Yield Forecasting Research Articles

Transforming Philippine agriculture through Agriculture 4.0

Agriculture 4.0 has paved the way for Philippine agriculture through the One Department of Agriculture (One DA) Agenda. The government is slowly paving its way to Agriculture 4.0 through the support of collaborating countries and agencies. The increasing digital integration and advanced technologies emphasizing precision farming have contributed to the industry's growth. The agriculture, forestry, and fishing sectors contributed 8.9% to the Philippines' GDP (PSA, 2022). Agriculture 4.0 also contributed to the increase in farm mechanization level of the country to 2.679 horsepower per hectare (hp/ha) from 2.31 hp/ha in 2013 (Bagoso, 2023). The government has recently developed a more straightforward approach to farming problems, such as monitoring farm activity through drones, remote sensing, and satellite technologies. The farmers can now access real-time updates on infrastructure and machinery projects from proposal to turning-over through ABEMIS. Using remote sensing and data analytics has become a game changer in agriculture, especially in weather prediction, forecast of future yields, disaster assessment and access to insurance schemes are available with PRiSM and RIICE projects. Drone use in different applications was strengthened through policies set by the DA. The continuous research for development and technology transfer with local government units through province-led agriculture and fisheries extension systems helps the government promote digital transformation in agriculture. Indeed, Agriculture 4.0 is the future of global agriculture and the engine of agriculture modernization in the Philippines.

PREDICTION SYSTEM FOR THE INFLUENCE OF PHOSPHORUS ON WHEAT YIELD: OPTIMAL HYPERPARAMETER SELECTION

Modern studies in the application of machine learning in agrotechnology have showcased significant advancements in accurate crop yield forecasting and agricultural production optimization. This study aimed to select optimal hyperparameters for a neural network designed to forecast the impact of phosphorus on the yield of spring wheat. Tasks to achieve this goal included the selection of optimal neural network hyperparameters and deployment of the prediction model into forecasting system, ensuring accessibility and user convenience. Input dataset consisted of year, region, climatic indicators (temperature of the soil surface, precipitation, humidity), applied phosphorus, and the target variable is the yield of spring wheat. To select the optimal hyperparameters, GridSearchCV was utilized and integrated into the experiment, enabling the achievement of more precise and impactful forecasting outcomes. Experiments aimed at hyperparameter tuning identified an optimal network configuration (3 layers, 32 neurons, 300 epochs, 32 batches). Results from training showcased that this network configuration achieved the best performance, demonstrating minimal mean squared error (MSE). These results introduced a novel scientific approach in training data using this neural network model for forecasting the yield of cereal crops. Developing the prediction system's interface relied on the Streamlit, allowing the creation of an intuitive and user-friendly interface. The visualization derived from this model empowers users to evaluate crop yield and receive recommendations for optimal phosphorus application to enhance spring wheat yield. These results highlighted practical significance of this prediction system, based on neural network utilizing optimal hyperparameters that are ready for practical implementation by system developers.

In-season forecasting of within-field grain yield from Sentinel-2 time series data

Precise knowledge of cropland productivity is relevant for farmers to enable optimizing managing practices; particularly with the perspective of anticipating crop yield ahead of harvest. The current availability of high spatiotemporal resolution Sentinel-2 satellite data offers a unique opportunity to monitor croplands over time. In this context, the recently introduced kernel NDVI (kNDVI) statistically optimizes the conventional NDVI formulation by applying a nonlinear function to the involved bands, and so maximizes the spectral information extraction. This study proposes a workflow for within-field yield forecasting from Sentinel-2 kNDVI time series analysis focusing on winter cereal croplands in Switzerland over three years, comparing with NDVI as baseline. For a temporally continuous modelling of crop yields, Gaussian Process Regression (GPR) was applied to reconstruct cloud-free time series of the complete crop growing seasons. Following, distinct machine learning regression models (GPR, Kernel Ridge Regression and Random Forest) were developed to forecast yield at any point in time throughout the cropland growing season. The integration of Growing Degree Days (GDD) information as temporal spacing reference of the time series considerably improved the accuracy and consistency of in-season yield forecasting. Training and testing within the same year demonstrated that yield can be accurately forecast approximately 2–2.5 months ahead of harvest, at crops’ anthesis (flowering) phase, with an RMSE up to 0.71 t/ha and a relative RMSE of 7.60%. Although the forecasting accuracy of the models decreased when predicting yield for the unseen years, still satisfactory results were obtained: RMSE = 0.97 t/ha, relative RMSE = 11.47%.

Integrating genetic algorithm with ARIMA and reinforced random forest models to improve agriculture economy and yield forecasting

Integrating genetic algorithm with ARIMA and reinforced random forest models to improve agriculture economy and yield forecasting

Deep Learning Model for Classifying and Evaluating Soybean Leaf Disease Damage.

Soybean (Glycine max (L.) Merr.) is a major source of oil and protein for human food and animal feed; however, soybean crops face diverse factors causing damage, including pathogen infections, environmental shifts, poor fertilization, and incorrect pesticide use, leading to reduced yields. Identifying the level of leaf damage aids yield projections, pesticide, and fertilizer decisions. Deep learning models (DLMs) and neural networks mastering tasks from abundant data have been used for binary healthy/unhealthy leaf classification. However, no DLM predicts and categorizes soybean leaf damage severity (five levels) for tailored pesticide use and yield forecasts. This paper introduces a novel DLM for accurate damage prediction and classification, trained on 2930 near-field soybean leaf images. The model quantifies damage severity, distinguishing healthy/unhealthy leaves and offering a comprehensive solution. Performance metrics include accuracy, precision, recall, and F1-score. This research presents a robust DLM for soybean damage assessment, supporting informed agricultural decisions based on specific damage levels and enhancing crop management and productivity.

Spatial correlation-based machine learning framework for evaluating shale gas production potential: A case study in southern Sichuan Basin, China

Assessment of production potential and prediction of sweet spots in unexploited shale gas wells are crucial technologies for achieving a high success rate in drilling. Most existing methods are only applicable for yield forecasting of shale gas wells with known geological and drilling/completion factors after production. In addition, the shale gas reservoir in the Longmaxi Formation in the southern Sichuan Basin of China is characterized by low porosity, low permeability, and diverse gas reservoir states. Thus, forecasting production from unexploited wells and predicting the exact location of the sweet spot under complex geological conditions have become challenging topic for both academia and industry. According to the first law of geography, shale gas wells that are closer in space have more similar geologic features. Based on this foundation, a machine-learning prediction model based on spatial correlation (SC-MLPM) is proposed for evaluating potential production of shale gas wells in the Changning play located in Longmaxi Formation in the southern Sichuan basin, China. Specifically, an improved K-nearest neighbor algorithm (SC-KNN) based on the spatial correlation characteristics of the exploited wells is designed to estimate the relevant geological factors of the unexploited wells in the same area. In addition, an extreme gradient lifting model based on Huber loss function (Hu-XGB) is developed to predict the potential production of unexploited shale gas wells, and then the shale gas sweet spot area in the research area is predicted quickly and accurately. The experimental results on 88 shale gas wells from Changning play show that the mean relative error of the prediction results of the proposed method is 9.8%. Furthermore, the results also determine that the sweet spots of shale gas wells in the research area are mainly developed in the northwest and northeast directions, and gradually become worse in the southeast direction. Eventually, experimental results for the Fox Creek play in Canada show that the proposed model obtains better predictive performance as the number of gas wells increases, and also verifies the generalizability. This work provides a basis for the efficient development of the research area.

Large-scale spatio-temporal yield estimation via deep learning using satellite and management data fusion in vineyards

Crop yield forecasting is essential for informed farm management decisions. However, most yield forecasting models have low spatial resolution, late-season predictions, and lack validation for unseen years or locations. These limitations likely stem from the scarcity of large-scale, high-resolution yield measurement data collected over multiple years, which is uncommon in commercial specialty crop operations. Additionally, these limitations raise concerns about the models’ utility and generalizability under new environmental and management conditions within or across farms. In this study, we develop a spatio-temporal deep learning model to forecast wine grape yield early in the season, utilizing a large dataset with high spatio-temporal resolution (i.e., yield data from ∼5 million grapevines of eight cultivars observed over four years). The model is trained on weekly 10 m RGB-NIR time-series satellite data from Sentinel 2A-B, fused with categorical and continuous management inputs, including cultivar type, trellis type, row spacing, and canopy spacing. We assess the model’s generalizability by examining its performance on data from unseen years and/or locations and at multiple spatial resolutions. Our results show that combining management data with satellite imagery significantly improves model performance on entirely unseen vineyard blocks at 10 m resolution, achieving an R2 of 0.76, a mean absolute error (MAE) of 4.21 tonnes/hectare, and a mean absolute percent error (MAPE) of 13%. We find that cultivars with considerable year-to-year yield variability tend to exhibit lower predictive performance and may benefit from longer time-series observations for model training to encompass a wide range of environmental and management conditions. We also observe improved estimations from the early season in April to the middle of the growing season in June. In conclusion, the yield forecasting and model validation framework established in this study lays the foundation for training spatio-temporally aware deep learning models on exceptionally large yield datasets with high spatio-temporal resolution. We anticipate that such models will become increasingly prevalent as yield monitors are more frequently deployed in specialty crop operations.

Tracing the Evolution and Legacy of Professor A.P. Iofinov’s Scientific School

The scientific ideas and contribution of Professor A.P. Iofinov to the development of agricultural science remain relevant in our time. The paper emphasizes Professor August Pavlovich Iofinov’s pivotal role in the exploration of quality management and the modeling of technological processes for agricultural machines. The scientific concepts introduced by A.P. Iofinov continue to hold significance in contemporary times. Their development and application contribute to improving the operational capabilities of agricultural machinery and bolstering the overall efficiency of the agricultural industry. (Research purpose) The study seeks to consolidate historical and scientific materials about the activities of professor A.P. Iofinov and the creation of a scientific school under his leadership at the Department of Agricultural Machinery of the Bashkir State Agrarian University. (Materials and methods) The establishment and evolution of the scientific school are examined through the problem-chronological method. Scientific issues articulated by Professor A.P. Iofinov, are presented chronological sequence of their progression, moving from hypothesis to practical testing and eventual generalization in scientific works. (Results and discussion) The paper chronicles the founding and progression of the scientific school affiliated with the Department of Agricultural Machines at the Bashkir State Agrarian University, guided by Professor A.P. Iofinov. The established scientific school delved into significant research domains, including the statistical dynamics of agricultural machinery, quality management of agricultural machinery, modeling of technological processes, and yield forecasting. At present, these research areas are experiencing renewed development, with new tasks being formulated to align with contemporary trends and the latest achievements in science and technology. (Conclusions) Professor A.P. Iofinov’s works and his scientific school have given rise to the development of new algorithms and computer programs essential for automating the control processes of agricultural machinery and forecasting yields. Furthermore, they have played a crucial role in the creation of digital twins for agricultural machines, based on models of technological processes. This advancement significantly enhances the precision and efficiency in resource utilization, improves the quality of work, and boosts productivity in agriculture.

Apple (Malus domestica) and pear (Pyrus communis) yield prediction after tree image analysis

Yield forecasting depends on accurate tree fruit counts and mean size estimation. This information is generally obtained manually, requiring many hours of work. Artificial vision emerges as an interesting alternative to obtaining more information in less time. This study aimed to test and train YOLO pre-trained models based on neural networks for the detection and count of pears and apples on trees after image analysis; while also estimating fruit size. Images of trees were taken during the day and at night in apple and pear trees while fruits were manually counted. Trained models were evaluated according to recall, precision and F1score. The correlation between detected and counted fruits was calculated while fruit size estimation was made after drawing straight lines on each fruit and using reference elements. The precision, recall and F1score achieved by the models were up to 0.86, 0.83 and 0.84, respectively. Correlation coefficients between fruit sizes measured manually and by images were 0.73 for apples and 0.80 for pears. The proposed methodologies showed promising results, allowing forecasters to make less time consuming and accurate estimates compared to manual measurements. Highlights: The number of fruits in apple and pear trees, could be estimated from images with promising results. The possibility of estimating the fruit numbers from images could reduce the time spent on this task, and above all, the costs. This allow growers to increase the number of trees sampled to make yield forecasts.

On the issue of adaptive landscape development of useless lands based on the results of barley yield forecasting

The research aimed to assess the suitability of useless lands of a particular farm for growing barley in pure and cover crops based on the analysis of data from long-term monitoring of its yield at an agro-ecological test site. The work used data from long-term (1997-2012) monitoring of the yield of barley variety Gonor in pure and cover (allowing for additional production) crops at the Gubino agricultural site of All- Russian Research Institute of Reclaimed Lands. The agricultural site is located 4 km east of the city of Tver, on a hill with a relative height of 15 m, consisting of a flat top, a northern gentle slope with a steepness of 2-3о, a southern slope (3-5 о) and interhill depressions (northern and southern). Using statistical and mathematical modelling, patterns of barley yield formation were identified. Based on the obtained formulas and archival data on the state of useless lands, forecast crop yield surfaces were calculated for the entire territory of the farm. Barley in pure crops responds only to the height of the location, while in cover crops its yield depends on the height and steepness of the surface, as well as the phosphorus content of the soil. A quarter of the farm's area (the upper parts of the hills in the west and south) is potentially capable of providing barley yields in pure crops from 1.4 to 3.3 t/ha, and an increased yield of this crop in cover crops (1.1-3.4 t/ha) can be obtained only on 20 % of the territory occupying the transition zones between moraine and valley landscapes. Using only archival data is not always justified - different approaches to the design of farming systems must complement each other, and to clarify the yield forecast, it is necessary to conduct special landscape and soil studies.

Hybrid time series models with exogenous variable for improved yield forecasting of major Rabi crops in India

Accurate and in-time prediction of crop yield plays a crucial role in the planning, management, and decision-making processes within the agricultural sector. In this investigation, utilizing area under irrigation (%) as an exogenous variable, we have made an exertion to assess the suitability of different hybrid models such as ARIMAX (Autoregressive Integrated Moving Average with eXogenous Regressor)–TDNN (Time-Delay Neural Network), ARIMAX–NLSVR (Non-Linear Support Vector Regression), ARIMAX–WNN (Wavelet Neural Network), ARIMAX–CNN (Convolutional Neural Network), ARIMAX–RNN (Recurrent Neural Network) and ARIMAX–LSTM (Long Short Term Memory) as compared to their individual counterparts for yield forecasting of major Rabi crops in India. The accuracy of the ARIMA model has also been considered as a benchmark. Empirical outcomes reveal that the ARIMAX–LSTM hybrid modeling combination outperforms all other time series models in terms of root mean square error (RMSE) and mean absolute percentage error (MAPE) values. For these models, an average improvement of RMSE and MAPE values has been observed to be 10.41% and 12.28%, respectively over all other competing models and 15.83% and 18.42%, respectively over the benchmark ARIMA model. The incorporation of the area under irrigation (%) as an exogenous variable in the ARIMAX framework and the inbuilt capability of the LSTM model to process complex non-linear patterns have been observed to significantly enhance the accuracy of forecasting. The performance supremacy of other hybrid models over their individual counterparts has also been evident. The results also suggest avoiding any performance generalization of individual models for their hybrid structures.

Model pruning enables localized and efficient federated learning for yield forecasting and data sharing

Federated Learning (FL) presents a decentralized approach to model training in the agri-food sector and offers the potential for improved machine learning performance, while ensuring the safety and privacy of individual farms or data silos. However, the conventional FL approach has two major limitations. First, the heterogeneous data on individual silos can cause the global model to perform well for some clients but not all, as the update direction on some clients may hinder others after they are aggregated. Second, it is lacking with respect to the efficiency perspective concerning communication costs during FL and large model sizes. This paper proposes a new technical solution that utilizes network pruning on client models and aggregates the pruned models. This method enables local models to be tailored to their respective data distribution and mitigate the data heterogeneity present in agri-food data. Moreover, it allows for more compact models that consume less data during transmission. We experiment with a soybean yield forecasting dataset and find that this approach can improve inference performance by 15.5% to 20% compared to FedAvg, while reducing local model sizes by up to 84% and the data volume communicated between the clients and the server by 57.1% to 64.7%. Our method demonstrates the potential to use efficient models that are more environmentally friendly to support the agri-food sector’s transition to net zero. Future enhancements of this method could further optimize distributed learning in agri-food, enhancing sustainability and applicability.

Machine learning for crop yield forecasting

Amid the persistent rise in global population, there has been a heightened focus on food security by academia, governmental initiatives, and international endeavors. Food security serves as a critical pillar in the national security framework, contributing to a nation’s sovereignty and self-sufficiency in food supply. To fulfill global requirements for essential food items, there is an imperative need to enhance agricultural efficiency across countries. Concurrently, agricultural practices must align with contemporary quality standards and meet consumer needs, drawing upon an integrated approach to crop cultivation technologies and yield classifications. Methodologies and tools for yield augmentation, grounded in scientific advancements in predictive modeling, are of paramount importance. Investigating the plethora of variables that contribute to optimal crop development, which in turn influences yield, poses significant challenges. Comprehensive inquiries that incorporate cutting-edge scientific and technological methodologies are essential for creating precise yield forecasts. The evolving landscape of yield modeling and prediction has emerged as a technologically sophisticated domain. Advanced methods such as machine learning and deep learning offer robust platforms for addressing crop yield forecasting, particularly when coupled with extensive datasets on environmental variables. A growing body of literature suggests the promising role of computational technologies and machine learning paradigms, inclusive of various forms of remote sensing data, in fine-tuning yield models. Yield prediction models are often characterized by intricate nonlinear equations influenced by a range of factors: seed quality and diversity, soil attributes, climatic variables, fertilizer usage, and other agronomic practices. The impacts of these variables on crop yield are varied, with some exerting greater influence than others. Additionally, crop yield is susceptible to adverse environmental and climatic conditions. While there exists a rich corpus of research on yield forecasting, addressing this issue remains an exigent priority in the agricultural sector.

Spatio-temporal multi-level attention crop mapping method using time-series SAR imagery

Accurate crop mapping is of great significance for crop yield forecasting, agricultural productivity development and agricultural management. Thanks to its all-time and all-weather capability, integrating multi-temporal synthetic aperture radar (SAR) for crop mapping has become essential and challenging task in remote sensing. In recent years, deep learning (DL) has demonstrated excellent crop mapping accuracy to interpret crop dynamics. However, existing DL-based methods tend to be incapable of capturing spatial and temporal features at different scales simultaneously, and this often leads to severe mis-classification due to the complex and heterogeneous distribution of crops and diverse phenological patterns. In this paper, we propose a novel spatio-temporal multi-level attention method, named as STMA, for crop mapping using time-series SAR imagery in an end-to-end fashion to increase the capability of crop phenology retrieval. Specifically, the multi-level attention mechanism is designed to aggregate multi-scale spatio-temporal representations on crops via cascaded spatio-temporal self-attention (STSA) and multi-scale cross-attention (MCA) modalities. To ensure a fine extraction of multi-granularity features, a learnable spatial attention position encoding is proposed to adaptively generate the position priors to facilitate multi-level attention learning. Experimental results on Brandenburg Sentinel-1 dataset, public PASTIS-R dataset and South Africa dataset demonstrated that STMA can achieve state-of-the-art performance in crop mapping tasks, with the accuracy of 96.54% in the Brandenburg Sentinel-1 dataset, 86.77% in the PASTIS-R dataset and 83.37% in the South Africa dataset, validating its effectiveness and superiority. Further comparison of spatio-temporal generalization capability reflected its excellent performance in spatio-temporal modeling on different crops and scenarios. This research provides a viable and intelligent spatio-temporal framework for large-area crop mapping using time-series SAR imagery in complex agricultural systems. The Brandenburg Sentinel-1 dataset and the STMA code will be publicly available at https://github.com/hanzhu97702/ISPRS_STMA.

High spatial resolution seasonal crop yield forecasting for heterogeneous maize environments in Oromia, Ethiopia

Seasonal climate variability determines crop productivity in Ethiopia, where rainfed smallholder farming systems dominate in the agriculture production. Under such conditions, a functional and granular spatial yield forecasting system could provide risk management options for farmers and agricultural and policy experts, leading to greater economic and social benefits under highly variable environmental conditions. Yet, there are currently only a few forecasting systems to support early decision making for smallholder agriculture in developing countries such as Ethiopia. To address this challenge, a study was conducted to evaluate a seasonal crop yield forecast methodology implemented in the CCAFS Regional Agricultural Forecasting Toolbox (CRAFT). CRAFT is a software platform that can run pre-installed crop models and use the Climate Predictability Tool (CPT) to produce probabilistic crop yield forecasts with various lead times. Here we present data inputs, model calibration, evaluation, and yield forecast results, as well as limitations and assumptions made during forecasting maize yield. Simulations were conducted on a 0.083° or ∼ 10 km resolution grid using spatially variable soil, weather, maize hybrids, and crop management data as inputs for the Cropping System Model (CSM) of the Decision Support System for Agrotechnology Transfer (DSSAT). CRAFT combines gridded crop simulations and a multivariate statistical model to integrate the seasonal climate forecast for the crop yield forecasting. A statistical model was trained using 29 years (1991–2019) data on the Nino-3.4 Sea surface temperature anomalies (SSTA) as gridded predictors field and simulated maize yields as the predictand. After model calibration the regional aggregated hindcast simulation from 2015 to 2019 performed well (RMSE = 164 kg/ha). The yield forecasts in both the absolute and relative to the normal yield values were conducted for the 2020 season using different predictor fields and lead times from a grid cell to the national level. Yield forecast uncertainties were presented in terms of cumulative probability distributions. With reliable data and rigorous calibration, the study successfully demonstrated CRAFT’s ability and applicability in forecasting maize yield for smallholder farming systems. Future studies should re-evaluate and address the importance of the size of agricultural areas while comparing aggregated simulated yields with yield data collected from a fraction of the target area.

Forecasting carrot yield with optimal timing of Sentinel 2 image acquisition

Accurate, non-destructive forecasting of carrot yield is difficult due to its subterranean growing habit. Furthermore, the timing of forecasting usually occurs when the crop is mature, limiting the opportunity to implement alternative management decisions to improve yield (during the growing season). This study aims to improve the accuracy of carrot yield forecasting by exploring time series and multivariate approaches. Using Sentinel-2 satellite imagery in three Australian vegetable regions, we established a time series of carrot phenological stages (PhS) from ‘days after sowing’ (DAS) to enhance prediction timing. Numerous vegetation indices (VIs) were analyzed to derive temporal growth patterns. Correlations with yield at different PhS were established. Although the average root yield (t ha−1) did not significantly differ across the regions, the temporal VI signatures, indicating different regional crop growth trends, did vary as well as the PhS at when the maximum correlation with yield occurred (PhSR2max)\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$PhS_{{R2_{max} }} )$$\\end{document} with two of the regions producing a delayed PhSR2max\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$PhS_{{R2_{max} }}$$\\end{document} (i.e. 90–130 DAS). The best multivariate model was identified at 70 DAS, extending the forecasting window before harvest between 20 to 60 days. The performance of this model was validated with new crops producing an average error of 16.9 t ha−1 (27% of total yield). These results demonstrate the potential of the model at such early stage under varying growing conditions offering growers and stakeholders the chance to optimize farming practices, make informed decisions on selling, harvesting, and labor planning, and adopt precision agriculture methods.

Enhancing state and parameter estimations of a dynamic crop model by a recombination particle filter

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.

Integrating Temporal Fluctuations in Crop Growth with Stacked Bidirectional LSTM and 3D CNN Fusion for Enhanced Crop Yield Prediction

Optimizing farming methods and guaranteeing a steady supply of food depend critically on accurate predictions of crop yields. The dynamic temporal changes that occur during crop growth are generally ignored by conventional crop growth models, resulting in less precise projections. Using a stacked bidirectional Long Short-Term Memory (LSTM) structure and a 3D Convolutional Neural Network (CNN) fusion, we offer a novel neural network model that accounts for temporal oscillations in the crop growth process. The 3D CNN efficiently recovers spatial and temporal features from the crop development data, while the bidirectional LSTM cells capture the sequential dependencies and allow the model to learn from both past and future temporal information. Our model's prediction accuracy is improved by combining the LSTM and 3D CNN layers at the top, which better captures temporal and spatial patterns. We also provide a novel label-related loss function that is optimized for agricultural yield forecasting. Because of the relevance of temporal oscillations in crop development and the dynamic character of crop growth, a new loss function has been developed. This loss function encourages our model to learn and take advantage of the temporal trends, which improves our ability to estimate crop yield. We perform comprehensive experiments on real-world crop growth datasets to verify the efficacy of our suggested approach. The outcomes prove that our unified strategy performs far better than both baseline crop growth prediction algorithms and cutting-edge applications of deep learning. Improved crop yield prediction accuracy is achieved with the integration of temporal variations via the merging of bidirectional LSTM and 3D CNN and a unique loss function. This study helps move the science of estimating crop yields forward, which is important for informing agricultural policy and ensuring a steady supply of food.

Comparison of Phenological Weather indices based Statistical and Machine Learning Models for Soybean Yield Forecasting in Pantnagar, Uttarakhand

Comparison of Phenological Weather indices based Statistical and Machine Learning Models for Soybean Yield Forecasting in Pantnagar, Uttarakhand

Prediction analysis of cotton yield based on BP neural network and multiple linear regression

With the rapid development of the economy and agricultural production, cotton is an indispensable crop in people's production and life, and the average annual cotton yield is particularly significant in production. The unit yield of cotton is affected by many factors, so it is extremely important to analyze the factors affecting future cotton yield forecast. This article varies seven relevant indicators, including fertilizer cost,consumption of agricultural plastic, pesticide cost, seed cost, irrigation cost, chlorpyrifos control area, and labor input, and establishes a prediction model based on Back Propagation (BP) neural network. The model uses the function approximation ability of BP neural network and gradient descent algorithm to predict the nonlinear functional relationship between the unit output and the seven factors. Meanwhile, a multiple linear regression model is used to forecast the unit yield and the prediction accuracy of the two methods for unit output is compared. The experimental results show that compared with multiple linear regression, BP neural network has higher accuracy in predicting cotton unit yield, which can be used as an idea for future prediction.