Study on the Group Threshing Characteristics of Maize Ear Kernels

Agriculture · 2026-04-16

To address the lack of direct experimental characterization of multi-kernel group threshing during maize ear threshing, an experimental study on maize ear group threshing was conducted based on kernel arrangement unit characteristics. By constructing a torque testing system for maize ear detachment, we analyzed the temporal variation in torque during detachment and its response to different experimental conditions. Statistical evaluation of torque variability and stability was performed using analysis of variance (ANOVA) and error bars. Furthermore, high-speed photography was employed to capture continuous images and analyze the trajectories of kernel motion during critical detachment stages, revealing the movement characteristics and shedding behavior of kernel clusters. The results indicate that the maize ear threshing process does not involve individual kernel detachment but primarily manifests as group threshing behavior with the arrangement unit as the fundamental unit. Furthermore, the characteristics of the variation in threshing torque correspond to the collective detachment process of the kernel group. This study provides direct experimental evidence for the group threshing mechanism of maize ears through both torque statistical analysis and high-speed visualization. These findings offer valuable insights for threshing mechanism research and the optimization of threshing components.

Keeping Pace With Intensifying Agricultural Field Inundation Events: A Framework for Testing the Mitigative Capacity of Current Best Management Practices

Global Change Biology · 2026-04-01

According to data from the USDA's Risk Management Agency, crop insurance indemnities related to precipitation, hurricanes, excess moisture, and field inundation have totaled approximately $3.65 billion across Illinois, Indiana, and Iowa over the past decade. Of this amount, an estimated $924 million (25.31%) was attributed to losses that occurred in the spring months. Cover crops and conservation tillage have been recommended as best management practices to mitigate financial impacts by reducing nutrient losses from erosion, runoff, and greenhouse gas (GHG) emissions, preventing disease and physical plant damage, and enhancing field access through improved landscape drainage. However, further intensification of field inundation events is projected in these three states as we approach the midcentury, which may lessen the mitigative capacity of these practices. Few studies have tested the resilience of these land management practices to intensifying field inundation. We propose a framework that integrates guiding research questions and field experiments to determine whether the mitigative capacity of cover crops and conservation tillage keeps pace with intensifying field inundation events. We also explore agricultural biologicals, precision agriculture, the introduction of perennial crops, and drainage management as measures to address inefficiencies associated with the mitigative capacity of cover crops and conservation tillage that may be identified during experimentation. This effort expands recommended best management practices and provides stakeholders with more options in an uncertain future due to climate change.

Characteristics and numerical simulation of maize ear threshing under mixed airflow

International journal of agricultural and biological engineering · 2025-01-01

Corrigendum to “An upscaling approach for estimating field-level irrigation water use through the Budyko framework” [J. Hydrol. 661 (2025) 133785]

Journal of Hydrology · 2025-09-26

Transfer learning for improved crop yield predictions in a cross-scale pathway: a case study for Brazilian national soybean

International Journal of Applied Earth Observation and Geoinformation · 2025-11-21

• Cross-scale pathway suffers from model scalability and data scarcity. • Transfer learning improves cross-scale pathway by 15% explained variance. • Transfer learning slightly alleviates the impact of model scalability. • Transfer learning significantly alleviates the impact of data scarcity. Crop yield prediction at a fine spatial scale is crucial for improving agricultural management and resource allocations. Many countries and regions lack fine-scale yield data for fine-scale modeling and thus have to use a “cross-scale pathway”, where coarse-scale (e.g., state-level) data are used to train a model for fine-scale (e.g., county-level or field-level) yield predictions. However, the cross-scale pathway has limited effectiveness in predicting yield due to issues with data availability and model scalability. In this study, we quantify the benefits of transfer learning in the cross-scale pathway. We applied transfer learning by fine-tuning a previously trained and validated AI-based machine learning model, originally developed for field-level soybean yield predictions in the United States, using Brazilian state-level data to predict Brazilian municipal-level soybean yield. Despite differences in environmental conditions, crop phenology, and yield responses between the U.S. and Brazil, we show that transfer learning improves the municipal-level predictions from the cross-scale pathway by increasing the R 2 from 0.29 (without transfer learning) to 0.44. Notably, this is achieved without using any municipal-level data and relying only on scarce state-level observations. When the municipal-level data were used, the transfer learning achieved an R 2 of 0.57, the most stable high performance compared with previous studies. The effectiveness of the cross-scale pathway, thus, increases from 50% to 78% with transfer learning. These findings demonstrate the benefits of transfer learning in the cross-scale pathway under data-limited conditions, and underscore the potential for global crop yield predictions across scales.

Embracing large language model (LLM) technologies in hydrology research

Environmental Research Water · 2025-05-27 · 3 citations

The growing complexity of hydrological systems necessitates innovative approaches to data management, knowledge management, and model development. Large language models (LLMs) have great potential to accelerate hydrological research by unifying and advancing these three critical aspects. In this perspective work, we review recent advances and applications of LLMs and exemplify using LLMs in hydrology studies. We demonstrate that LLMs can enhance data accessibility by efficiently extracting and organizing information from diverse sources and formats. LLMs also facilitate comprehensive knowledge management through knowledge retrieval and synthesis, enabling the integration of various datasets. Furthermore, LLMs, combined with modular development, Chain-of-Thought reasoning, and the intent-based network framework, hold immense promise for transforming physical model development and fostering model unification across scales. LLMs are powerful tools for integrating domain hydrological knowledge and advances in machine learning. Their potential in hydrological studies and the mitigation of their risks will require rigorous assessment, domain-specific regulations and guidelines, and significant contributions from hydrologists. We envision LLMs become indispensable resources for meeting the evolving demands of transdisciplinary hydrological research.

Aligning satellite-based phenology in a deep learning model for improved crop yield estimates over large regions

Agricultural and Forest Meteorology · 2025-06-06 · 11 citations

• We introduce phenology alignment for input time series in crop yield modeling. • The phenology alignment increases the accuracy of crop yield estimates by 10 %. • The benefit of phenology alignment is most significant in fields with yield loss. • This framework of phenology alignment can be widely applied. Accurate estimate of crop yield is crucial for agricultural planning and resource allocation. Crop yield estimation over large regions is especially challenging because of large variations of crop phenology and environmental conditions. Environmental stresses, such as drought and heat, during different phenological stages impose different impacts on plant growth and thus on final grain yield. We hypothesize that incorporating the phenology information, specifically, aligning environmental conditions with plant phenological timing, can better simulate the impact of environmental conditions on crop yield and thus improve crop yield estimates. To test our hypothesis, we build a deep learning model to predict crop yield using growing-season weather and satellite time series that are either aligned with field-level phenology or not. Based on rainfed corn fields in the 12 states in the US Midwest between 2011 and 2020, we show that applying the aligned phenology in the deep learning model decreases the error of corn yield estimates from 34.1 bu/ac (or 21.8 %) to 29.2 bu/ac (or 18.7 %) and increase the explained yield variability from 61 % to 71 % for leave-one-year-out predictions. The improvements are most significant in fields with yield loss that are associated with early phenological timing. Specifically, in the extreme drought year 2012, the phenology was ∼20 days earlier than normal years, aligning the input sequences with field-level phenological timing captures 78 % of the yield loss, reduces the error by 15 %, and increases the explained variability over 50 %. Our results demonstrate the importance of aligned phenological timing for capturing environmental stress in crop yield estimates and show great potential of phenology alignment in accurate crop yield estimates over large regions.

Comparing continuous-corn and soybean-corn rotation cropping systems in the U.S. central Midwest: Trade-offs among crop yield, nutrient losses, and change in soil organic carbon

Agriculture Ecosystems & Environment · 2025-06-07 · 8 citations

How do Hydrological Variability and Human Activities Control the Spatiotemporal Changes of Riverine Nitrogen Export in the Upper Mississippi River Basin?

Environmental Science & Technology · 2025-12-20 · 4 citations

Excessive nitrogen export from agricultural watersheds remains a critical water quality challenge, with the Upper Mississippi River Basin (UMRB) significantly contributing to downstream eutrophication and hypoxia in the Gulf. This study investigates the spatiotemporal dynamics of riverine nitrate plus nitrite (NO3– + NO2–-N) export across the UMRB at high spatial resolution (12-digit Hydrologic Unit Codes or HUC12 subwatershed scale) during 2001–2020 and quantifies the effects of anthropogenic activities and hydrological variability on riverine NO3– + NO2–-N export changes in the region between 2001–2005 and 2016–2020. Our results revealed hotspots of substantial increases in NO3– + NO2–-N yields across the UMRB, with distinct regional patterns in driving factors. Over the entire UMRB, NO3– + NO2–-N yields increased by 9.7 kg/ha/yr on average from 2001–2005 to 2016–2020, with anthropogenic activities contributing 4.8 kg/ha/yr and hydrological variability contributing 4.9 kg/ha/yr. The northern and western UMRB had combined influences from both anthropogenic activities and hydrological variability, while the east-central regions had predominantly hydrologically driven changes. Agricultural sources, including fertilizer, manure, and biological nitrogen fixation, collectively contributed over 80% of NO3– + NO2–-N loading throughout the basin. This framework for disentangling human and hydrological impacts provides critical insights for developing effective and targeted watershed management strategies to reduce nutrient losses and improve water quality.

Glance and Gaze: Progressive end-to-end deep learning for fine-grained 3D shape classification

2025-05-05

Fine-grained 3D shape classification (FGSC) poses a unique challenge due to subtle inter-class differences and intricate structural information, often leading to suboptimal performance. This paper introduces GGView-Net, a novel approach for FGSC utilizing a progressive Glance and Gaze mechanism inspired by human object recognition processes. GGView-Net comprises three key stages - Glance, Gaze, and Joint Classification. The Glance phase mimics the holistic perception of a 3D shape by extracting global features, while the Gaze stage focuses on generating intra-view attention features guided by the global context. Finally, the Joint Classification step tailors losses to individual views and FGSC requirements. Through comprehensive exper-iments on FG3D, GGView-Net demonstrates superior FGSC performance using only class labels, with visualization results confirming its interpretability. Moreover, GGView-Net achieves state-of-the-art results in meta-category 3D shape classification on the ModelNet40 dataset, showcasing remarkable versatility.