Crop models can serve as decision-support tools, but their uncertainty must be accounted for. While previous research has shown effective calibration of crop models using remote sensing (RS) data, the remaining uncertainty is rarely quantified. This study investigated the propagation of errors associated with RS data in a coupled crop-radiative transfer model in two steps. First, the results of a Particle Filter (PF) process were examined to assess the uncertainty of the model parameters and outputs. Next, the Winding Stairs (WS) method was used to quantify the contribution of crop model parameters uncertainty to the total model uncertainty. The results show that parameters related to crop growth rate contribute more to the variance of simulated Leaf Area Index (LAI) and yield than the phenology-related parameters. These findings can guide future research to improve the model reliability by focusing on calibrating the parameters with a higher impact on model outcome uncertainty.

Climate change and public policies restricting freshwater use for agricultural irrigation are compelling farmers to maintain production with limited water. Advanced irrigation scheduling tools that combine data and computer simulations are needed to optimize water use and maximize crop productivity. Limited studies have evaluated the performance of data assimilation and model-based simulation optimization of irrigation scheduling under field conditions. The objective of this study was to evaluate model-based irrigation scheduling with and without assimilation of LAI in processing tomatoes. The treatments included two DSSAT CropGro-Tomato models. The treatments were T1 (TM0005 with LAI data assimilation), T2 (TM0030 with LAI data assimilation), and T3 (Control: TM0005 without LAI data assimilation). This study was conducted near Davis, California. Model performance was evaluated using applied water, soil water content, growth, yield, and fruit quality. Results showed no significant yield differences between treatments that assimilated LAI and the control. All the models accurately predicted LAI and yield within one standard deviation of measured values, suggesting that model-based optimization was effective with or without data assimilation. The framework reduced applied water by 26

Accurate monitoring of nitrogen (N) levels, while accounting for spatiotemporal variability is crucial for optimizing fertilization in citrus orchards. Traditional methods, such as frequent leaf and soil sampling followed by laboratory analysis, are costly, labor-intensive, and prone to human error. Remote sensing (RS) technologies, including unmanned aerial vehicles (UAVs) and satellite platforms, offer scalable and precise alternatives for N management. However, integrating these platforms poses challenges due to significant differences in spatial, temporal, and spectral resolution. This study presents a novel approach incorporating multispectral and temporal data from UAVs and Sentinel-2 satellites to estimate canopy N content (CNC) in citrus orchards. This method captures spatiotemporal variability across multiple citrus cultivars, aiming to enhance nitrogen use efficiency (NUE) while reducing environmental impact, ultimately promoting sustainable orchard management practices. The study was conducted in commercial citrus plots in the Hefer Valley, Israel, and spanned two phases. The first phase (May 2019 to April 2022) focused on four plots of the 'Newhall' cultivar, while the second phase expanded to twelve additional plots featuring five different citrus cultivars. The methodology consisted of six key steps: (1) Leaf samples from the study area were collected for laboratory nitrogen (N) analysis. (2) Acquiring and preprocessing bimonthly UAV multispectral images and Sentinel-2 satellite images to ensure data quality and consistency. (3) Segmenting individual trees using UAV imagery and extracting structural features through Structure-from-Motion (SfM) photogrammetry. (4) Processing images and extracting spectral and structural features relevant to N estimation. (5) Developing Random Forest (RF) models to estimate CNC using UAV-derived vegetation indices (VIs) and SfM data and combining these with Sentinel-2 VIs to generate canopy-scale CNC heatmaps. (6) Analyzing the relationship between CNC and yield to understand nitrogen dynamics and their impact on productivity. The integrated RF model, which combined UAV-VIs, Sentinel-2 VIs, and SfM-derived structural data, achieved superior performance (R² = 0.80, RMSE = 0.17 kg/m²) compared to models relying solely on UAV-VIs (R² = 0.68, RMSE = 0.23 kg/m²) or Sentinel-2 VIs (R² = 0.48, RMSE = 0.30 kg/m²). Additionally, CNC expressed as mass per tree demonstrated a strong positive correlation with yield (R² = 0.66), highlighting the relationship between nitrogen dynamics and orchard productivity. These results underscore the robustness of the integrated model and the clear advantage of multi-platform data fusion over single-source approaches. The study provides compelling evidence for the potential of combining UAV and Sentinel-2 data to improve CNC estimation and its correlation with yield in citrus orchards. The findings contribute to advancements in precision agriculture by offering a scalable, data-driven framework to enhance nutrient management and support sustainable orchard practices.

The increasing availability of remote sensing (RS) data and the advancement of computation abilities, combined with the demands for enhancing crop production, encourages the creation of a framework in which crop growth simulation can be updated sequentially to serve as a yield predictor and be part of a decision support system. However, crop model outputs and RS data must be linked via a radiative transfer model (RTM), which simulates the interaction between the crop and the intercepted radiation. In this study, a comprehensive coupling scheme between a crop model (DSSAT-CROPGRO-tomato) and an RTM (SCOPE-RTMo) was formulated and investigated through global sensitivity analysis (SA) and by testing the coupled model in a synthetic data assimilation (DA) experiment. The DA experiment utilized a sensitivity-based particle filter (PF) in which the SA results were used to enhance the PF convergence rate and accuracy. The SA results provide the sensitivity of simulated reflectance at different wavelengths to DSSAT-CROPGRO parameters throughout the season. This information can help guide future data assimilation experiments by choosing imaging instruments with appropriate spectral bands and timing the measurements to enhance model calibration. The results of the synthetic DA experiment showed a good convergence of the particle filter towards the ground truth. The results also demonstrated the strong relation between LAI and reflectance, as several model runs with different initial values of DSSAT-CROPGRO parameters all converged and predicted the synthetic LAI observations very well. The convergence of DSSAT-CROPGRO parameters to their ground truth values was only partial, and phenology-related parameters tended to converge better than growth-related parameters.

The increasing availability of remote sensing (RS) data and advancements in data assimilation (DA) techniques facilitate the non-destructive calibration of mechanistic crop models but necessitate a framework that digitally represents cropping systems and their spectral properties. This study implemented a coupling scheme linking the outputs of a crop model (DSSAT-CROPGRO) with a radiative transfer model (RTMo module in SCOPE). Reflectance data acquired from a multispectral camera mounted on a UAV were assimilated into the coupled model. The DA scheme was tested in an irrigation and fertilization trial with processing tomatoes, a row crop, requiring the adjustment of the model in order to reflect the vegetation and soil pixel proportions. Examining the relative contribution of dynamically updating specific RTMo parameters showed that the coupled model performed better when parameters were adjusted than when using their nominal values. Applying the DA scheme improved the normalized root mean square error (NRMSE) of the Leaf Area Index (LAI) from 59% to 42% and yield from 64% to 35%. The best performance was achieved when the most water-stressed treatment was excluded, resulting in NRMSE of 34% for LAI and 16% for yield. Since the DA scheme presented here performed well at low to moderate water stress, it should be further tested in assimilating space-borne RS data into simulations of large-scale, commercial fields.

Multi-year planning of allocation of agricultural land and irrigation water remains a major challenge, which is exacerbated by decreasing arable land and increasing water scarcity in many regions. This paper presents a model-based framework to address this challenge. One of the key elements of the proposed framework is that it takes into account explicitly the need to rotate crops according to some agronomically-based sequences. The framework consists of three nested optimizations: Innermost: Optimize water allocation assuming pre-divided fields and pre-determined crop rotations. Middle: Optimize crop rotation sequences within each field. Outermost: Optimize fields geometry to maximize net income. These computations leverage crop- and soil-specific 'Yield value vs. Irrigation' functions derived from an auxiliary multi-objective optimization problem, namely maximizing yield and minimizing water use. In this manner, all the planning is based on the knowledge contained in complex crop growth models (rather than simplistic models), without having to actually run these models a prohibitively high number of times. The procedure is illustrated on two 67 ha areas near Davis, CA, that altogether contained seven types of soil. Planning was performed for a 10-year planning horizon, assuming seven crops (&fallowing) and six crop rotation patterns were available to choose from. Several scenarios that differed in terms of water availability are presented. The results demonstrate the strong impact that crop rotation requirements have on the overall performance.

Irrigation forecasting is essential for improving water management in agriculture. This study proposed a novel and practical framework for irrigation forecasting for paddy rice. Public weather forecasts in China were selected to forecast ETo and quantitative rainfall. For the latter, a simple deterministic model and a probabilistic model based on a Pearson-III representation of the data were considered. A modified Python version of the AquaCrop model (ACOP-Rice model) was adopted to generate the probability distribution of forecasted irrigation events and rule-based irrigation recommendations. The performance of weekly irrigation forecasting was then evaluated. The analysis of the weather forecasts revealed that the accuracy of the temperature and ETo predictions was acceptable, whereas the quality of the rainfall forecasts was poor. The performance of the scenarios that using probabilistic rainfall forecasts outperformed scenarios that relied on deterministic rainfall forecasts, despite the lower quality of the probabilistic rainfall forecasts compared to the deterministic forecasts. The irrigation recommendations exhibited a systematic bias (i.e., the inclination toward overirrigation or underirrigation) when using weather forecasts, and this bias was lower when using probabilistic forecasts. Additionally, compared to ETo forecasts, the inaccuracy of rainfall forecasts had a much greater impact on irrigation forecasts.

Fire blight disease causes significant losses in pear orchards. Fire blight infection is accompanied by visual symptoms that can easily be recognized by a grower or adviser, but such visual inspection is time consuming. The present work focused on the use of convolutional neural networks (CNNs) to identify one type of visual symptom (cankers on the main trunk of dormant trees) as well as autumn blooming, which in Israel plays an important role in the epidemiology of Erwinia amylovora —the bacterium responsible for fire blight. Images of dormant trees were acquired with a tripod-mounted DSLR camera while, for autumn blooming detection, the images were acquired using a small unmanned aerial vehicle flying a few meters above the trees. In both cases, several Faster R-CNNs were trained and tested with several datasets acquired at various locations and over several years. Overall, the CNNs for canker detection achieved precision and recall rates that exceeded 90% while, for autumn blooming detection, the precision and recall rates exceeded 80% in all but one case. These trained CNNs were used to analyze automatically geo-references images, hence generating infection/blooming maps. Such maps could be one of the information layers used by growers for managing the orchard, for instance to determine whether winter sanitation is needed and/or if it was carried out properly, or to decide when and where costly manual removal of autumn flowers or copper application is required.

Estimating crop nitrogen status to optimize production and minimize environmental pollution is a major challenge for modern agriculture. The study objective was to develop a multivariate spatiotemporal dynamic clustering approach to generate Nitrogen (N) Management Zones (MZs) in a citrus orchard during the growing season. The research was conducted in four citrus plots in the coastal area of Israel. Five variables were selected to characterize each plot’s spatiotemporal variability of canopy N content. These were split into constant (i.e., elevation, northness, and slope) and non-constant (i.e., canopy N content and tree height) variables. The non-constant data were obtained via bi-monthly imaging campaigns with a multispectral camera mounted on an unmanned aerial vehicle (UAV) throughout the growing season of 2019. The selected variables were then standardized to define the clusters by applying the Getis-Ord Gi* z-score. These were used to develop a spatiotemporal dynamic clustering model using Fuzzy C-means (FCM). Four input variables were investigated in this final stage, including the constant variables only and different combinations of constant and non-constant variables. The support vector machine regression model results for estimating canopy N-content from multispectral images were R 2 = 0.771 and RMSE = 0.227. This model was used to predict monthly canopy-level N content and classify the N content levels based on the October N-to-yield content envelope curve. Delineating MZs was followed by the comparison of spatial association among cluster maps. This process may support site-specific and time-specific nitrogen management.

Crop growth simulation models are important components in agricultural management. Such models include parameters that should be calibrated locally, which can be achieved by assimilating observations using a particle filter. The number of model parameters is usually high while the number of observations is limited and adjusting parameters that are non-influential under the specific environmental conditions and growth stages should be avoided. This study suggests a novel particle filter-based framework in which sensitivity analysis (SA) is embedded in the filter so that at each data assimilation step only a subset of influential parameters is adjusted. The proposed framework was implemented in two synthetic study cases with the open-source AquaCrop model (v5.0a), assuming weekly observations of canopy cover and soil water content, and in some cases biomass. In the first case study, the adjustment of a subset of influential parameters identified by SA was compared with the adjustment of all candidate parameters, the impact of the SA screening threshold and the variance of the parameter perturbation included in the filter were investigated, and the addition of biomass measurements was examined. In the second case study, different irrigation treatments were implemented to demonstrate the impact of the growing conditions on the subset of parameters that could be calibrated. The performance of the proposed framework was evaluated by computing the root mean square error (RMSE) and normalized RMSE (NRMSE) of the states and parameters estimations, and of the final biomass and yield forecasts. Overall, all state predictions were very accurate. Estimation of the model parameters was more challenging and not all the parameters converged toward their true values. This result is not surprising considering the low sensitivity of certain pa-rameters and correlations that exist within the model. Nonetheless, and more importantly, after the assimilation of relatively few observations, the model was able to forecast final biomass and yield quite accurately. Compared to standard data assimilation in which all parameters were adjusted, the SA-embedded filter performed better according to all the indicators considered:NRMSE of canopy cover and soil water content decreased from 2.4% to 1.3% and from 3.2% to 2.4%, respectively, NRMSE of parameter estimations decreased from 13.5% to 11.4%, and NRMSE of forecasted yield decreased from 5.9% to 5.2%. Overall, a relative improvement of 16% was obtained in the average NRMSE. Assimilation of additional biomass measurements improved only the ability of the model to forecast biomass but had no positive impact on yield forecasting. The next step should be to test such an approach thoroughly with experimental data.

Crop simulation models are essential tools in supporting sustainable agricultural management. However, due to uncertainty in model parameters, the model predictions may not be sufficiently accurate. Data assimilation (DA) is a common approach to improve dynamic crop modeling by combining it with observation data. Among DA approaches, particle filter (PF) is a popular choice. Since conventional PF (CPF) may suffer from sample impoverishment problems, various filter modifications have been proposed in the literature, including the application of genetic operators (arithmetic cross-over and mutations). In this study, a novel PF approach inspired by the gene's recombination process has been developed. In this new filter, named "recombination" PF (RPF), particle diversity is increased via information exchange between surviving particles and intermediate particles which are located close to existing particles. In turn, increased particle diversity reduces the chances of sample impoverishment and thus improves filter performance. The proposed method was tested on two synthetic study cases using the open-source AquaCrop model (v5.0a) and assuming weekly observations of canopy cover and soil water content. When CPF was implemented, the overall average normalized root mean square error (NRMSE), combining state and parameter estimations, and yield forecasts, all performed throughout the season, ranged from 4.0 to 5.1 % for ensemble sizes ranging from 150 to 500 particles. When RPF was implemented with a similar number of particles, the overall average NRMSE decreased to 3.6-3.7 %, corresponding to a 7-26 % improvement. Furthermore, higher stability of the results was observed, and the final parameter estimations improved in all the ensemble sizes investigated by approximately 40 %, which would be very beneficial for predicting crop growth in the next season.

Due to climate change, increased regulation of water resources and competition from other beneficial uses, the agricultural sector is under pressure to use water more efficiently. This paper reports the field evaluation of two model-based simulation-optimization approaches for irrigation scheduling: deterministic optimization and stochastic optimization. The field experiments were conducted at the University of California Davis research farm with a processing tomato crop. The crop growth simulation model used in the study was DSSAT-CROPGRO processing tomato. In order to mitigate the impact of weather forecasts inaccuracies, irrigation schedules were updated every 5 to 10 days, depending on operational constraints. These updates were performed via custom graphical user interfaces that enabled the user to visualize the expected outcomes of various decisions or scenarios before choosing which irrigation schedule to implement. An irrigation treatment based on manual monitoring of soil water content with a field-calibrated neutron probe served as benchmark. The model-based treatments achieved yields and water use efficiencies that were not significantly different from those obtained in the neutron probe-based treatment. These results demonstrate the high potential of model-based simulation-optimization approaches for in-season adaptive irrigation scheduling, especially since neutron probes, which are considered one of the most accurate indirect methods of measuring soil water content, are not commonly used by growers.

Sustainability in our food and fiber agriculture systems is inherently knowledge intensive. It is more likely to be achieved by using all the knowledge, technology, and resources available, including data-driven agricultural technology and precision agriculture methods, than by relying entirely on human powers of observation, analysis, and memory following practical experience. Data collected by sensors and digested by artificial intelligence (AI) can help farmers learn about synergies between the domains of natural systems that are key to simultaneously achieve sustainability and food security. In the quest for agricultural sustainability, some high-payoff research areas are suggested to resolve critical legal and technical barriers as well as economic and social constraints. These include: the development of holistic decision-making systems, automated animal intake measurement, low-cost environmental sensors, robot obstacle avoidance, integrating remote sensing with crop and pasture models, extension methods for data-driven agriculture, methods for exploiting naturally occurring Genotype x Environment x Management experiments, innovation in business models for data sharing and data regulation reinforcing trust. Public funding for research is needed in several critical areas identified in this paper to enable sustainable agriculture and innovation.

Precision drip irrigation of horticultural crops is receiving interest, mostly in applications in large orchards and vineyards where spatial variability results in costs to yields, quality and water productivity. We review the issues and status of precision and variable rate drip irrigation and summarize advancements and issues regarding opportunity to consider spatial irrigation management in orchards and vineyards. Topics discussed include: the conflict between "smart" and "precise" irrigation; challenges and advancement in variable-rate drip irrigation application technologies and; the use of data, including that acquired from sensors and remote sensing, for both the delineation of management zones and decision making for irrigation scheduling within zones. Prospects for future work and progress for drip irrigated horticultural application of precision water management include variable rate dripper technologies, utilization of big data from sensors and remote sensing, and unique multivariate data processing including spatial-temporal modeling.

Current rice production in China is associated with low rainfall use efficiency. In order to increase rainfall use efficiency and develop simple water-saving irrigation modes that could be readily implemented by farmers, a new multi-objective optimization framework for irrigation modes of paddy rice was developed, based on a modified version of the AquaCrop model called ACOP-Rice model. The optimization focused on the water level at which irrigation was triggered for five growth periods and the irrigation frequency, rainfall use efficiency and yield were optimized. The procedure was tested on nine rice production cases in China for which over 60 years of historical meteorological data and irrigation guidelines were available. Analysis of the weather data showed that rainfall distribution varied greatly between the different locations and growth periods. The results obtained by following the current guidelines were compared to three optimal solutions that corresponded to minimum number of irrigation events, maximum rainfall use efficiency and “balanced” performance in which equal attention was given to rainfall use efficiency and irrigation frequency, respectively. Overall, the optimization led to lowering the water depth at which irrigation was triggered. The optimal water level after irrigation varied between the different cases, depending on the combined effects of rainfall distribution, operation constraints and length of growth period. Compared to the current guidelines, the optimized irrigation modes reduced the proportion of drainage caused by rainfall after irrigation. For optimal solutions with minimum number of irrigation events, maximum rainfall use efficiency and “balanced” performance, the number of irrigation events was reduced by 57 %, 18 % and 44 % on average (9.4, 3.0 and 7.4 fewer irrigation events per year) while on average rainfall use efficiency improved by 5 %, 19 % and 17 % without significant yield loss.

Real-time in situ measurements are increasingly being used to improve the estimations of simulation models via data assimilation techniques such as particle filter. However, models that describe complex processes such as water flow contain a large number of parameters while the data available are typically very limited. In such situations, applying particle filter to a large, fixed set of parameters chosen a priori can lead to unstable behavior, i.e., inconsistent adjustment of some of the parameters that have only limited impact on the states that are being measured. To prevent this, in this study correlation-based variable selection is embedded in the particle filter, so that at each step only a subset of the most influential parameters is adjusted. The particle filter used in this study includes genetic algorithm operators and Monte Carlo Markov Chain for alleviating filter degeneracy and sample impoverishment. The proposed method was applied to a water flow model (Hydrus-1D) in which soil water content at various depths and soil hydraulic parameters were updated. Two case studies are presented. Overall, the proposed method yielded parameters and states estimates that were more accurate and more consistent than those obtained when adjusting all the parameters. Furthermore, the results show that the higher the influence of a parameter on the model output under the current conditions, the better the estimation of this parameter is.

This paper presents a scheme for applying two-stage explicit stochastic optimization to seasonal irrigation scheduling. It is assumed that an ensemble of Ns weather forecasts (scenarios) is available. At each decision point during the season up to Ns Nsmulti-objective optimization problems are solved by assuming a specific scenario for the immediate decision period and all possibleNs scenarios for the subsequent periods. The irrigation schedule selected for implementation during the immediate decision period is the one that produces the highest worst-case yield, which mimics the traditional risk-adverse farmers' strategy. The procedure is illustrated for a maize crop at Davis, CA, modeled with DSSAT. The optimization was performed for ten years, using as forecasts the weather recorded on the previous 15 years. The proposed approach yielded consistently results that were very close to truly optimal, i.e. results that could have been obtained if perfect weather forecasts were available at the beginning of the season. These results were better than those obtained with a deterministic approach that relied on the same data and decision rules but used only a single forecast that consisted of the average weather of the 15 previous years. However, these improved results came at the expense of a significant increase of the computation burden. In addition to the overall improved performance in terms of yield, a main advantage of the stochastic approach is that, since the solution for implementation is selected from an ensemble of solutions, it is possible to develop a selection strategy that mimics farmers' traditional selection strategy. This could prove a key factor toward the adoption of decision support tools that involve model-based optimization.

A challenge in modern farming is to find a sustainable way of achieving sufficient production. Precision in dosage, timing and allocation of water, biocides, fertilizer and other inputs is essential, as are such management actions as harvesting, pruning and weeding. Despite the increasing availability of sensor and actuator technologies, decision-making is still largely left to the farmer. This is creating a strong demand for support in operational management. This paper presents an overview of methods involving the use of technology and data to develop model-based management support and automation for productive and input-efficient farming. For each method, the main advantages and drawbacks relating to typical farm characteristics are discussed and summarized. Three case studies are presented, to illustrate the design steps involved in developing a model, observer and controller. The overall design procedure is summarized in a flowchart, and serves as a basic guide for method selection and model development.

Delineation of management zones for applying Variable Rate Irrigation (VRI) in drip irrigation presents unique challenges due to the directionality of the drip lines and the fact that the water flow rate can not be varied along a drip line. This work presents a method for optimal delineation of irrigation zones under such constraints. It is assumed that historical weather records, a soil map, a simulation model that predicts the crop development in response to weather and irrigation, as well as crop and water prices, are available. The first step consists of estimating the water productivity function of the crop for each soil type if managed optimally, using multi-objective optimization. The second step consists of estimating the water productivity function of the crop on each soil if the irrigation is managed according to another soil. Optimal delineation of the irrigation zones is then obtained by solving a non-linear optimization problem in which the decision variables are the positions of the zones boundaries. This involves an intermediate step in which the area of each soil in each irrigation zone is calculated, and the soil according to which the irrigation should be managed in each zone is determined. The procedure is illustrated on two examples involving three soils and a cotton crop in Greece. In both cases, optimal management zones delineation and management was obtained using the average weather of years 1982–1991 and performance was estimated for years 1992–2016. The results show that splitting the field into several management zones increases profit slightly via yield increase and/or water savings, depending on the relative price of water.

Site-specific agricultural management relies on identifying within-field spatial variability and is being used for variable rate input of resources. Precision agricultural management commonly attempts to integrate multiple datasets to determine management zones (MZs), homogenous units within the field, based on spatial characteristics of environmental and crop properties (i.e., terrain, soil, vegetation conditions). This study aims to develop a novel statistical multivariate spatial clustering approach to determine MZs for precision nitrogen fertilization in a citrus orchard along the growing season. Five variables were used to characterize spatial variability (i.e., N spectral index, crop water stress index (CWSI), tree height, elevation, and slope) within four plots based on a monthly thermal and multispectral high-resolution imagery acquired from an unmanned aerial vehicle (UAV). The UAV data was tested against leaf N samplings based on samples taken from 48 trees within the four craters plots, which were selected based on a stratified random design (SRD) model. A Support Vector Machines-Regression (SVM-R) model was applied to develop a prediction N spectral index for canopy N levels. The clustering model included the following components — spatial representation of the data based on Getis Ord Gi*. Then variable weights were assigned based on their relative contribution to principal component analysis. Fuzzy C-means algorithm was applied to the weighted spatial representation and was found to generate spatially continuous and homogeneous MZs with similar numbers of trees. In addition, we analyzed the temporal dynamics in the MZs and clustering patterns throughout the year, using information based on the monthly UAV imagery. Management of the sub-units, or plots, using spatial representation rather than the measured values, is suggested as a more suitable platform for agricultural practices. Future development of fertilization applications for individual trees will require adjusting the statistical approach to support tree-specific management. The proposed model composite is flexible and may be composed of different models and/or variables for developing optimal MZ delineation for specific plots.