region. Outcomes show that RVI, proportion of the double-bounce within total scattering, and the relative comparison between the double-bounce and the volume scattering have moderate to strong mechanistic ability to identify rice inundation status with Overall Accuracy (OA) achieving 75%. The use of relative ratios further helped mitigate the impacts of far range incidence angles. The RF approach, which requires training data, achieved a higher OA and Kappa of 88% and 71%, respectively, when leveraging multiple SAR parameters. Thus, the combination of physical characterization and ML provides a powerful approach to retrieving cropland inundation under the canopy. The growth of polarimetric L-band availability should enhance cropland inundation metrics beyond open water that are required for tracking water quantity at field scale over large areas.

Methane Emission Reductions from the Alternate Wetting and Drying of Rice Fields Detected Using the Eddy Covariance Method

Rice cultivation contributes 11% of the global 308 Tg CH₄ anthropogenic emissions. The alternate wetting and drying (AWD) irrigation practice can conserve water while reducing CH₄ emissions through the deliberate, periodic introduction of aerobic soil conditions. This paper is the first to measure the impact of AWD on rice field CH₄ emissions using the eddy covariance (EC) method. This method provides continuous, direct observations over a larger footprint than in previous chamber-based approaches. Seasonal CH₄ emissions from a pair of adjacent, production-sized rice fields under delayed flood (DF) and AWD irrigation were compared from 2015 to 2017. Across the 2 fields and 3 years, cumulative CH₄ emissions in the production season were in the range of 7.1 to 31.7 kg CH₄-C ha^-1 for the AWD treatment and in the range of 75.7-141.6 kg CH₄-C ha^-1 for the DF treatments. Correcting for field-to-field differences in CH₄ production, the AWD practice reduced seasonal CH₄ emissions by 64.5 ± 2.5%. The AWD practice is increasingly implemented for water conservation in the mid-south region of the United States; however, based on this study, it also has great potential for reducing CH₄ emissions.

Greenhouse Gas Emissions and Management Practices that Affect Emissions in US Rice Systems

Previous reviews have quantified factors affecting greenhouse gas (GHG) emissions from Asian rice ( L.) systems, but not from rice systems typical for the United States, which often vary considerably particularly in practices (i.e., water and carbon management) that affect emissions. Using meta-analytic and regression approaches, existing data from the United States were examined to quantify GHG emissions and major practices affecting emissions. Due to different production practices, major rice production regions were defined as the mid-South (Arkansas, Texas, Louisiana, Mississippi, and Missouri) and California, with emissions being evaluated separately. Average growing season CH emissions for the mid-South and California were 194 (95% confidence interval [CI] = 129-260) and 218 kg CH ha season (95% CI = 153-284), respectively. Growing season NO emissions were similar between regions (0.14 kg NO ha season). Ratoon cropping (allowing an additional harvestable crop to grow from stubble after the initial harvest), common along the Gulf Coast of the mid-South, had average CH emissions of 540 kg CH ha season (95% CI = 465-614). Water and residue management practices such as alternate wetting and drying, and stand establishment method (water vs. dry seeding), and the amount of residue from the previous crop had the largest effect on growing season CH emissions. However, soil texture, sulfate additions, and cultivar selection also affected growing season CH emissions. This analysis can be used for the development of tools to estimate and mitigate GHG emissions from US rice systems and other similarly mechanized systems in temperate regions.

Water quality of surface runoff and lint yield in cotton under furrow irrigation in Northeast Arkansas

Use of furrow irrigation in row crop production is a common practice through much of the Midsouth US and yet, nutrients can be transported off-site through surface runoff. A field study with cotton (Gossypium hirsutum, L.) was conducted to understand the impact of furrow tillage practices and nitrogen (N) fertilizer placement on characteristics of runoff water quality during the growing season. The experiment was designed as a randomized complete block design with conventional (CT) and conservation furrow tillage (FT) in combination with either urea (URN) broadcast or 32% urea ammonium nitrate (UAN) injected, each applied at 101kgNha^-1. Concentrations of ammonium (NH₄-N), nitrate (NO₃-N), nitrite (NO₂-N), and dissolved phosphorus (P) in irrigation runoff water and lint yields were measured in all treatments. The intensity and chemical form of nutrient losses were primarily controlled by water runoff volume and agronomic practice. Across tillage and fertilizer N treatments, median N concentrations in the runoff were <0.3mgNL^-1, with NO₃-N being relatively the highest among N forms. Concentrations of runoff dissolved P were <0.05mgPL^-1 and were affected by volume of runoff water. Water pH, specific electrical conductivity, alkalinity and hardness were within levels that common to local irrigation water and less likely to impair pollution in waterways. Lint yields averaged 1111kgha^-1 and were higher (P-value=0.03) in FT compared to CT treatments. Runoff volumes across irrigation events were greater (P-value=0.02) in CT than FT treatments, which increased NO₃-N mass loads in CT treatments (394gNO₃-Nha^-1season^-1). Nitrate-N concentrations in CT treatments were still low and pose little threat to N contaminations in waterways. The findings support the adoption of conservation practices for furrow tillage and N fertilizer placement that can reduce nutrient runoff losses in furrow irrigation systems.

Continuous Instream Monitoring of Nutrients and Sediment in Agricultural Watersheds

Pollutant concentrations and loads in watersheds vary considerably with time and space. Accurate and timely information on the magnitude of pollutants in water resources is a prerequisite for understanding the drivers of the pollutant loads and for making informed water resource management decisions. The commonly used "grab sampling" method provides the concentrations of pollutants at the time of sampling (i.e., a snapshot concentration) and may under- or overpredict the pollutant concentrations and loads. Continuous monitoring of nutrients and sediment has recently received more attention due to advances in computing, sensing technology, and storage devices. This protocol demonstrates the use of sensors, sondes, and instrumentation to continuously monitor in situ nitrate, ammonium, turbidity, pH, conductivity, temperature, and dissolved oxygen (DO) and to calculate the loads from two streams (ditches) in two agricultural watersheds. With the proper calibration, maintenance, and operation of sensors and sondes, good water quality data can be obtained by overcoming challenging conditions such as fouling and debris buildup. The method can also be used in watersheds of various sizes and characterized by agricultural, forested, and/or urban land.