Precision Agriculture最新文献

Purpose

Management zones (MZs) are the subdivision of a field into a few contiguous homogeneous zones to guide variable-rate application. Delineating MZs can be based on geostatistical or clustering approaches, however, the joint use of these approaches is not usual. Here, we show a joint use of both techniques. The objective of this manuscript is twofold: (1) compare different procedures for creating management zones and (2) determine the relation of the MZs delineated with i) coffee yield maps and ii) the summarizing power of each method for each input variable inside the MZs delineated.

Methods

The techniques compared to summary spatial data were: (1) summarizing the variables into a soil fertility index (SFI), (2) the MULTISPATI-PCA technique, and (3) the multivariate Min/Max autocorrelation factors (MAF) approach. Then, clustering methods were applied to perform field partition into binary MZs (grouping lower and higher values of input variables).

Results and discussion

The MAF approach achieved the best field partition regarding clustering metrics (McNemar’s test, Silhouette Score Coefficient, and variance reduction). In this paper we did not use yields as a cluster variable but as a measure of success. MAF also was the best one for separating low- from high-yielding areas over the MZs. The results show that the proposed approach could be effectively used for management zone delineation.

Conclusions

This methodology facilitates evaluating innovative approaches in challenging spatial modeling scenarios, such as low-sampled fields with outliers. A wide range of summarization methods and clustering techniques are available, making this agnostic approach quite interesting for delivering MZ maps. This flexible approach can guide precision nutrient management in low-sampled areas, allowing the joint use of data science tools and agronomical knowledge to delineate variable rate application strategies.

Graphical abstract

Purpose

The use of UAVs (Unmanned Aerial Vehicles) equipped with sensors such as laser scanners offers an alternative to conventional, labor-intensive manual measurements in agriculture, as they enable precise and non-destructive field surveys.

Methods

This paper evaluates the use of UAV-based laser scanning (RIEGL miniVUX-SYS) for estimating the crop height and the plant area index (PAI) of winter wheat. (Methods) It further introduces a novel ground classification method, enhancing early growth stage classification through sensor attributes like intensity and pulse shape deviation.

Results

The crop height estimation shows a high (R^2) score with (99.69~%) but a systematically lower estimate with a mean absolute error of 7.4 cm. The potential of PAI derivation is analyzed with three different estimation strategies and provides an overview and limitations of the approach. Additional weighting based on the scan angle and the adaptation of the extinction coefficient present results with (R^2) of (97.66~%) and a mean absolute error of 0.25.

Conclusion

The investigation discusses further the impact of the calculated gap fraction, which describes the ratio of laser beams penetrating through the crop canopy in comparison to the total number of measurements.

The provision of ecosystem services from Mediterranean permanent grasslands is threatened due to shifting management practices and environmental pressures. This observational study tested the hypothesis that Land Surface Phenology (LSP) parameters from high-resolution satellite data can characterize various permanent grasslands to support conservation and improvement practices. The potential of LSP derived from Sentinel-2 data in identifying the multi-layer mixed vegetation of Mediterranean grasslands, including silvopastoral systems, that were well-characterized from an agronomic and ecological perspective through field surveys, was assessed. Forty-nine polygons, representing eleven sites characterized by different grassland vegetation, soil, climate and management, were identified in Sardinia (Italy). Sentinel-2 satellite images from 2017 to 2023 were processed to derive NDVI, and LSP parameters were calculated using TIMESAT 3.3 software. The Canonical Correspondence Analysis showed a significant association (p < 0.05) between a combination of LSP metrics used as proxies of a set of relevant agronomical indicators. It was then possible to differentiate managed vs. abandoned grasslands (e.g., start and peak of the season significantly later under unmanaged grasslands, p < 0.0001), wooded grasslands vs. open grasslands(e.g., base value significantly higher in woodlands and wooded grasslands, p < 0.0001) across environmental gradients (altitude) and management practices (green-down rate significantly higher under mown than unmown areas, p < 0.0001). The LSP parameters proved to be promising proxies to characterize agronomic features (e.g., length of the growing season, earliness, forage availability, mowing and grazing intensity, unpalatable species) of Mediterranean permanent grasslands. The characterization can support management design or monitoring to detect abandonment or environmental pressures early.

The article defines the key determinants of adopting precision agriculture technologies and digitalisation. The research objectives are fulfilled by the systematic review and meta-analysis of relevant studies, identified and selected in accordance with the PRISMA protocol in the Web of Science and Scopus databases. The findings emphasize the importance of socio-economic factors, such as education, age, and farm size. High technical literacy and adequate information about new technologies—including their expected profitability—are crucial for assessing the benefits of precision agriculture and digitalisation, on which a more considerable expansion of these technologies into the practice of agricultural entities depends. Large and capital-intensive enterprises are more likely to implement new technologies in production practices, especially if they are led by younger and more educated managers who are more open to modern technologies and are more willing to take risks.

The global attention to the utilization of unmanned aerial vehicle remote sensing drones in crop disease-wide detection has led to the urgent need to find an adapted model for different environmental conditions. Therefore, the current study has focused on spatiotemporal usage of different multispectral cameras in acquiring spectral reflectance models of in-field rice bacterial blight stresses. Where, long short-term memory (LSTM) model was compared with the other models in transfer learning strategy for assessing the blight stress severity. The results revealed that by extracting 30% of the data from the target domain and transferring it to the source domain, the adaptability of the model across different sites was effectively enhanced. Besides, LSTM showed high tuning transfer efficiency that demonstrated optimal predictive performance and the shortest training time in transfer tasks. Its coefficient of the prediction set was 0.82, and its residual prediction deviation has reached 2.26. In practice, LSTM enabled the acquisition of reliable prediction results at a minimal sample collection cost while circumventing feature reduction resulting from inter-domain data alignment. When the transfer ratio reached 20%, the coefficient of determination of the prediction set reached 0.71, and the residual prediction deviation reached 1.79. The novelty of this study came from the transfer learning efficiency in improving the model’s application capabilities across the different sites, environment, and unmanned aerial vehicle in farmland disease detection.

The first stage in the precision agriculture cycle has been a vital study area in recent years because it allows soil testing followed by data analysis. In this stage, a strategic delineation of site-specific management zones acquires a particular interest because it enables site-specific treatment to improve crop yield by efficiently using the input of resources. The delineation of site-specific management zones problem is to determine the minimum number of zones that cover the entire field so that each zone’s homogeneity is significant according to a specific biological, chemical, or physical soil property. Furthermore, the delineated zones should be orthogonal-shaped to be practical for agricultural machinery. This work has proposed a new bio-inspired algorithm, specifically an Estimation of Distribution Algorithm, based on a decoder that heavily relies on the Disjoint-Set algorithm and a new reactive penalized fitness function that detects unfeasible solutions. The new methodology improves the solutions presented in the literature by using a new search engine that drastically reduces the computational times of similar algorithms. Our algorithm has been tested with the literature benchmark, considering a new reactive penalization in the fitness function. It obtains the best solutions for 66.66% of the instances benchmark compared to the best literature method. Due to the algorithm’s efficiency, a new set of larger instances is introduced to test the scalability and robustness of the method. It obtained an efficiency of 79.3%.

Introduction

Optimizing fruit quality and yield in agriculture requires accurately monitoring leaf nitrogen (N) status spatially and temporally throughout the growing season. Standard remote sensing approaches for assessing leaf N rely on proxies like vegetation indices or leaf chlorophyll a + b (C_ab) content. However, limitations exist due to the C_ab-N relationship’s saturation and early nutrient deficiency insensitivity.

Methods

The study utilized Sentinel-2 satellite imagery to estimate a set of plant biochemical traits in large almond orchards in a two-year study. These traits, including leaf dry matter, leaf water content, and leaf C_ab retrieved from the radiative transfer model, were used to explain the observed variability of leaf N. Airborne hyperspectral imagery-derived leaf N using C_ab and solar-induced fluorescence served as a benchmark for validation.

Results

Results demonstrate that plant traits quantified from Sentinel-2 were strongly associated with leaf N variability across the orchard, with a strong contribution from the estimated leaf C_ab content and leaf dry matter biochemical constituent, outperforming the consistency of vegetation indices. The Sentinel-2 model explaining leaf N variability yielded r² = 0.82 and nRMSE = 13% in a two-year dataset, obtaining consistent performance and trait contribution across both years.

Conclusion

This study highlights the potential application of Sentinel-2 satellite imagery for monitoring leaf N variability in almond tree orchards. Incorporating plant biochemical traits allows for a more consistent and reliable prediction of leaf N compared to traditional vegetation indices over two years, making it a promising method for precision agriculture applications.

On-farm experiments (OFE) typically do not account for limitations of grain yield monitors such as the dynamics of grain flow through a large combine. A common question asked within OFE is how ground speed impacts yield estimates from grain yield monitors. Therefore, the objective of this study was to determine if combine ground speed influences the ability of grain yield monitors to report yield differences for OFE. Six sub-plot treatment resolutions that differed in length (7.6, 15.2, 30.5, 61.0, 121.9, and 243.8 m) of imposed yield variation were harvested at combine ground speeds of 3.2 and 6.4 km h⁻¹. Treatments were replicated 3 times. The intentional yield variability in maize (Zea mays L.) was created by alternating nitrogen application (0–202 kg N ha⁻¹) across the treatment lengths. A factory installed yield monitor (YM3) and a third-party platform (P1) using the controller area network (CAN) bus data were used to collect yield data and compared to plot combine data collected from adjacent rows for each treatment length along a pass. Comparisons were made between each YM and plot combine yield estimates for each low and high yield treatment lengths. Combine ground speed did not significantly impact yield estimates (p ≥ 0.31 for all speed interactions) except speed * method due to lack of calibration. There were no significant differences the computed yield differences (all speed interactions p ≥ 0.40). Combine ground speed did not significantly influence the ability of yield monitoring technologies (i.e. mass flow sensor) to estimate the average low and high yields (p ≥ 0.31 for all speed interactions for individual plot lengths except when operating outside the calibrated flow range of the mass flow sensor. Operating outside the calibrated flow range of the mass flow sensor resulted in mass flow rate being overestimated by an average of 23% for both yield monitors (YM3 and P1).

Spatial and temporal variability of the soil’s apparent electrical conductivity (ECa) and other soil attributes can be analyzed using specific digital platforms for precision agriculture, contributing to agricultural management decision-making. Understanding these variations enables more efficient and sustainable management practices tailored to each area’s characteristics, leading to higher crop yields and reduced environmental impacts. A critical question arises: should ECa measurement be done regularly or just once? This study aims to evaluate the spatial and temporal variability of soil’s apparent electrical conductivity to determine if a single ECa measurement can characterize spatial soil variability. The experiment was conducted in two areas under different management practices in Céu Azul, PR, Brazil. One area operates under a direct planting system, cultivating soybeans in the summer and rotating with wheat or corn during the winter. The second area is used as pasture during the winter and planted with corn or soybeans in the summer. ECa data from 2013 to 2016, along with chemical and physical soil attributes from 2013, were retrieved from our laboratory database. Additionally, ECa data were collected on 19/05/2022, 18/10/2022, and 10/03/2023. All ECa measurements were performed using an EM38-MK2 conductivity meter in horizontal dipolar and drag mode. ECa normalization methods such as range, average, and standard score were employed to mitigate temporal influences partially. Data was processed using the AgDataBox web platform, which included data cleaning, data interpolation, creation of thematic maps, delineation of management zones, and spatial correlation matrix procedures. Thematic maps revealed that ECa spatial variability exhibited a stable pattern. Both areas showed significant cross-correlation among topography and most soil chemical and physical attributes. The study concluded that ECa measurement could be performed once as a co-variable for interpolating other variables since the ECa pattern remained stable in both areas. The average method was the most effective normalization method in both areas. Furthermore, management zones (MZs) were delineated using equivalent normalized ECa (ECa_Eq) (mS/m) with the three data normalization methods. The agreement between MZs was sufficient to conclude that the influence of the normalization methods can be ignored.

Nitrogen (N) is vital for plant growth, but its imbalance can negatively affect crop yields, the environment, and water quality. This is especially crucial for California’s almond orchards, which are the most N-hungry nut crop and require substantial N for high productivity. The current practices of uniform and extensive N application lead to N leaching into the groundwater, creating environmental hazards. Traditional remote sensing methods often rely on data-driven approaches that work well statistically (achieving a high R² value) with one dataset but aren’t adaptable across different datasets. To create a more robust, data-driven model, one would typically need a vast and varied collection of datasets. Our goal, however, is to develop a more universally applicable model using smaller datasets, typical of commercial orchards, that can accurately estimate N content in tree canopies, regardless of differences in spatial, spectral, and temporal data. In this study, we investigate and evaluate multiple remote sensing approaches for estimating N concentration in Californian almonds, utilizing hyperspectral imaging at the canopy level. We assess various classical vegetation indices, machine learning models, and a physics-informed 3D radiative transfer model. While cross-validated results show comparable results for radiative transfer models and best-performing machine learning models, most single vegetation indices are not capable of exceeding the baseline model (:fleft(mathbf{x}right)=bar{y}) and thus had R² value less than 0. Despite being less commonly used, 3D radiative transfer modeling shows promise as a strong and adaptable method, producing results that are comparable to the best machine learning models.