Yield Model Research Articles

Integrated Iot Approaches for Crop Recommendation and Yield-Prediction Using Machine-Learning

In this study, we present an integrated approach utilizing IoT data and machine learning models to enhance precision agriculture. We collected an extensive IoT secondary dataset from an online data repository, including environmental parameters such as temperature, humidity, and soil nutrient levels, from various sensors deployed in agricultural fields. This dataset, consisting of over 1 million data points, provided comprehensive insights into the environmental conditions affecting crop yield. The data were preprocessed and used to develop predictive models for crop yield and recommendations. Our evaluation shows that the LightGBM, Decision Tree, and Random Forest classifiers achieved high accuracy scores of 98.90%, 98.48%, and 99.31%, respectively. The IoT data collection enabled real-time monitoring and accurate data input, significantly improving the models’ performance. These findings demonstrate the potential of combining IoT and machine learning to optimize resource use and improve crop management in smart farming. Future work will focus on expanding the dataset to include more diverse environmental factors and exploring the integration of advanced deep learning techniques for even more accurate predictions.

Prediction of Maize Crop Yield Using Principal Component Analysis of Weather Parameters

The use of principal component analysis in the development of statistical models for crop yield forecasting has been demonstrated. Maize crop yield data for a period of 21 years (2001-2021) were drawn from the Dacnet website and the weather data were collected from the Meteorological Observatory, Department of Agrometeorology, College of Agriculture, G.B. Pant University of Agriculture and Technology Pantnagar, Uttarakhand. Maximum temperature, Minimum temperature, Relative Humidity A.M, Relative Humidity P.M, Total rainfall, Sunshine hours, Wind velocity and Evapotranspiration were the weather parameters considered for the study. Out of the 21-year data, 17-year data were used for training the model while remaining 4 years data were used for testing the model. Weekly data on weather variables was used to create weather indices [1]. This work involves developing forecasting models using Principal Component Analysis (PCA) and Multiple Linear Regression (MLR). Applied to extract principal components (PCs) from the correlation matrix of predictors. Five models (Model 1 to Model 5) are developed using MLR, with varying numbers of PCs as regressors which also include time trend and maize yield as dependent variable. The model performance was measured using Adjusted R-squared (adj R2) and Root Mean Square Error (RMSE) as goodness of fit criteria. On the basis of adj R2 and RMSE, model 1 which includes all the calculated weather indices, was found to be best suited model with high adj R2 (74.18 %) and least RMSE (276.36). Hence, this model can be used to forecast maize yield for the studied region.

Increasing bio-hydrogen production from microbial electrolysis cell using artificial gorilla troops optimization

BackgroundThe target of this paper is to improve the performance of the microbial electrolysis cell (MEC). The performance of MEC including bio-hydrogen production and energy recovery is depending on the values of three controlling parameters including buffer concentration, dilution factor, and applied voltage.ProblemTherefore, defining the optimal values of three controlling parameters is the challenge of the work.MethodologyIn this paper the artificial gorilla troops optimization has been combined with and ANFIS modelling to increase the bio-hydrogen production from MEC. At first, using measured data, a model is created to simulate the MEC in terms of three controlling parameters. Then, for first time, an artificial gorilla troops optimization (AGTO) has been used to determine the optimal values of buffer concentration, dilution factor, and applied voltage to boost simultaneously bio-hydrogen production and energy recovery of MEC. To demonstrate the superiority of integration between ANFIS modelling and AGTO, the obtained results are compared with RSM methodology, and artificial neural network integrated with particle swarm optimization.FindingsFor hydrogen yield model, the RMSE lowered from 67.5 using RSM to 5.562 using ANFIS (decreased by 91.7%) as compared to RSM. The R-square for prediction rises from 0.94 (using RSM) to 0.99 (using ANFIS) by about 5.32%. For the ANFIS model of energy recovery, the RMSE decreased from 31.7 to 2.83 utilising ANFIS, a decrease of 91%. The R-square for prediction rises from 0.95 (using RSM) to 0.986 (using ANFIS) by about 3.8%. Compared with measured data, the integration between ANFIS and AGTO succeed to increase the hydrogen yield from 576.3 mL/g-VS to 843.32 mL/g-VS. in sum, the total performance of the MEC has been increased by 34.74%, 29.9% and 24.38% respectively compared to measured data, RSM and ANN-PSO.

Comparative evaluation of statistical and machine learning models for weather-driven wheat yield forecasting across different districts of Punjab

Comparative evaluation of statistical and machine learning models for weather-driven wheat yield forecasting across different districts of Punjab

Particle Properties and Flotation Characteristics of Difficult-to-Float Lean Coal

The flotation effect of lean coal is crucial for its clean utilization. Therefore, the flotation characteristics of difficult-to-float lean coal were studied. The analysis results of the feed properties showed that the ash content of the feed was high and the particle size was very fine. The minerals in the gangue mainly included sericite, kaolinite, quartz, white mica, and other substances. After flotation, the functional groups of the coal particles in the tailings decreased, and the absorption peak intensity weakened. Furthermore, the results of multi-factor flotation experiments showed that the dosages of the collector and the frother were significant factors affecting the yield of clean coal. The clean coal yield gradually increased with an increase in the two factors. The ash content of the clean coal increased with an increase in the frother dosage. Within the range of feed concentrations used in this work, the feed concentration was not a significant factor affecting the clean coal’s yield and ash content. Prediction models for the clean coal yield and ash content were proposed. Under optimized experimental conditions, the clean coal yield and the flotation perfection index were 72.15% and 46.63%, respectively, indicating a good flotation effect.

A study on anisotropic hardening of 7075 aluminum alloy based on non-associated flow rules

The accurate description of anisotropic plastic deformation is key to accurately predicting the stamping forming of metal sheets. The anisotropic yield criterion, when based on the assumption of isotropic hardening, often leads to significant inaccuracies. To address this issue, this paper observes the anisotropic hardening phenomenon through non-associated flow rules and evaluates the anisotropy of 7075-O aluminum alloy. Through a series of tensile tests, we determined the mechanical properties of 7075-O aluminum alloy in three distinct orientations. To describe the metal hardening behavior, we employed the Swift-Voce hardening criterion. From the hardening curves in three different directions, it was found that AA7075-O exhibits plastic anisotropy. Based on the VUMAT subroutine, finite element simulation of AA7075-O tensile tests was conducted through Abaqus. Compared with the Hill48 model, it was found that the simulated values of the S–Y2009 anisotropic hardening model have a higher degree of agreement with the experimental curves. The S–Y2009 anisotropic hardening model was adopted to predict the earing behavior of AA7075-O during circular cup deep drawing. The root mean square error between the predicted values of the S–Y2009 model and the experimental values was only 0.1795, which is far smaller than that of the Hill48 yield model. Therefore, the SY2009 model has important guiding significance for the stamping forming of metal sheets.

Short Fatigue-Crack Growth from Crack-like Defects under Completely Reversed Loading Predicted Based on Cyclic R-Curve.

Understanding short fatigue-crack propagation behavior is inevitable in the defect-tolerant design of structures. Short cracks propagate differently from long cracks, and the amount of crack closure plays a key role in the propagation behavior of short cracks. In the present paper, the buildup of fatigue-crack closure due to plasticity with crack extension from crack-like defects is simulated with a modified strip yield model, which leaves plastic stretch in the wake of the advancing crack. Crack-like defects are assumed to be closure-free and do not close even under compression. The effect of the size of crack-like defects on the growth and arrest of short cracks was systematically investigated and the cyclic R-curve derived. The cyclic R-curve determined under constant amplitude loading of multiple specimens is confirmed to be independent of the initial defect length. Load-shedding and ΔK-constant loading tests are employed to extend the cyclic R-curve beyond the fatigue limit determined under constant amplitude loading. The initiation stage of cracks is taken into account in R-curves when applied to smooth specimens.

Enhanced Tobacco Yield Prediction Using Spatial Information and Exogenous Variable-driven Machine Learning Models

Remote sensing technology has been essential in studying the relationship between tobacco canopy spectral characteristics and biomass yield. This study has been conducted in Garnepudi, Andhra Pradesh, employed satellite imagery obtained between 2015 and 2023 to extract vegetation indices (VI’s). Accurately predicting yield is crucial for India's economy. This study investigates the efficacy of various predictive models for tobacco yield forecasting using multiple vegetation indices: Normalized Difference Vegetation Index (NDVI), Green Normalized Difference Vegetation Index (GNDVI), Soil Adjusted Vegetation Index (SAVI), Modified Soil Adjusted Vegetation Index (MSAVI), Leaf Area Index (LAI) and Leaf Surface Water Index (LSWI). The models assessed include traditional parametric approaches (ARIMAX, MLR), machine learning techniques (ANN, SVR, RFR), and advanced ensemble methods like XGBoost. The results highlight XGBoost as the most accurate model, consistently delivering the lowest error metrics, including RMSE and MAE, across all vegetation indices. Specifically, XGBoost achieved the best performance with LAI showing RMSE of 86.657, MAE of 58.324, sMAPE of 14.354, MASE of 1.001, and QL of 29.162 respectively. They exhibited lower error metrics, as compare to the statistical and ML models underscoring their effectiveness and potential in tobacco yield prediction. This study highlights the significant role of remote sensing technology in capturing crop development patterns and accurately forecasting tobacco yield, thereby offering valuable insights for agricultural planning and decision-making. The study also addresses challenges such as data quality and model generalization, providing a comprehensive view of the research impact and future directions.

Detrital K-feldspar as a novel archive of continental crustal evolution using coupled in situ Rb[sbnd]Sr dating and Pb isotope analysis

Here we present in situ RbSr and Pb isotopic analysis of K-feldspar from the Shap Granite, (Cumbria, UK) and detrital K-feldspar from the associated Shap Wells Conglomerate Formation, to investigate the potential of the detrital K-feldspar record as an archive of crustal composition. We use a unique, collision cell – multi collector inductively coupled plasma tandem mass spectrometer to make the in situ isotope measurements, using ‘chemical resolution’ of 87Rb+ and 204Hg+ isobaric interferences on 87Sr+ and 204Pb+, achieved with SF6 and NH3 reaction gases, respectively. Using this technique, we successfully demonstrate that the majority of detrital K-feldspar RbSr ages, as well as initial Sr and Pb isotope ratios, from the Shap Wells Conglomerate are within uncertainty of those for parental Shap Granite K-feldspar (398 ± 1 Ma, n = 6). Some detrital K-feldspar grains record perturbed RbSr ages, reflecting post depositional alteration events. The majority of samples with disturbed RbSr ages carry distinct petrographic features evident in SEM imaging, prior to in situ dating, which indicates a means to help screen for reliable ages.We further illustrate the potential value of combined RbSr and Pb isotopic information preserved in K-feldspar to estimate the timing and chemical composition of juvenile continental crust formed in the past. Where UPb has been fractionated during juvenile crust formation, the initial Pb isotope ratio provided by the K-feldspar yields model crustal U/Pb (diagnostic of tectonic setting) and extraction age. The contrast in the U/Pb composition of juvenile crust formed in intraplate or subduction settings allows the model U/Pb ratios of the detrital K-feldspars to be used to constrain the style of juvenile crust formation. Additionally, the initial 87Sr/86Sr isotope ratio measured in detrital K-feldspar can be combined with the calculated crustal extraction age and RbSr K-feldspar age to model the Rb/Sr of the juvenile crustal protolith.

Predicting dryland winter wheat yield in cold regions of Iran

AbstractWheat is a crucial staple worldwide, serving both human and animal needs. In Iran, where climate conditions vary widely, wheat farming faces significant challenges, especially in areas facing freezing winters and unfavorable temperatures during reproductive stages. Unfortunately, existing models often fail to account these extreme and specific climate conditions, leading to inaccurate predictions, notably in cold areas. To address this issue, wheat dryland farming (WDF) model was evaluated in predicting dryland winter wheat yields in five distinct areas including Shahrekord, Borujen, Koohrang, Farsan, Lordegan, and Ardal in Chahar‐Mahal and Bakhtiari province, Iran. The results showed that changes in precipitation and temperature significantly impacted dryland wheat production. While higher precipitation generally associates with higher yields, this relationship is not always straightforward due to factors like unfavorable precipitation patterns and types (i.e., rainfall or snow). Likewise, unfavorable temperatures, particularly during crucial growth stages and winter freezes, pose significant challenges to wheat growth and yield modeling. The WDF model's performance was evaluated across various temperature conditions in the study area, and it was more accurate in regions with certain minimum and maximum temperature values above thresholds. However, the model performance was poor in colder areas, where freezing temperatures were occurred in winter duration (Shahrekord, Borujen, Koohrang, and Farsan). In order to improve the model's accuracy, a correction factor based on the minimum and maximum air temperatures was incorporated in the model. The findings emphasized the importance of considering both precipitation and temperature dynamics when modeling winter wheat yields, especially in regions with diverse climates. By refining models like WDF, agricultural planners can better forecast the yield fluctuations and address the impacts of climate variability on food security in Iran and similar regions worldwide.

Data-driven modeling of bio-oil yield in agricultural biomass pyrolysis with machine learning

Abundant amounts of solid waste are produced due to the processing of various agricultural goods, such as garbage incineration. This agricultural waste management strategy causes a significant increase in air pollution and environmental deterioration. Several methods are widely applied to treat agricultural waste; the pyrolysis process, which includes burning biomass feedstocks without oxygen, is one of the most recommended. Pyrolysis of agricultural waste can produce biochar, bio-oil, and syngas. Pyrolysis is a complicated procedure that depends on feedstock characteristics such as feed composition (i.e., carbon, hydrogen, and oxygen elemental composition), fixed carbon, and volatile matter. Investigation of the individual property of the feedstock on the overall product yield by experimental or computational study is laborious and time-consuming. Therefore, the machine learning technique is applied in this paper to develop a robust and accurate model to predict the agricultural waste bio-oil yield. A total of 387 data points are gathered from various literature sources. This paper utilizes the proximate analysis, ultimat analysis of the agricultural feedstock, and pyrolysis reaction conditions as input features. Eight different machine-learning algorithms, i.e. adaptive boosting, artificial neural network, categorical boosting, decision tree with bagging, k nearest neighbor, light gradient boosting, random forest, and extreme gradient boosting are deployed for this objective. In addition, a SHAP analysis with a bar plot and a beeswarm plot is built to identify the input variables that had the greatest impact on bio-oil production. The categorical boost model gave the lowest root mean squared error of 2.65 out of all the models and had the greatest R2 value of 93.2% on the testing dataset. The SHAP analysis of the developed model for identifying the feature importance and the beeswarm plot is also performed, which shows that for the pyrolysis of agricultural waste, the hydrogen content is one of the most crucial parameters for the maximization of bio-oil yield.

Integrated modeling of anisotropic neutron yields of classical and spherical tokamaks

Estimations of counting rates of neutron spectrometers in experiments on controlled fusion with magnetic confinement, as well as calculations of energy resolved flux densities of fusion neutrons from plasma to the walls of a reactor require spatial integration of the local, usually anisotropic function of the neutron source. The integrated modeling consists of three main stages. First, sources of fast particles in beam- or wave-heated plasma are calculated. The next stage deals with spatial, energetic, and angular velocity distributions of plasma ions. Finally, double differential rate coefficients of nuclear fusion reactions are computed. This article describes calculations of spatial distributions of nuclear fusion reaction rates in classical and spherical tokamaks and the anisotropy of the neutron yield and spectra. The results are based on analytical formulas for energetic and angular distributions of the local source of fusion products in plasma. Examples of energetic spectral densities of neutron fluxes on first walls are presented, as well as energy resolved counting rates of collimated neutron spectrometers for perpendicular and tangential lines of sight.

Mesoscale mechanism of damage in fracture process zone of CFRP laminates simulated with triaxial stress state-dependent constitutive equation of matrix resin

The three-dimensional failure process experimentally observed by synchrotron radiation X-ray computed tomography (SR X-CT) regarding the influence of the interfiber distance is discussed on the basis of the results of numerical experiments. Triaxial stress states in the fracture process zone of carbon fiber reinforced polymers were analyzed on the mesoscale under mode I and mixed-mode (mode I + II) loading. Yield and damage models depending on stress triaxiality were used to accurately simulate three-dimensional stress states in the damage zone around the crack tip. Owing to the heterogeneity of composites, deviatoric stress is prominent in the thin resin region where the interfiber distance is small under mode I loading. On the other hand, matrix resin is triaxially stressed in the middle point between carbon fibers in the thick resin region where the interfiber distance is large. Under mode II loading, the shapes of fiber/matrix debonding depended on the interfiber distance. Areas with stress concentration were found owing to a large debonding area in the thick resin region resulting in a matrix cracking-prone stress state. These findings explain the damage and failure processes well observed by SR X-CT and provide a fundamental understanding of the damage mechanism at a mesoscale.

Improved Winter Wheat Yield Estimation by Combining Remote Sensing Data, Machine Learning, and Phenological Metrics

Accurate yield prediction is essential for global food security and effective agricultural management. Traditional empirical statistical models and crop models face significant limitations, including high computational demands and dependency on high-resolution soil and daily weather data, that restrict their scalability across different temporal and spatial scales. Moreover, the lack of sufficient observational data further hinders the broad application of these methods. In this study, building on the SCYM method, we propose an integrated framework that combines crop models and machine learning techniques to optimize crop yield modeling methods and the selection of vegetation indices. We evaluated three commonly used vegetation indices and three widely applied ML techniques. Additionally, we assessed the impact of combining meteorological and phenological variables on yield estimation accuracy. The results indicated that the green chlorophyll vegetation index (GCVI) outperformed the normalized difference vegetation index (NDVI) and enhanced vegetation index (EVI) in linear models, achieving an R2 of 0.31 and an RMSE of 396 kg/ha. Non-linear ML methods, particularly LightGBM, demonstrated superior performance, with an R2 of 0.42 and RMSE of 365 kg/ha for GCVI. The combination of GCVI with meteorological and phenological data provided the best results, with an R2 of 0.60 and an RMSE of 295 kg/ha. Our proposed framework significantly enhances the accuracy and efficiency of winter wheat yield estimation, supporting more effective agricultural management and policymaking.

Feasibility of machine learning-based rice yield prediction in India at the district level using climate reanalysis and remote sensing data

CONTEXTYield forecasting, the science of predicting agricultural productivity before the crop harvest occurs, helps a wide range of stakeholders make better decisions around agricultural planning. OBJECTIVEThis study aims to investigate whether machine learning-based yield prediction models can capably predict Kharif season rice yields at the district level in India several months before the rice harvest takes place. METHODOLOGYThe methodology involved training 19 machine learning models such as CatBoost, LightGBM, Orthogonal Matching Pursuit, and Extremely Randomized Trees on 20 years of climate, satellite, and rice yield data across 247 of India's rice-producing districts. In addition to model-building, a dynamic dashboard was built understand how the reliability of rice yield predictions varies across district. RESULTS AND CONCLUSIONSThe results of the proof-of-concept machine learning pipeline demonstrated that rice yields can be predicted with a reasonable degree of accuracy, with out-of-sample R2, MAE, and MAPE performance of up to 0.82, 0.29, and 0.16 respectively. This performance outperformed test set performance reported in related literature on rice yield modelling in other contexts and countries. In addition, SHAP value analysis was conducted to infer both the importance and directional impact of the climate and remote sensing variables included in the model. Important features driving rice yields included temperature, soil water volume, and leaf area index. In particular, higher temperatures in August correlate with increased rice yields, particularly when the leaf area index in August is also high. Building on the results, a proof-of-concept dashboard was developed to allow users to easily explore which districts may experience a rise or fall in yield relative to the previous year. The dashboard show that the model may perform better in some regions than in others. For instance, the absolute percentage error for predicted versus actual yields ranged from an average of 7.1 % in districts in Uttarakhand to an average of 14.7 % in Uttar Pradesh. SIGNIFICANCEThis study underscores the potential for policymakers to consider scaling and operationalizing machine learning approaches to rice yield prediction in the context of agricultural early warning systems to deliver timely crop yield forecasts on a rolling basis throughout the season, thereby equipping agricultural decision-makers with the ability to make informed choices on irrigation scheduling, fertilizer application, and harvest planning to optimize crop output and resource use.

Towards improved description of plastic anisotropy in sheet metals under biaxial loading: A novel generalization of Hill48 yield criterion

Lightweight design and development of automobiles are strongly dependent on lightweight high-strength sheet metals. An accurate prediction of anisotropy and yield surface evolution of these materials under different sampling directions of biaxial tension (BT) is a necessary condition for the numerical analyses of stamping forming. In this work, the Hill48 model considering the principal stress direction is generalized to describe the anisotropic evolution and the yield loci on the normal and diagonal planes. The material parameters of the generalized M-Hill48 model (GM-Hill48) have analytical form and can be calibrated by the yield stresses and r-values expressed by the continuous equivalent plastic strain (EPS). A quadratic function interpolation method based on the near-plane strain (NPS) stress is established to control the yield locus under different plane stress states. In addition, a method for predicting the yield loci in arbitrary sampling directions under BT is proposed by tensor coordinate transformation. Different from the traditional model which can only determine material parameters by using equi-biaxial tension (EBT) or NPS stress along the rolling direction (RD) and transverse direction (TD), the yield stresses of BT in different sampling directions can be incorporated into the generalized model. The predictive capability of the GM-Hill48 model is verified by capturing the continuous evolution of yield stresses and strain for DP590 and AA6016-T4 materials. The convexity condition of the established modified model is given and discussed. Finally, the advantages of new model are further expounded in terms of the prediction accuracy and identified parameters compared with the advanced yield models.

Energy Yield Modeling of Perovskite–Silicon Tandem Photovoltaics: Degradation and Total Lifetime Energy Yield

This study investigates the impact of degradation in perovskite‐silicon tandem solar cells by means of energy yield (EY) modelling over the entire lifetime. First, we assess the impact on EY of degradation in the individual solar cell parameters of the perovskite top cell. Our analysis reveals that degradation in fill factor, due to a decline in perovskite top cell shunt resistance (RSh), has the most severe impact on the EY, emphasizing the imperative to rectify perovskite imperfections in thin film processing causing RSh decline. Second, we investigate implications of degradation in the perovskite top cell on the EY of current mismatched tandem solar cells. Third, we examine critical thresholds for the “acceptable degradation levels” in the perovskite top cell with regard to degradation in each solar cell parameter, assuming that the total loss in EY must be comparable to the degradation in state‐of‐the‐art silicon. Overall, our study highlights that degradation of the perovskite top cell needs to be assessed with care when extrapolating the impact on the lifetime EY of perovskite‐silicon tandem solar cells. The severity of degradation for different degradation mechanisms in a single junction perovskite solar cell cannot be translated one‐to‐one to tandem devices.

Assessments of 12 Commercial Species Stocks in a Subtropical Upwelling Ecosystem Using the CMSY and BSM Methods

Twelve commercial species exploited in the eastern Guangdong and southern Fujian waters were assessed using the Catch-Maximum Sustainable Yield (CMSY) and Bayesian Schaefer Model (BSM) methods. The carrying capacity (k), intrinsic rate of population growth (r), maximum sustainable yield (MSY), and relative biomass (Bend/k and B/BMSY) were estimated. The current stock status was defined by B/BMSY and fishing mortality (F/FMSY). The results indicate that seven stocks were overfished or below safe biological limits (B/BMSY < 0.5 or F/FMSY > 1), two stocks were in a recovery phase (0.5 < B/BMSY < 1, F/FMSY < 1), and three stocks were under sustainable fishing pressure with healthy biomass, capable of producing yields close to the MSY (B/BMSY > 1, F/FMSY < 1). The stock statuses are consistent with previous studies on the utilization of pelagic fisheries in the eastern Guangdong and southern Fujian waters and with those assessments in other waters. The results of the assessments suggest that these stocks could be expected to produce higher sustainable catches if permitted to rebuild; thus, more effective and proactive management is needed in this upwelling fishing ground.

Evaluation of CMIP6 models for sea surface temperature and sea surface salinity variability over the Arabian Sea

The capability of Coupled Model Inter-comparison Project Phase-6 (CMIP6) is analysed in simulating the seasonal variability of Sea Surface Temperature (SST) and Sea Surface Salinity (SSS) over the Arabian Sea in this study. The historical simulations of 11 CMIP6 models, which come from various source are validated with the observational data (HadISST for SST and SODA for SSS). This study covers the Arabian Sea, spanning 5°S to 32°N and 33°E to 80°E from 1985 to 2014 (30 years). Seasonal mean climatology and annual cycle of SST and SSS are estimated over the Arabian Sea. The deviation of coupled model-simulated SST and SSS from the observation is quantified by determining the biases during all the seasons. Four distinct model evaluation measures are used to access the performance of CMIP6 models: root mean square error (RMSE), standard deviation (STD), correlation coefficient (CC) and interannual variability skill score (IVS). CMCC-CM2-SR5, NESM3 and IITM-ESM models for SST yield CC/RMSE values of 0.988/0.174, 0.983/0.204 and 0.963/0.299, respectively. CMCC-CM2-SR5, IPSL-CM6A-LR and CanESM5 models for SSS yield CC/RMSE values of 0.956/0.035, 0.937/0.042 and 0.820/0.069, respectively. Thus, these models having higher CC and lower RMSE among all models, give better performance. The present study is important as the future scenarios can be predicted using the best models.

Correlation and regression analysis of the influence of economic factors on wheat yield

The article presents the results of econometric modeling of wheat yield based on microdata from reports on the financial and economic status of agricultural producers. The relationship between wheat yield and economic factors is studied: indicators of the level of intensity (costs of mineral fertilizers, cost of fixed assets, direct labor costs for grain production, number of workers employed in agriculture per unit area) and specialization of agricultural production (proportion of revenue from sales of crop products in revenue from sales of agricultural products). Using the stepwise variable selection approach, statistically significant linear regression models were constructed that can be used for analysis and forecasting. The paired model included the cost of mineral fertilizers per 1 ha of sown area as a factor in wheat yield (coefficient of determination R2 = 0.503), and multiple regression was obtained by adding a second factor — the cost of fixed assets per 100 ha of agricultural land (multiple coefficient of determination R2 = 0.562). Decomposition of the multiple coefficient of determination into separate coefficients of determination allowed us to estimate the impact of each factor included in the model on yield: thus, the impact of mineral fertilizer costs was 42 %, and the impact of fixed assets was 15 %. According to the approach to decomposition of R2 based on β-coefficients, the net impact of mineral fertilizer costs on wheat yield was 35 %, the net impact of fixed assets was 5 %, while the impact of the systemic effect from the interaction of the two factors, which reflects the level of production intensification, was 17 %. Based on a multiple regression model, forecast values were constructed and estimated, and reserves for yield growth were determined.