Accurate and timely crop yield prediction and forecasting are important for improving agricultural productivity and supporting informed management decisions. In this study, we developed and evaluated a framework for estimating in-season potato yield at the field scale using Sentinel-2 satellite time series. A Grouped TreeSHAP Stability Selection (GTSS) was first applied to identify a compact, phenology-aware subset of spectral bands and vegetation indices, thereby reducing redundancy and mitigating overfitting in small-data settings. Two deep learning architectures tailored for limited training data were then introduced: LiteTemporalConv, a lightweight temporal convolutional network, and MS-ConvBiGRU-Attn, a hybrid encoder combining multi-scale convolutions, bidirectional GRUs, and an attention mechanism. Both models were benchmarked against widely used machine learning methods, including Random Forest, Support Vector Machine, Extreme Gradient Boost, Partial Least Squares, as well as standard deep learning baselines (CNN and GRU). Results showed that the proposed models outperformed both machine learning and conventional deep learning baselines, with LiteTemporalConv achieving the highest accuracy under 10-fold cross-validation (R2 = 0.84; RMSE = 3.18 t ha-1; rRMSE = 6.36%) and MS-ConvBiGRU-Attn yielding similarly strong performance (R2 = 0.82; RMSE = 3.53 t ha-1; rRMSE = 7.03%). By comparison, the best baseline, XGB, achieved an R2 of 0.79 with an rRMSE of 9.8%. The two best-performing models were further evaluated on an independent spatial dataset to assess their generalization beyond the training region. In an additional experiment, both deep learning models trained on mid-season observations showed predictive stability for late-season yield estimation. Overall, the results highlight the importance of targeted feature selection and lightweight encoders for yield modeling in data-scarce conditions.

This literature review examines 100% bio-based binders (bio-binder) derived from algal biomass as a sustainable alternative to petroleum-based bitumen in the pavement construction industry. The traditional use of bitumen contributes significantly to greenhouse gas (GHG) emissions and environmental deterioration, highlighting the critical need for renewable, low-carbon emitting materials in response to global initiatives. Algal biomass, particularly through hydrothermal liquefaction (HTL), offers a promising alternative due to its quick growth, high lipid content, and capacity to survive in various settings, including non-arable lands and wastewater. HTL effectively transforms wet algal biomass into high-quality bio-crude oil, which may then be processed into bio-binders. Studies show that these bio-binders can match or outperform standard bitumen in terms of durability and mechanical performance. The economic feasibility of bio-binders is increasing as crude oil prices rise and biotechnology advances. Additionally, algae cultivation sequesters CO2, which may offset emissions from bio-binder manufacturing. The transition to bio-binders has major environmental and economic benefits, such as lower carbon footprints and new market potential in renewable energy. However, issues including scalability, production costs, and energy usage arise. This paper provides a complete overview of current research, technological achievements, and future possibilities, emphasizing bio-binders' potential to alter road construction sustainably. Collaboration between governments, industry, and researchers is required to develop supportive policies and technology for widespread adoption.

Context: Managing livestock manure is a key sustainability challenge because it represents both a source of nutrient pollution and a resource for renewable energy production. In Italy, the spatial concentration of livestock systems generates heterogeneous patterns of nitrogen pressure and energy potential, requiring targeted, spatially explicit policy approaches. Objective: This study uses a GIS-based spatio-temporal framework to quantify nitrogen loads and manure-derived energy potential at the municipal level in Italy. The aim is to identify environmental and energy hotspots and assess the trajectory of the livestock sector up to 2026. Methods: Municipal-level livestock data for cattle, buffalo, and pigs (2011-2023) were integrated with geospatial datasets and analysed using spatial statistics and spatio-temporal pattern mining. Spatial clustering was assessed using local indicators (Getis-Ord Gi*), while temporal dynamics were captured through a space-time cube framework. Nitrogen loads and energy potential were estimated using standardised coefficients, and forecasting models were applied to project herd and farm trends up to 2026, assuming that historical trends represent shortterm future dynamics Results and conclusions: The results reveal clear spatial patterns in nitrogen pressure and energy potential. There are persistent and emerging hotspots, which are mainly concentrated in northern Italy and partially overlap with Nitrate Vulnerable Zones. Areas with high livestock density demonstrate the greatest potential for biogas production, which confirms the spatial relationship between nutrient surpluses and opportunities for energy recovery. Temporal analysis indicates a general decline in cattle and pig herds and holdings, alongside a localized increase in buffalo farming, particularly in Campania and Lazio. Forecast results suggest continued restructuring of the sector, with an estimated 27% reduction in the number of farms between 2011 and 2026, primarily impacting small and medium-sized holdings. These findings highlight the spatial reorganisation of livestock systems, with environmental pressures and energy opportunities becoming increasingly concentrated in the intensive livestock regions of northern Italy, particularly in the Po Valley and parts of Piedmont, as well as in the buffalo-farming areas of Campania and Lazio. The proposed framework shows how combining spatial, temporal and predictive analyses can help identify critical and opportunity areas for targeted policy interventions. Significance: The results highlight areas of high nitrogen pressure and energy potential, as well as short-term structural changes in the livestock sector. By integrating spatial, temporal, and predictive analyses within a single framework, the proposed approach provides a comprehensive assessment that supports the identification of critical and opportunity areas for targeted interventions.

Understanding the spatiotemporal dynamics of urban surface temperatures is essential for climate-resilient green planning under intensifying heat stress and rapid land-use change. This study investigates seasonal thermal behavior at a microscale analytical resolution (200 × 200 m) across Bologna, a densely urbanized city with diverse morphology. Satellite-derived land surface temperature (LST) was statistically downscaled to 10 m resolution and subsequently aggregated to the 200 m grid, and integrated with urban morphological and socioeconomic variables, including built surface, tree cover, population density, land-use classes, and settlement zones. A twofold approach was developed: (1) unsupervised k-means clustering to identify thermally distinct clusters per season, revealing consistent spatial patterns where tree-dominated areas remain cooler and dense urban cores warmer; and (2) an explainable deep learning model based on a convolutional neural network (CNN) to quantify variable contributions using permutation-based feature importance, benchmarked against Support Vector Machine, Random Forest, and XGBoost models. Model performance differences were evaluated using paired statistical tests with significance assessed at accepted thresholds. The CNN demonstrated competitive performance, effectively capturing nonlinear relationships in the data and, when combined with subsequent analysis, providing insights into feature relevance. Built surface was the most influential driver (>37% across all seasons), while tree cover (7%-11%) and population density (5%-35%) showed pronounced seasonal variability, reflecting shifts in vegetation activity and energy demand. This integrative, interpretable modeling framework establishes a foundational step toward a neighborhood-scale urban climate digital twin, capable of linking multi-source data to simulate and explain intra-annual heat dynamics. The findings provide critical insights for adaptive, fine-scale heat mitigation and climate-resilient urban green planning.

Vertical farming represents a promising innovative solution to the challenges of sustainable food production, reducing land consumption and optimizing production. This agricultural system relies on vertical spaces and advanced technologies to grow plants in controlled environments, although it is typically associated with high energy consumption. However, to assess its actual environmental sustainability, a Life Cycle Assessment (LCA) approach is applied, enabling the evaluation of impacts across the entire life cycle, including production, energy use, transportation, and end-of-life stages. This study considers the LCA of a productive pilot site at the National Agriculture University of Athens, which has been used to produce Valerianella locusta and Lactuca sativa. The LCA analysis primarily takes into consideration and compares vertical farming to the open field production of Valerianella locusta and Lactuca sativa. Upscaling and improved scenarios of vertical farm production have been studied to evaluate the environmental impact trends of this technology based on specific set up variations also assessing the pros and cons of the production method. The study shows that, despite the high energy consumption for lighting and climate control of indoor production, vertical farming can outperform open-field production in several important environmental categories, depending on the crop-specific productivity. For Lactuca sativa, reductions in water consumption (-41%), marine eutrophication (-86%), freshwater ecotoxicity (-91%), and land use (-95%) compared to open-field production were observed primarily in the baseline scenario. However, freshwater eutrophication increased significantly (+425%) in the VF baseline scenario, mainly driven by upstream emissions associated with electricity production. Therefore, VF does not systematically outperform open-field systems across all impact categories. The situation is different for Valerianella locusta, which, due to lower crop yields, shows higher impacts, although these are comparable to those of lettuce under different energy supply choices and increased yields, with an unchanged system setup. The integration between vertical farming and LCA allows identifying optimization strategies, promoting technological and policy innovations that favor resilient and sustainable transition to sustainable urban agriculture.

Urban overheating poses major challenges in Mediterranean cities, affecting public health and well-being. This study comparatively evaluates how alternative greening configurations influence urban microclimate and outdoor thermal comfort in a brownfield regeneration site in Imola, Italy, using ENVI-met simulations under a representative extreme summer condition. Eight scenarios with varying vegetation density, structure, and spatial arrangement were modelled on the hottest day of the year, and the Physiological Equivalent Temperature (PET) was evaluated at representative times. Results show that greening reduces heat stress, though its effectiveness varies over time and across configurations. No meaningful cooling occurred at 5:00 a.m., confirming that vegetation has a limited impact during nocturnal radiative processes. At 9:00 a.m., the medium-density scenario (S2b) achieved the greatest PET reduction (similar to 2 degrees C), suggesting favorable evapotranspiration conditions under moderate radiation. At 4:00 p.m., the distributed high-density scenario (S3.2b) provided the strongest mitigation (similar to 1.8-2 degrees C). Distributed layouts outperformed clustered ones, highlighting the non-linear nature of vegetation cooling. Zonal analysis showed the largest cooling in public green areas, followed by parking, building, and path zones, demonstrating the influence of surface type and shading geometry. Greening also produced modest improvements in surrounding neighborhoods (up to 0.8 degrees C in the morning), although impacts remained localized. Overall, results highlight how vegetation quantity, structure, and spatial distribution influence cooling performance under critical summer conditions, supporting climate-responsive urban regeneration design. These findings contribute to sustainable urban planning by supporting nature-based strategies for climate adaptation and improved environmental quality in regenerating urban districts. Future work should consider seasonal vegetation dynamics and multi-objective design optimization.

The challenge of increasing the economic and environmental sustainability of the dairy cattle sector involves several aspects and among these, the milk production throughout the career of a cow, is perhaps the parameter that all farmers would like to know for a more efficient planning of entries and exits. In fact, if on one hand numerous researchers are studying the problem selecting most efficient animals, on the other hand, few studies have focused on the definition of tools forecasting the productivity of dairy cows in the future lactations starting from the data collected in past lactations. This aspect has a particular importance in the first years of a cow since, as well known, first lactation usually has lower production than subsequent lactations. For a farmer it is important to know, as soon as possible, if a specific animal will have on a long term, lactations with high, medium or low milk productivity. The current availability of large dataset collected by automatic milking systems or by electronic milking parlors, paves the way for application of big data approaches based on machine learning algorithms with classification learner representing one of the most promising data-driven tools. In this study, firstly two supervised learning methods, i.e., Super Vector Machine and K-Nearest Neighbors, have been applied to a large dataset of 720 complete lactations, with the object to train machine learning tools able to classify and separate first and second lactation. The two classification algorithms have been applied to the raw dataset and after the application of a dimensionality reduction method. Four different dimensionality reduction methods (i.e., ISOMAP, UMAP, MDS and t-SNE) have been tested to evaluate the most efficient for this application. Finally, the same two classification algorithms have been used for the attribution of the productivity level of the second lactation starting from data of the first lactation. The two classification methods reached very encouraging accuracy values, ranging from 70% to 73%, indicating that selected predictors despite their simplicity look very promising and entail for the definition of enhanced future models. In fact, the method is particularly interesting for practical applications, as it represents a viable support-to-decision tool for selecting the most productive animals.

In intensive farming systems the facilities have a central role on both animal welfare and animal production all this paving the way of researching new housing systems and management strategies for reducing the impacts. In particular, in the dairy cattle sector, the early detection of irregular productions is fundamental for animal health and safety. On the other hand, despite the growing interest concerning the modelling and forecasting daily production data, there is lack of studies devoted to identification of anomalous data. To this regard, in this work, a data driven approach for detecting milk production and behaviour anomalies is presented and applied to three farms selected as case study. The DAIRY CHAOS procedure proposed in this paper bases on two numerical algorithms having the scope of separately detect anomalies daily data for a single cow. Both the algorithms presented hereinafter have statistical foundations and take in input daily resting time, milk yield and climate data respectively recorded by pedometer worn by the cow, automatic milking robot and a thermo-hygrometer data logger installed in each barn. The first algorithm takes into consideration three indicators, namely Relative Yield Difference, Relative Laying time Difference and Cumulative Discomfort Index. An anomaly, i.e. a deviation from a normal value, is determined, for a single cow, for a specific day, if the three conditions assessing a noticeable deviation from the normal values of the three indicators above are contemporary verified. The second algorithm, by means of a multifit procedure, introduces the concept of reliability of robust statistics and provides statistically solid, since not affected by outlier values, milk yield and laying time trends for each animal. The application, in a production context, of the procedure proposed here can result extremely useful for the identification of animals suffering heat stress and therefore can become a support to the farmer's decisions for the mitigation of the heat stress effects and a more efficient management of the animals.

Light quality is a recognized driver of plant growth and secondary metabolism in Coleus blumei, a valuable source of rosmarinic acid (RA) and quercetin (QU), whereas its combination with salinity stress represents a potential strategy that still requires further investigation. We evaluated four LED spectra, red–blue (RB) (6:1, control), blue (B), red (R), and RB + Far-Red, under both control (0 mM NaCl) and moderate salt stress (120 mM NaCl), measuring biomass (dry weight) and RA/QU in leaves and roots after three (T1) and five weeks (T2). Blue light produced the greatest root biomass, while the leaf dry weight under B did not differ significantly from RB or RBfr. RA peaked at T2 under B in leaves and under R in roots; QU was maximal under B in leaves and under RB in roots. Extending exposure from T1 to T2 markedly increased both metabolites’ yield. Salinity had little effect on biomass, increased the total QU yield, and did not enhance the total RA yield. These results indicate that targeted LED regimes and longer exposure can raise the yields of bioactive compounds, and that combining specific spectra with moderate salinity is an effective strategy for selectively increasing quercetin accumulation in indoor-grown C. blumei.

Efficient nitrogen (N) management is essential for environmental sustainability, soil fertility, and the long-term viability of the livestock sector by reducing nutrient losses, improving resource efficiency, and mitigating environmental impacts. However, accurately estimating N loads and managing them spatially remains challenging. This study evaluates the impact of using two different data sources, traditional land-use maps (CUAS09) and satellite-derived land cover data (WC20), to estimate N loads at a large scale, focusing on livestock manure in the Campania Region, Southern Italy. Specifically, the study aims to assess how differences in spatial resolution and classification accuracy between these datasets influence N load estimation and the identification of suitable manure spreading areas. Furthermore, we investigate the role of Geographic Information Systems (GIS) in integrating multiple data sources to improve spatial analysis. Spatial analyses compared the data sources under two scenarios: (S1) traditional manure spreading techniques and (S2) advanced methods involving rapid manure incorporation. Results showed notable differences in N load estimation and the area identified as suitable for manure spreading. In Caserta, traditional land-use maps identified 21,899 ha, while satellite-based data estimated 20,571 ha. In Salerno, satellite-based data identified 11,019 ha, compared to 9,706 ha using land-use maps. The total N produced in the two study areas amounted to approximately 4,877 Mg. The overall accuracy (OA) between the two data sources was moderate (51.38 %), with a Kappa Coefficient (KC) of 23 %, indicating discrepancies in spatial agreement. These differences highlight the importance of selecting appropriate data sources for N load estimation and their implications for developing precise N management strategies. The findings emphasize the need for integrating advanced and up-to-date spatial datasets to improve accuracy, identify N hotspot zones, and support sustainable agricultural practices.

Urbanization transformed global landscapes, intensifying Urban Heat Islands (UHI), further exacerbated by climate change. Sustainable green urban design offers cooling effects through evapotranspiration and shading with varying effectiveness across regions. This study investigates the role of urban vegetation, particularly trees and grassland, in moderating temperatures across nine European cities from 40 degrees N to 53 degrees N, with Temperate to Mediterranean climates. High-resolution Land Surface Temperature (LST) data, downscaled using a Gradient Tree Boosting model, were integrated with the De Martonne Aridity Index and a Contribution Index (CI) to quantify vegetation-driven cooling across a latitudinal gradient. The results show that tree and grassland cooling effects are not spatially uniform: vegetation in cooler, less arid cities provides stronger thermal mitigation. Regression analysis using Random Forest and Generalized Additive Models revealed that vapor pressure deficit (VPD) most strongly influences vegetation cooling, followed by precipitation and solar radiation. Even similar vegetation types demonstrate differing cooling performance depending on local climatic conditions. This study emphasizes the importance of optimizing urban greening strategies to geographic and climate-specific contexts, offering actionable insights for designing climate-responsive green infrastructure to reduce urban heat.

Light quality and biostimulants regulate alkaloid biosynthesis and promote plant growth, but their combined effects on vindoline (VDL) and catharanthine (CAT) production in Catharanthus roseus remain underexplored. This study investigated the impact of different LED spectra and an arbuscular mycorrhizal fungi-based biostimulant (BS) on VDL and CAT production in indoor-grown C. roseus. After a 60-day pretreatment under white LEDs, plants were exposed to eight treatments: white (W, control), red (R), blue (B), and red-blue (RB) light, and their combinations with BS. Samples were collected before treatments (T0) and 92 days after pretreatment (T1). No mycorrhizal development was observed. VDL was detected in both roots and leaves, with higher levels in roots. R produced significantly higher mean concentrations of both VDL and CAT than W. BS significantly increased mean concentrations and total yields of both alkaloids than the untreated condition. The combination of R and BS produced the highest mean concentrations and total yields of VDL and CAT. In particular, it resulted in a significantly higher mean concentration and total yield of VDL compared to sole W. Total yields increased from T0 to T1, primarily due to a substantial rise in root yield. In conclusion, combining R and BS proved to be the most effective strategy to enhance VDL and CAT production by maximizing their total yields, which also increased over time due to greater root contribution. This underscores the importance of combining targeted treatments with harvesting at specific stages to optimize alkaloid production under controlled conditions.

This research developed a predictive model using NeuralProphet to estimate energy consumption in the ventilation system of a dairy cattle farm. The necessity for energy management in livestock farming has increased due to the growing energy demands associated with climate control systems. Approximately two years of historical energy consumption data, collected through a smart monitoring system deployed on the farm, were utilized as the primary input for the NeuralProphet model to predict long-term trends and seasonal variations. The computational results demonstrated satisfactory performance, achieving a coefficient of determination (R2) of 0.85 and a mean absolute error (MAE) of 27.47 kWh. The model effectively captured general trends and seasonal patterns, providing valuable insights into energy usage under existing operational conditions. However, short-term fluctuations were less accurately predicted due to the exclusion of exogenous climatic variables, such as temperature and humidity. The proposed model demonstrated superiority over traditional approaches in its capacity to forecast long-term energy demand, providing critical support for energy management and strategic decision-making in dairy farm operations.

Monitoring changes in the feeding behaviour of dairy cows is essential for assessing their feeding preferences, milk production, and health status. Sick cows often exhibit altered feeding patterns, such as reduced feeding time and frequency, making early detection crucial for effective farm management. Traditional methods for monitoring feeding behaviour are labour-intensive, time-consuming, and prone to errors. To address these challenges, precision livestock farming technologies have gained increasing attention. While wearable sensors, such as accelerometers and RFID tags, provide accurate data, they have limitations, including high costs and potential stress on animals. Alternatively, computer vision-based approaches offer a non-invasive and efficient solution for monitoring feeding behaviour. Deep learning techniques, particularly the YOLO (You Only Look Once) object detection model, have been widely applied in animal husbandry. Despite advancements in object detection, individual cow recognition in operational environment remains a challenge due to the lack of a standardized and viable approach.The main aim of the paper is to evaluate the reliability and validate a deep learning-based computer vision model for automatically recognizing individual cows at the feeding lane in a relevant environment. By identifying individual cows, it is possible to determine their feeding time, feeding duration and daily frequency. The paper describes the work phases from data collection to analysis and validation of an improved YOLOv8n model that, after a fine-tuning on the collected video set, achieved a precision of 85 %, a recall of 62 % (F1 score 0.72) at IoU 0.5 and processes a 640 × 640 pixels frame in just 12 ms on an NVIDIA RTX 2080. The promising results presented here contribute to the advancement and validation of computer vision applications in herd monitoring, supporting the commercial adoption of these technologies for analysing cow behaviour so increasing animal welfare and the sustainability of the animal production.

Computer vision is rapidly transforming the field of dairy farm management by enabling automated, non-invasive monitoring of animal health, behavior, and productivity. This review provides a comprehensive overview of recent applications of computer vision in dairy farming management operations, including cattle identification and tracking, and consequently the assessment of feeding and rumination behavior, body condition score, lameness and lying behavior, mastitis and milk yield, and social behavior and oestrus. By synthesizing findings from recent studies, we highlight how computer vision systems contribute to improving animal welfare and enhancing productivity and reproductive performance. The paper also discusses current technological limitations, such as variability in environmental conditions and data integration challenges, as well as opportunities for future development, particularly through the integration of artificial intelligence and machine learning. This review aims to guide researchers and practitioners toward more effective adoption of vision-based technologies in precision livestock farming.

Research in landscape planning highlights the benefits of urban green systems' ecosystem services, and the need for integrated approaches to urban green planning that can adapt to different local conditions worldwide. Green roofs can reduce urban heat, improve air quality, and manage stormwater: research indicates that plant selection and substrate management are crucial for optimizing these performances. However, relatively few works have studied the effects of combined factors on the growth of native medicinal plants suitable for low-maintenance, multi-purpose urban GRs. Filling this gap is essential to increase the knowledge for an integrated approach to urban planning.This study performed an interdisciplinary, multi-factorial evaluation of the performances of three native medicinal forbs suitable for urban GRs: Trifolium repens, Melissa officinalis, and Hypericum perforatum. They were grown on experimental GR modules, comparing single-species vs. mixed-species cultivation, and chemical fertilizers vs. mycorrhizal bio-stimulants. Biomass, cover ratio, chlorophyll content, and metabolite production were measured. Results showed that these species can be effective for sustainable management of GRs: T. repens and M. officinalis may enhance multi-purpose GR's functionality, due to their production of secondary metabolites in quantities comparable to those of plants grown using other cultivation methods, while the fast growth and uniform cover of T. repens make it suitable for extensive green roofs, if properly managed. Mycorrhizal bio-stimulation was also effective on M. officinalis and H. perforatum growth, although to a lesser extent than chemical fertilization.Overall, the results proved that low-maintenance, native medicinal plants can perform well on GRs, provide sustainable, multifunctional solutions, and produce useful quantities of secondary metabolites.

Vertical farming is a sustainable solution for urban agriculture by optimizing space and resources. However, this requires ideal indoor climatic conditions to achieve maximum crop yield and quality. This research develops and validates a prediction model based on NeuralProphet algorithm to assess the vapor pressure deficit in a vertical farming facility. The model uses environmental data such as temperature, relative humidity, and solar radiation to predict vapor pressure deficit (VPD), a key indicator of vegetation health and crop growth status. The model shows high accuracy and reliability with a root mean squared error (RMSE) of 34.80 and a mean absolute error (MAE) of 25.28. The model, demonstrating satisfactory performance in predicting VPD, enables optimization of indoor growth conditions, thereby improving resources use efficiency and minimizing operational costs. Finally, it indicates a promising application of advanced artificial intelligence tools in vertical farming management to establish a sustainable and economically feasible agricultural practice since the model can help to produce high quality crops through a precise control of environmental parameters.

A GIS-based multicriteria decision analysis (MCDA) is presented to evaluate the suitability of land for the implementation of nature-based solutions (NbS) to enhance carbon sequestration in Emilia-Romagna, Italy. Excessive carbon emissions into the atmosphere have caused rapid and profound climate change that needs to be mitigated. The use of NbS has emerged as an effective strategy to sequester atmospheric carbon and improve environmental resilience. This study focuses on identifying the best NbS to maximise carbon sequestration for three environmental zones: urban, peri-urban and agricultural. The analysis identifies optimal locations for three area-specific NbS: street trees, green spaces and buffer strips. The region was divided into 30 x 30 m grid pixels, with each grid cell assigned a value from 1 (least suitable) to 5 (most suitable). The results show that most of the high-quality pixels are located near the main urban centres and along the coastline. These results provide useful information for policy makers and urban planners who can be guided in the strategic implementation of NbS to achieve maximum environmental benefits. The work also includes an individual sensitivity analysis to validate the robustness of the proposed model and a quantitative estimate of the carbon that can be sequestered by these NbS.