Precision Agriculture最新文献

Soil sampling is crucial for capturing soil variability and obtaining comprehensive soil information for agricultural planning. This article evaluates the potential of MaxVol, an optimal design method for soil sampling based on selecting locations with significant dissimilarities. We compared MaxVol with conditional Latin hypercube sampling (cLHS), simple random sampling (SRS) and Kennard-Stone algorithm (KS) to evaluate their ability to capture soil data distribution. We modeled spatial distributions of soil properties using simple kriging (SK) and regression kriging (RK) interpolation techniques and assessed the interpolation quality using Root Mean Square Error. According to the results, MaxVol performs similarly or better than popular sampling designs in describing soil distributions, particularly with a smaller number of points. This is valuable for costly and time-consuming field surveys. Both MaxVol and Kennard-Stone are deterministic algorithms, unlike cLHS and random sampling, providing a reliable sampling scheme. Thus, the proposed MaxVol algorithm enables obtaining soil property distributions based on environmental features.

Efficient mechanical harvesting of wild blueberries across uneven topographies calls for precise header height adjustments to optimize fruit picking. Conventionally, an operator requires manual adjustment of the harvester header to accommodate the spatial variations in plant height, fruit zone, and field terrain. This can result in inadequate header positioning, which leads to berry losses and increased operator stress. This study aimed to investigate the integration of machine learning techniques with real-time geo-location data to develop an innovative system to automate harvesting operations. A supervised machine learning Random Forest (RF) model was trained based on pre-defined header setting data and integrated with the harvester’s controller to predict and position the header height using real-time geo-location data from the Starfire (SF) 6000 Global Positioning System (GPS) receiver. During harvesting, the system’s performance was evaluated at tractor ground speeds (0.31, 0.45, and 0.58 ms⁻¹) and segment lengths (5, 10, and 15 m). Results indicated that segment size minimally affected the system’s ability to adjust header height. However, at the lowest segment length, 5 m, the coefficient of determination was 97.24, 98.12, and 82.71% for the 0.31, 0.45, and 0.58 ms⁻¹, respectively. This study provided convincing results for automating the harvester header based on pre-defined settings, marking a significant step toward complete automation of the wild blueberry harvester. Automation of wild blueberry harvesting can help to increase picking efficiency and enhance profit margins for growers to justify the ever-increasing cost of production.

Crop models can improve our understanding of crop responses to environmental conditions and farming practices. However, uncertainties in model inputs can notably impact the quality of the outputs. This study aimed at quantifying the uncertainty in soil information and analyse how it propagates through the Quantitative Evaluation of Fertility of Tropical Soils model to affect yield and fertilizer recommendation rates using Monte Carlo simulation. Additional objectives were to analyse the uncertainty contributions of the individual soil inputs to model output uncertainty and discuss strategies to communicate uncertainty to end-users. The results showed that the impact of soil input uncertainty on model output uncertainty was significant and varied spatially. Comparison of the results of a deterministic model run with the mean of the Monte Carlo simulation runs showed systematic differences in yield predictions, with Monte Carlo simulations on average predicting a yield that was 0.62 tonnes ha⁻¹ lower than the deterministic run. Similar systematic differences were observed for fertilizer recommendations, with Monte Carlo simulations recommending up to 59, 42, and 20 kg ha⁻¹ lower nitrogen (N), phosphorous (P), and potassium (K) fertilizer applications, respectively. Stochastic sensitivity analysis showed that pH was the main source of uncertainty for K fertilizer (81.6%) and that soil organic carbon contributed most to the uncertainty of N fertilizer application (97%). Uncertainty in P fertilizer application mostly came from uncertainty in extractable phosphorus (55%) and exchangeable potassium (20%). A threshold probability map designed using statistical predictions served as a visual aid that could enable farmers to swiftly make informed decisions about fertilizer application locations. The study highlights the importance of refining the accuracy of soil maps as well as incorporating uncertainty in input data, which improves QUEFTS model predictions and offers valuable insights into the relationship between soil information accuracy and reliable crop modeling for sustainable agricultural decisions.

The design and analysis of on-farm experimentation (OFE) have received growing attention because of the availability of precision machinery that promotes data collection. Even though replicated trials are the most recommended designs, on-farm trials with no replication are used in scenarios where variable rate technology is not available. Despite the abundance of georeferenced data within each plot harvested with yield monitor, treatments are not replicated. This paper presents an approach to statistically analyze unreplicated OFE promoting field-specific inference of treatment effects. Statistical tools for spatial data are coupled with permutation tests to determine the statistical significance between treatment means. The new methodology (OFE-mean test) involves: (1) calculation of effective sample size (ESS) given the underlying spatial structure, (2) ANOVA permutation test on a random sample of ESS, and (3) generation of the empirical distribution of p-values from repetition of step two. The median of this empirical distribution is regarded as the p-value associated with the no treatment effect hypothesis. The OFE-mean test is illustrated using several OFE trials comparing two treatments under different scenarios: with and without treatment differences. Additional assessment is carried out under simulated scenarios with different levels of spatial correlation, variability, and mean differences between treatments. The OFE-mean test had high power to detect mean differences higher than 15% for all spatial structures when total variability was lower than 30%. After treatment effects were removed, no type I error occurred in real data. The test can be easily extended to cover scenarios with more than two treatments. R scripts and sample files to run the OFE-mean test are provided.

Investigating soil properties and yield variability in farming systems is crucial for delineating Management Zones (MZs). The objectives of study were to investigate the spatiotemporal variability of soil properties, identify spatial and temporal yield-limiting factors of soil and delineate MZs based on these factors. This study was conducted at the Xinghua Rice Smart Farm (33.08°E, 119.98°N) in Jiangsu Province, China, and the experiment covered five consecutive years of soil and rice yield testing from 2017 to 2021, with 933 geo-referenced soil samples and 140 rice yield samples collected annually. Soil samples were analyzed for pH, soil organic matter (SOM), total nitrogen (TN), available phosphorus (AP), available potassium (AK), and apparent soil conductivity (ECa). Spatial and temporal variability of soil properties and RY were analyzed using statistical and geostatistical methods. Ordinary Kriging (OK) interpolation characterized these distributions, and the random forest (RF) algorithm identified key yield-limiting factors. Subsequently, the effectiveness of using all variables to delineate the MZ was compared against the approach of defining MZs based solely on the identified yield-limiting factors. The study also compared Fuzzy C Means (FCM) and Spatial Fuzzy C-Means (sFCM) clustering to evaluate MZs and their temporal stability. Results showed that the coefficients of variation for soil properties ranged from low to medium (7.7-77.4%), with semi-variational function analyses showing moderate to high spatial dependence for most properties. Temporally, soil nutrients and ECa exhibited a slow increase, whereas pH decreased, showing the highest temporal stability for pH and the lowest for AP. RF analysis identified SOM, TN, and ECa as primary influencers of spatial variability of RY, and SOM, pH, and TN as main contributors to its temporal variability. The integration of yield-limiting factors with the sFCM method improves performance of MZ delineation, maintaining stability over the five-year period.

Rapid rangeland monitoring is critical for implementing management actions effectively and therefore, various remote sensing methods are used for rangeland monitoring. Prices of high-resolution imagery and cloud problems could avoid practicing satellite based-methods. UAV- or ground-based high resolution RGB imagery suggested as an alternative to monitor rangelands. In this study, the performance of smartphone RGB imagery was evaluated over prediction of biomass yield and forage quality of steppe rangelands. Besides, the performance of a mobile application (Canopeo) over rangeland cover was evaluated. RGB band reflection values of smartphone images were determined using a simple open-source software, ImageJ. A total of thirteen different vegetation indices (eleven commonly used and two newly introduced) were estimated and their relations with ground data were evaluated over simple linear and quadratic regression models. AGB and DMY were predicted with moderate accuracy via the newly introduced modified blue-red-green index (MBRGI) (R² = 0.5 for AGB) and recently used normalized difference blue-red index (NDBRI) (R² = 0.46 for DMY) through quadratic regression models. Green leaf index (Gli), visible atmospheric resistant index (Vari), and red green blue vegetation index (RGBVI) gave better results for forage quality predictions among the other VI’s. Gli was an accurate predictor (R² = 0.78) of forage dry matter content. However, prediction performances of VI’s were low for CP (Vari, R² = 0.26), NDF, and ADF contents (RGBVI, R² = 0.31 and 0.37 respectively). Cover data of Canopeo highly correlated both with transect (R² = 0.99) and modified wheel loop (R² = 0.73) data. These results showed that Canopeo might be a useful tool for cover predictions and smartphone-based RGB imagery has good potential for managing rangeland in terms of yield and dry matter content but the accuracy of both yield and forage quality predictions still needs to be improved.

Nitrogen Sufficiency Index (NSI) is an important nitrogen (N) stress indicator for precision N management. It is usually calculated using variables such as leaf chlorophyll meter readings (SPAD) and vegetation indices (VIs). However, no consensus has been reached on the most preferred variable. Additionally, conventional NSI (NSI_uni) calculation assumes N being the sole yield-limiting factor, neglecting other factors such as soil water variability. To tackle these issues, this study compared various variables for NSI calculation and evaluated two new N stress indicators in minimizing the impact of confounding water treatment. The following ground- and aerial-derived variables were compared for NSI_uni calculation: SPAD, sampled leaf and canopy N content (LNC, CNC), LNC and CNC estimated using hyperspectral images acquired by an Unmanned Aerial Vehicle, and three VIs (Normalized Difference Vegetation Index (NDVI), Normalized Red Edge Index (NDRE), and Chlorophyll Index) from the hyperspectral images. Results demonstrated that ground-measured variables outperformed aerial-based variables in deriving N-responsive NSI. Especially, LNC derived NSI_uni responded to N treatment significantly in ten out of thirteen site-date datasets. For the second objective, a modified NSI (NSI_w) and the NDRE/NDVI ratio were compared to NSI_uni. NSI_w reduced water treatment effects in over 80% of the datasets where NSI_uni showed evident impacts. NDRE/NDVI performed similarly to NSI_w, with the notable advantage of not requiring prior knowledge of soil water spatial distribution. This research pioneers the optimization of N stress indicators by identifying the best variables for NSI and mitigating the effects of soil water variability. These advancements significantly contribute to precision N management in complex field conditions.

In-season crop growth and yield prediction at high spatial resolution are essential for informing decision-making for precise crop management, logistics and market planning in horticultural crop production. This research aimed to establish a plant-level cabbage yield prediction system by assimilating the leaf area index (LAI) estimated from UAV imagery and a segmentation model into a crop simulation model, the WOrld FOod STudies (WOFOST). The data assimilation approach was applied for one cultivar in five fields and for another cultivar in three fields to assess the yield prediction accuracy and robustness. The results showed that the root mean square error (RMSE) in the prediction of cabbage yield ranged from 1,314 to 2,532 kg ha^–1 (15.8–30.9% of the relative RMSE). Parameter optimisation via data assimilation revealed that the reduction factor in the gross assimilation rate was consistently attributed to a primary yield-limiting factor. This research further explored the effect of reducing the number of LAI observations on the data assimilation performance. The RMSE of yield was only 107 kg ha^–1 higher in the four LAI observations obtained from the early to mid-growing season than for the nine LAI observations over the entire growing season for cultivar ‘TCA 422’. These results highlighted the great possibility of assimilating UAV-derived LAI data into crop simulation models for plant-level cabbage yield prediction even with LAI observations only in the early and mid-growing seasons.