Search Results (1,902)

Intelligent agents are emerging as an important system-level paradigm for smart agriculture. This review focuses on modern agricultural intelligent agents driven by large language models and related multimodal foundation models and examines how this emerging field is reshaping the organization of intelligent agricultural systems. It first clarifies the conceptual boundaries of agricultural intelligent agents and distinguishes them from traditional multi-agent systems, agent-based modeling, agricultural foundation models, and static retrieval-augmented question-answering systems. It then synthesizes their architectural foundations, key capabilities, application scenarios, deployment challenges, and future research directions. The reviewed literature indicates that agricultural intelligent agents are moving beyond isolated perception, prediction, and response generation toward the goal-oriented coordination of agricultural knowledge, dynamic data, external tools, and decision-making processes across agricultural task chains. They are beginning to support more integrated forms of knowledge services, crop monitoring and diagnosis, decision support, and farm-level collaborative management. Nevertheless, their transition from prototype systems to dependable and deployable agricultural systems remains constrained by context-aware knowledge grounding, heterogeneous data and tool integration, long-horizon reliability, the stability of multi-agent collaboration, and system security. This review further introduces an assessment perspective based on evidence reported in the original studies, comparing representative agricultural intelligent agents in terms of task decomposition, agronomic evidence applicability, tool-use validity, workflow reliability, multi-agent coordination, and deployment-related evidence. By distinguishing demonstrated capabilities from unevaluated dimensions, this review provides a structured framework for understanding the current status of agricultural intelligent agents and for guiding their future development toward reliable, deployable, and domain-oriented intelligent systems for smart agriculture. Full article

Adequate soil quality is essential for ensuring the productivity and sustainability of agriculture, with soil pH being a key variable due to its influence on nutrient availability, microbial activity, and plant development. This study proposes a Mamdani fuzzy inference system (FIS) optimized through metaheuristic algorithms for soil pH prediction in papaya (Carica papaya L.) cultivation, a crop highly sensitive to pH fluctuations within the rhizosphere. Soil temperature and soil moisture were used as independent variables, while the estimated soil pH constituted the dependent variable of the system. Three optimization techniques—genetic algorithms (GAs), Differential Evolution (DE), and Particle Swarm Optimization (PSO)—were evaluated to optimize the membership functions and fuzzy rule base of the Mamdani FIS. Model performance was assessed through 30 independent runs using 1500 records collected from a commercial papaya plantation. Across the 30 independent runs, the GA-optimized model achieved the best overall predictive performance, with a Mean Absolute Error (MAE) of 0.3636 ± 0.0035, Mean Relative Error (MRE) of 0.0554 ± 0.0004, and mean coefficient of determination (r²) of 0.8699 ± 0.0048. The best observed GA values were an MAE of 0.3305, MRE of 0.0513, and r² of 0.8957. In addition, a web-based decision support platform and an automated Telegram alert system were developed for event-driven pH alert notification. The results confirm that GA-optimized fuzzy systems constitute an effective, interpretable, and practical tool for pH-based soil condition monitoring and agronomic decision support in precision agriculture. Full article

This study proposes an innovative framework that combines hydroclimatic and agro-environmental predictors, including nitrate concentration and NDVI, with climatic factors such as Rainfall, temperature, and evapotranspiration to forecast piezometric level variations in the Saïss Basin, Morocco. Eight Machine Learning (ML) models were benchmarked, and feature importance was assessed using Shapley Additive exPlanations (SHAP) to ensure model transparency and interpretability. In parallel, the PELT algorithm was applied to detect structural change points, while Sen’s slope estimator and the Mann–Kendall test were used to quantify long-term trends. The Extra Trees (ET) model achieved the best performance (R² = 0.92), with Rainfall emerging as the most influential predictor, followed by nitrate concentration, confirming the added value of hydrochemical indicators for groundwater forecasting. Change-point analysis revealed significant declines during the 1980s and 1990s, followed by lower-amplitude fluctuations since the late 2000s. Projections toward 2050 suggest partial stabilization in the central part of the basin under favorable recharge conditions, whereas persistent declines are expected to continue in peripheral areas subjected to sustained groundwater abstraction pressure. These findings provide a robust and transferable decision-support tool for the sustainable management of groundwater resources in semi-arid agricultural area. Full article

Digital transformation in agricultural production requires automated approaches for monitoring, risk assessment, and decision support under uncertainty. This study proposes an automated risk assessment and control framework for feed production in digital agroengineering systems. The framework is designed for the quantitative evaluation of producer and consumer risks arising during the control of key feed-quality parameters, using the case of feed production for cattle in the OHMK agricultural holding. The proposed approach integrates probabilistic modeling, simulation-based risk estimation, fuzzy logic, expert evaluation, and a multi-agent representation of agroengineering processes. A three-dimensional risk model is developed to represent producer risk, consumer risk, and actuarial risk as interconnected components of a digital control environment. In addition, a fuzzy model is introduced to assess the robustness and digital maturity of management functions, including organization, planning, motivation, and control. Computer experiments based on statistical data for crude protein content in silage demonstrate that control risks depend nonlinearly on measurement uncertainty, parameter variability, and normative thresholds. In the analyzed single-indicator case study, the arithmetic mean of crude protein content in silage was 7.5% of dry matter, the standard deviation was 0.5, and the Weibull approximation parameters were α = 1.0, β = 2.5, and γ = 6.0. Under the most sensitive normative threshold scenario, producer risk increased to approximately 25%, while consumer risk showed a lower but nonlinear increase with measurement uncertainty. The results show that producer risk may reach significant levels when measurement uncertainty becomes comparable with the variability of the controlled parameter. The proposed framework can serve as a computational basis for future automated monitoring, risk-aware control, and decision-support systems in Industry 4.0-oriented agricultural production. Full article

Vernacular heritage conservation often separates material fabric, environmental adaptation, and social meaning. This study addresses that separation for one settlement. Socio-environmental heritage is defined here as heritage whose significance resides in the documented interdependence of environmental conditions, material practice, social organisation, and customary governance, rather than in fabric or imagery alone. The single purpose of the article is to develop and demonstrate an evidence-based method for interpreting and conserving Thee Ain Heritage Village in Al-Baha, Saudi Arabia, as such a system. A qualitative architectural case-study design combines a structured literature search, regional comparison, the author’s 2014 field observations and photographs, and published digital-heritage, energy-retrofit, and conservation studies; no human-participant data are analysed. Evidence is organised through three analytical layers—climatic material, socio-spatial, and customary governance—and each evidence–interpretation proposition is classified as directly observed, supported architectural inference, or hypothesis requiring measurement; conservation translation is treated as the output of this sequence rather than as a parallel analytical layer. Coded claim units are documented individually so that every interpretation and implication can be traced to its source, strength, and limitation. The analysis links rocky siting, stone and timber assemblies, thick load-bearing madameek walls, limited openings, vertical domestic hierarchy, controlled thresholds, and the agricultural setting to conservation priorities at landscape, construction, spatial, and adaptation scales. These priorities include compatible repair, retention of wall depth and opening logic, protection of privacy gradients and threshold sequences, and service integration without reducing the village to stone-clad imagery. Unlike previous work centred on digital documentation, energy modelling, or policy-level preservation, the contribution is an evidence-structured method linking architectural observation to bounded interpretation, conservation decisions, and explicit future testing requirements. Full article

The greenhouse whitefly (Trialeurodes vaporariorum) causes substantial yield losses and reduced market value of produce in protected agricultural systems worldwide. This study examined the biocontrol potential of recombinant Bacillus thuringiensis (Bt) toxin proteins harboring the LBW08 gene. Within a decision-support system (DSS) framework, we evaluated the effects of different competent cell lines on toxin expression profiles and assessed the insecticidal performance of two Bt toxin variants (08 and 08bt), both individually and in combination with yellow sticky traps. Our goal was to generate quantitative parameters on population suppression dynamics and cost-effectiveness, providing a basis for DSS-driven pest management recommendations. The 08Bt toxin expressed in JM110 competent cells exhibited superior purity and consistent insecticidal activity against T. vaporariorum nymphs. Notably, the integrated application of 08Bt and yellow sticky traps achieved the highest level of population suppression among all treatments, highest numerical level of population suppression among all treatments, although pairwise differences were not statistically significant, suggesting strong potential for synergistic control. In contrast, no synergistic interaction was observed with the 08 variant. However, the 08 variant combined with traps did not show a comparable numerical improvement. These findings support the development of cost-effective, environmentally sound tactics that can be incorporated into broader IPM strategies for protected agriculture. Full article

The potato production process in Konya, Türkiye, was evaluated through a cumulative and system-oriented approach. A functional unit of one ton of potatoes produced was used for all analyses, using region-specific agricultural input data. In this study, the cumulative energy consumption (CEnC), exergy consumption (CExC), and CO₂ emissions (CCO₂E) of the agricultural production process were determined. Specifically, the sustainability performance of the potato production process was examined through thermodynamic indicators. The results indicate that nitrogen fertilizer accounts for the highest CEnC, reaching 368.94 MJ per ton of potato produced, followed by diesel fuel at 179.64 MJ/ton and electricity at 91.79 MJ/ton. However, the CExC assessment revealed a different pattern, with electricity emerging as the dominant source of exergy depletion. Electricity consumption accounted for 382.75 MJ/ton, representing the largest exergy burden among all inputs, while diesel (166.21 MJ/ton) and nitrogen (154.28 MJ/ton) followed as secondary contributors. A similar trend was observed in the carbon emission analysis. Electricity use resulted in the highest CCO₂E value at 12.85 kg CO₂/ton, whereas diesel contributed 2.94 kg CO₂/ton. Emissions from chemical fertilizers remained notably low, with nitrogen, phosphorus and potassium generating only 0.42, 0.22 and 0.75 kg CO₂/ton, respectively. The sustainability indicators further highlighted the system’s performance. The cumulative degree of perfection (CDP) was calculated as 7.34, while the renewability indicator (RI) reached 0.86, suggesting that potato production in Konya demonstrates relatively high thermodynamic efficiency and a strong potential for renewable energy integration. Under a scenario in which agrivoltaic systems (AVS) and fully electric agricultural machinery replace conventional energy inputs, the CDP increased markedly to 21.61 and the RI to 0.95. To the authors’ knowledge, this study is the first thermodynamic analysis of potato production in Türkiye that integrates sustainability indicators with an AVS integration scenario. The proposed framework provides a practical decision support approach for evaluating the integration of renewable energy into agricultural production systems. Full article

Irrigation scheduling under seasonal water-use restrictions is a pressing challenge in water-scarce agricultural regions, where finite volumetric allocations demand careful timing and depth decisions to sustain profitability. Deep reinforcement learning (DRL) offers promise for optimizing such sequential decisions, yet existing observation designs rely on backward-looking weather statistics and omit near-term forecasts that may support the management of limited water across a growing season. This study evaluates whether augmenting the DRL agent’s observation space with seven-day precipitation and reference evapotranspiration forecasts and refactoring existing allocation information into three budget-pacing features can improve irrigation scheduling most effectively under seasonal water scarcity while retaining benefits as restrictions are relaxed. Proximal policy optimization policies were trained within the AquaCrop framework for irrigated maize in southwest Nebraska under 50, 75, 100, 125 mm and unrestricted seasonal water caps. Under the 50 mm cap, the augmented feedforward policy (FB-MLP) achieved 144.76 $/ha, which was 32.2% above the baseline policy and 22.3% above the optimized Soil Moisture Target benchmark against the validation set. Its best-run gains over the baseline across the other four scenarios averaged 2.52%, including a 1.2% improvement under unrestricted irrigation. Under the 75 mm cap, the augmented policy allocated 83.3% of its irrigation to flowering and yield formation. These findings show that the combined observation design improves irrigation scheduling most strongly where water scarcity is binding. Full article

Plant disease forecasting plays an important role in modern crop protection by enabling early disease prediction and supporting optimized management decisions. With the rapid development of digital agriculture, artificial intelligence, and environmental monitoring technologies, forecasting systems have evolved from traditional empirical and mechanistic models to machine learning, deep learning, multi-source data fusion, and hybrid forecasting frameworks. Unlike previous reviews that mainly focused on specific model types, decision support systems, or disease recognition technologies, this review provides a comprehensive synthesis of different forecasting approaches and their practical applications. The strengths and limitations of various models are comparatively analyzed in terms of predictive performance, interpretability, fungicide reduction potential, and practical applicability. In addition, recent advances in climate-driven disease forecasting, precision disease management, and intelligent decision support systems are discussed. Finally, current challenges and future directions, including AI-mechanistic model integration, multi-disease forecasting, IoT and remote sensing data fusion, and climate-adaptive forecasting systems, are highlighted to support the development of sustainable and intelligent crop protection strategies. Full article

Urban and peri-urban regions increasingly face climate-related pressures, competing land-use demands, and the need to expand renewable-energy infrastructure while maintaining agricultural land and landscape functions. Agrivoltaics, which combines photovoltaic energy generation with agricultural production, represents a potentially multifunctional approach to land use; however, regional GIS assessments often stop at environmental suitability surfaces without translating those results into planning-relevant cadastral inventories. This study develops and applies a planning-oriented Geographic Information System (GIS) framework for preliminary statewide agrivoltaic screening in Connecticut. Annual global solar radiation and terrain slope were integrated through a weighted suitability model, while incompatible land-cover classes were treated as hard exclusions through a binary land-cover mask. The workflow subsequently excluded protected and open-space lands, associated suitable areas with cadastral parcels, normalized and dissolved parcel identifiers using ParcelKey, and a recalculated suitable area from the resulting unique parcel geometries and then applied a minimum requirement of 1 ha of cumulative suitable area per retained parcel. The final baseline inventory contained 3497 normalized unique cadastral parcels encompassing 16,366.49 ha of GIS-identified suitable area, with suitable land representing an average of 42.46% of total parcel area. Peri-urban contexts accounted for the largest share of the final suitable area, containing 2497 parcels and 73.16% of the total, compared with 476 urban and 524 rural parcels. Sensitivity analysis indicated strong stability under alternative weighting schemes, with spatial overlap exceeding 99% relative to the baseline. Reducing the suitability-score threshold from 3.0 to 2.5 produced only minor changes, whereas increasing it to 3.5 reduced the inventory to 3095 parcels and 13,712.89 ha. From a sustainability perspective, the framework provides a spatial decision-support approach for coordinating renewable-energy planning with agricultural land stewardship, conservation constraints, and more efficient use of already fragmented land resources. By making the effects of exclusions, parcel thresholds, and analytical assumptions explicit, the approach supports more transparent and reproducible evaluation of land-use trade-offs relevant to sustainable development. The resulting inventory is intended as a first-stage planning resource rather than a determination of project feasibility or site-level sustainability performance. Full article

Accurate identification of the spatial distribution of agricultural greenhouses is essential for regional agricultural management in Japan. However, an efficient method for systematically mapping greenhouse locations onto existing farmland maps has not yet been established, and field surveys over large areas require substantial time and labor. This study aims to develop an automated approach for detecting agricultural greenhouses and integrating their locations into farmland maps by combining PlanetScope satellite imagery with farmland polygon data developed by the Japanese government. To improve the efficiency of the extraction process, farmland polygons were used to restrict the analysis to known agricultural areas, thereby reducing false detections originating from non-agricultural land. Within these predefined regions, three machine learning algorithms—Random Forest (RF), Support Vector Machine (SVM), and Isolation Forest (ISF)—were applied to classify and extract greenhouse features from satellite imagery. After optimizing the hyperparameters of all models, RF and SVM achieved an equivalent peak performance, with an F1-score of 0.86, while ISF reached 0.72. RF was, however, markedly more robust to the polygon-level decision threshold, demonstrating a practical advantage in situations where the threshold cannot be optimized in advance. In addition, an ablation experiment confirmed that without pre-masking with farmland polygons, 81.9% of the pixels predicted as greenhouse were distributed outside the agricultural parcels. The proposed method is expected to serve as an effective approach for efficiently identifying the distribution of agricultural facilities and integrating them with existing farmland information in regions characterized by small and fragmented agricultural fields, such as Japan. Full article

Olive cultivation is an important agricultural sector in Albania, where disease monitoring remains challenging due to fragmented orchards, heterogeneous management practices, and limited adoption of advanced sensing technologies. Recent studies have shown the potential of unmanned aerial vehicles (UAVs), sensor systems, and artificial intelligence (AI) for monitoring specific olive diseases in different olive-growing regions. However, the suitability of olive diseases reported in Albania for monitoring with UAVs and AI has not yet been systematically assessed. This paper examines the main olive diseases relevant to Albania using a semi-quantitative, literature-based multicriteria framework in which five monitoring criteria are scored from 1 to 3 and combined using equal weights. It also formalizes a UAV-first screening workflow that links image acquisition, AI-based canopy segmentation, feature extraction, anomaly scoring, decision thresholds, and targeted field or laboratory confirmation. Based on this assessment, the study identifies the most promising disease targets for future research and outlines key considerations for sensor selection and validation. The paper provides a context-specific foundation for future UAV- and AI-supported disease monitoring in Albanian olive groves. The revised analysis also distinguishes indicative acquisition targets from experimentally validated detection limits and specifies practical requirements for ground truth, radiometric calibration, dataset design, and geospatial validation. Full article

The objective of this research was to develop a system capable of detecting diseases in maize leaves through the analysis of foliar images, using machine learning techniques to support early diagnosis in the agricultural sector. A custom dataset of field-captured images was built, and a preprocessing workflow was applied that included data augmentation, segmentation, feature extraction, and normalization. Subsequently, three models: EfficientNetB0, DenseNet121, and EKNN were trained and evaluated to determine the architecture with the best classification performance. The results showed that model performance varied according to how each approach processed visual features, with the EKNN model achieving the highest overall accuracy of 94.33%, outperforming EfficientNetB0 (89.70%) and DenseNet121 (88.04%). The CNN-based architectures achieved adequate classification in diseases with well-defined patterns but presented limitations when dealing with visually similar lesions. In contrast, the EKNN model, which relies on segmentation and enhanced feature extraction, achieved the best overall performance, demonstrating the importance of preprocessing in diagnostic accuracy. Finally, the selected model was integrated into a functional web application, validating its practical utility as a tool for the early detection of diseases in maize leaves. This research demonstrates that machine learning can effectively assist farmers and agricultural technicians in the efficient identification of plant diseases, contributing to improved productivity and better decision-making in the field. Full article

Background: Managing agrologistics supply chains under infrastructure scarcity requires integrative, spatially explicit decision-support tools. This study develops a macro-level digital twin of the multimodal agricultural supply chain in Kazakhstan’s Almaty region to optimize freight allocation and guide strategic investment planning. Methods: Our methodology integrates Earth observation data (ESA WorldCover 10 m) with a large-scale multimodal road–rail graph network (1.39 million nodes) to identify 135 crop production clusters. Using linear programming in MATLAB, we optimize the regional distribution of 322.2 thousand tons of seasonal maize, wheat, and soybeans while localizing new storage silos using Green Field Analysis. Results: The baseline simulation reveals a critical storage capacity deficit, yielding a Capacity Coverage Ratio of only 23.8%. However, implementing optimal multimodal rail-road routing mathematically reduces the Logistics Cost Index from 8,642,195 to 4,716,175 units, achieving overall cost savings of 45.4%. Conclusions: The proposed digital twin and its performance metrics provide a scientifically grounded, data-driven toolkit for public–private partnerships, ensuring robust infrastructure investment localization and facilitating the transition toward the Agriculture 4.0 paradigm. Full article

Yield maps (YMs) are essential precision agriculture tools for assessing spatial variability in crop yields within an agricultural area. However, their quality is often compromised by duplicate points, negative or null points, outliers, and inliers, requiring the cleaning of raw harvest monitor data. This study aimed to develop, implement, and evaluate AgDataBox-Map-Clean v4.0.2 (ADB-Clean v4.0.2), a computational web module for cleaning raw harvest monitor data and integrating it with the AgDataBox-Map (ADB-Map) application from AgDataBox v4.0.2 (ADB). ADB is a web platform that integrates data, software, procedures, and methodologies for digital agriculture, and it is the only free, all-inclusive platform worldwide. The module cleans raw yield monitor data by removing duplicate points, negative or null points, outliers, and inliers. In a case study, three cleaning software programs (web module ADB-Clean v4.0.2, Map Filter v2.0, and Yield Editor v2.0.7) were compared. The YMs constructed from data processed by the three software programs were visually similar, but their similarity to raw-data YMs (without cleaning) was much worse. On average across the evaluated datasets, the cleaning process reduced the standard deviation (SD) by 68%, the coefficient of variation (CV) by 70%, and the dataset size by 23%, while increasing the mean yield by 11%. The results demonstrate that, for the datasets evaluated in this study, automated error removal from yield monitor data is feasible and can support improved agricultural decision-making. Because its cleaning algorithms operate on georeferenced point data, the module is expected to apply to other agricultural datasets, although this capability was not evaluated in the present study. Full article