Search

Accurate estimation of mineral composition is essential for refined reservoir characterization, thermal conductivity and mechanical determinations of sedimentary rocks, but is extremely challenging in shale units due to the mineralogical complexity, low porosity and ultra‐low permeability. Direct mineral measurements derived from laboratory X‐ray diffraction (XRD) analysis on core samples and borehole geochemical logging tool (GLT), and conventional geophysical logs from vertical wells penetrating sediments are widely available in some basins, which enables detailed mineralogical characterization of a well. A hybrid machine learning (ML) architecture that improves model training and validation by combining convolutional neural network (CNN) with XGBoost allows accurate description of the mineralogical compositions across a basin. We applied this ML approach to predict the mineral compositions using conventional well logs from the Horn River Basin, northeast British Columbia, Canada, where extensive drilling for shale‐gas and conventional hydrocarbon resources, complemented by high temperature geothermal energy potential is ideal for case testing. The predicted mineral compositions from the ML approach are consistent with the mineralogical readings from the GLT and are confirmed by the XRD mineral measurements. This allows basin‐wide mineral compositions mapping that reveals spatial trends of major mineral compositions and their relationship with the previously recognized geomechanical and geological features, which have important implications for thermal conductivity modeling, reservoir evaluation and extensive geological studies. Key Points: A hybrid machine learning approach is applied to develop log‐mineral models for mineralogy characterization using geophysical well logsThe model‐predicted major mineral compositions are consistent with the mineral data from geochemical logging tool and XRD analysisThe Horn River Group shale is dominated by quartz, clay, and carbonate and shows great variability among the three formations in the basin [ABSTRACT FROM AUTHOR]

Remote sensing has been used for assisting the precision nitrogen (N) management in wheat (Triticum aestivum) production. This study aimed to develop an efficient and energy saving N management strategy based on multi-source data for winter wheat at the farm scale. Five field experiments involving different cultivars and N treatments were conducted to establish and validate the N management strategy in 2017–2021. UAV multi-spectral images, plant sampling, weather and field management data collection were carried out synchronously at Feekes 6.0. Four machine learning methods were used to integrate multi-variate information to determine the optimal parameters in N regulation algorithm. The results showed random forest (RF) algorithm performed best for plant dry matter (R2 = 0.78) and plant N accumulation (R2 = 0.83) estimation, a N nutrition index optimized algorithm (NNIOA), driven by multi-source data, was developed and used for guiding in-season N application. The NNIOA efficiently regulated the deficient, optimal and excessive N status through up- (54.17%), fine- (0.67%) and downward- (18.18%) adjustment of N fertilizers, respectively, while the optimal N treatment achieved highest net profit, energy use efficiency (EUE) and energy productivity (EP). Compared with farmer's practices, the NNIOA increased partial factor productivity (PFP), net profit, EUE and EP by 19.60–27.94%, 22.47–45.13 $ ha−1, 6.94–13.07% and 8.36–12.29%, respectively, while reduced N input (16.77–21.67%), energy input (8.13–10.74%) and CO2 emission (7.60–10.11%) without any yield reduction at study farms. In conclusion, this study supplied a precision N management strategy to implement variable N application for sustainable wheat production at farm scale. [ABSTRACT FROM AUTHOR]

Wheat is an important food crop in China. The quality of wheat affects the development of the agricultural economy. However, the high-quality wheat produced in China cannot meet the demand, so it would be an important direction for research to develop high-quality wheat. Grain protein content (GPC) is an important criterion for the quality of winter wheat and its content directly affects the quality of wheat. Studying the spatial heterogeneity of wheat grain proteins is beneficial to the prediction of wheat quality, and it plays a guiding role in the identification, grading, and processing of wheat quality. Due to the complexity and variability of wheat quality, conventional evaluation methods have shortcomings such as low accuracy and poor applicability. To better predict the GPC, geographically weighted regression (GWR) models, multiple linear regression, random forest (RF), BP neural networks, support vector machine, and long-and-short-term memory algorithms were used to analyze the meteorological data and soil data of Jiangsu Province from March to May in 2019–2022. It was found that the winter wheat GPC rises by 0.17% with every 0.1° increase in north latitude at the county level in Jiangsu. Comparison of the prediction accuracy of the coefficient of determination, mean deviation error, root mean square error, and mean absolute error by analyzing multiple algorithms showed that the GWR model was the most accurate, followed by the RF model. The regression coefficient of precipitation in April showed the smallest range of variation among all factors, indicating that precipitation in April had a more stable effect on GPC in the study area than the other meteorological factors. Therefore, consideration of spatial information might be beneficial in predicting county-level winter wheat GPC. GWR models based on meteorological and soil factors enrich the studies regarding the prediction of wheat GPC based on environmental data. It might be applied to predict winter wheat GPC and improve wheat quality to better guide large-scale production and processing. [ABSTRACT FROM AUTHOR]

With the rapid popularization of the network, under the increasingly complex network security situation and the increasingly prominent network security problems, network security occupies an important field in the wireless communication base station monitoring system, and has become a hot research direction. It is to design a wireless communication base station monitoring system based on artificial intelligence and network security. In the experiment, using the supervised machine learning algorithm, the program of the wireless communication base station monitoring system is designed by setting the working frequency of the GSM-based wireless communication system to the wireless communication base station monitoring system. [ABSTRACT FROM AUTHOR]

Anaerobic digestion (AD) is a promising way to produce renewable energy. The solid-state anaerobic digestion (SSAD) with a dry matter content more than 15% in the reactors is seeing its increasing potential in biogas plant deployment. The relevant processes involve multiple of evolving chemical and physical phenomena that are not crucial to conventional liquid-state anaerobic digestion processes (LSAD). A good simulation of SSAD is of great importance to better control and operate the reactors. The modeling of SSAD reactors could be realized either by theoretical or statistical approaches. Both have been studied to a certain extent but are still not sound. This paper introduces the existing mathematical tools for SSAD simulation using theoretical, empirical and advanced statistical approaches and gives a critical review on each type of model. The issues of parameter identifiability, preference of modeling approaches, multiscale simulations, sensibility analysis, particularity of SSAD operations and global lack of knowledge in SSAD media evolution were discussed. The authors call for a stronger collaboration of multidisciplinary research in order to further developing the numeric simulation tools for SSAD. [ABSTRACT FROM AUTHOR]

Timely and accurate estimation of plant nitrogen (N) status is crucial to the successful implementation of precision N management. It has been a great challenge to non-destructively estimate plant N status across different agro-ecological zones (AZs). The objective of this study was to use random forest regression (RFR) models together with multi-source data to improve the estimation of winter wheat (Triticum aestivum L.) N status across two AZs. Fifteen site-year plot and farmers' field experiments involving different N rates and 19 cultivars were conducted in two AZs from 2015 to 2020. The results indicated that RFR models integrating climatic and management factors with vegetation index (R2 = 0.72–0.86) outperformed the models by only using the vegetation index (R2 = 0.36–0.68) and performed well across AZs. The Pearson correlation coefficient-based variables selection strategy worked well to select 6–7 key variables for developing RFR models that could achieve similar performance as models using full variables. The contributions of climatic and management factors to N status estimation varied with AZs and N status indicators. In higher-latitude areas, climatic factors were more important to N status estimation, especially water-related factors. The addition of climatic factors significantly improved the performance of the RFR models for N nutrition index estimation. Climatic factors were important for the estimation of the aboveground biomass, while management variables were more important to N status estimation in lower-latitude areas. It is concluded that integrating multi-source data using RFR models can significantly improve the estimation of winter wheat N status indicators across AZs compared to models only using one vegetation index. However, more studies are needed to develop unmanned aerial vehicles and satellite remote sensing-based machine learning models incorporating multi-source data for more efficient monitoring of crop N status under more diverse soil, climatic, and management conditions across large regions. [ABSTRACT FROM AUTHOR]

Patients with carcinoma of unknown primary (CUP) account for 3–5% of all cancer cases. A large number of metastatic cancers require further diagnosis to determine their tissue of origin. However, diagnosis of CUP and identification of its primary site are challenging. Previous studies have suggested that molecular profiling of tissue-specific genes could be useful in inferring the primary tissue of a tumor. The purpose of this study was to evaluate the performance somatic mutations detected in a tumor to identify the cancer tissue of origin. We downloaded the somatic mutation datasets from the International Cancer Genome Consortium project. The random forest algorithm was used to extract features, and a classifier was established based on the logistic regression. Specifically, the somatic mutations of 300 genes were extracted, which are significantly enriched in functions, such as cell-to-cell adhesion. In addition, the prediction accuracy on tissue-of-origin inference for 3,374 cancer samples across 13 cancer types reached 81% in a 10-fold cross-validation. Our method could be useful in the identification of cancer tissue of origin, as well as the diagnosis and treatment of cancers. [ABSTRACT FROM AUTHOR]

Leaf area index (LAI) and leaf dry matter (LDM) are important indices of crop growth. Real-time, nondestructive monitoring of crop growth is instructive for the diagnosis of crop growth and prediction of grain yield. Unmanned aerial vehicle (UAV)-based remote sensing is widely used in precision agriculture due to its unique advantages in flexibility and resolution. This study was carried out on wheat trials treated with different nitrogen levels and seeding densities in three regions of Jiangsu Province in 2018–2019. Canopy spectral images were collected by the UAV equipped with a multi-spectral camera during key wheat growth stages. To verify the results of the UAV images, the LAI, LDM, and yield data were obtained by destructive sampling. We extracted the wheat canopy reflectance and selected the best vegetation index for monitoring growth and predicting yield. Simple linear regression (LR), multiple linear regression (MLR), stepwise multiple linear regression (SMLR), partial least squares regression (PLSR), artificial neural network (ANN), and random forest (RF) modeling methods were used to construct a model for wheat yield estimation. The results show that the multi-spectral camera mounted on the multi-rotor UAV has a broad application prospect in crop growth index monitoring and yield estimation. The vegetation index combined with the red edge band and the near-infrared band was significantly correlated with LAI and LDM. Machine learning methods (i.e., PLSR, ANN, and RF) performed better for predicting wheat yield. The RF model constructed by normalized difference vegetation index (NDVI) at the jointing stage, heading stage, flowering stage, and filling stage was the optimal wheat yield estimation model in this study, with an R2 of 0.78 and relative root mean square error (RRMSE) of 0.1030. The results provide a theoretical basis for monitoring crop growth with a multi-rotor UAV platform and explore a technical method for improving the precision of yield estimation. [ABSTRACT FROM AUTHOR]

• LSTM model integrating multi-source data performed the best for wheat yield prediction. • A spatio-temporal variation-based LSTM can capture the cumulative effect of wheat. • LSTM model presented a robust yield prediction under normal years and extreme years. • Meteorological factors, especially precipitation, were the primary drivers for yield. Accurately and timely predicting yield and identifying its drivers are crucial for adjusting agricultural interventions, thereby responding to climate change and ensuring food security. Winter wheat phenology exhibited pronounced temporal and spatial variations at the county level in Jiangsu province. This study constructed a robust long short-term memory (LSTM) model that incorporates multiple data sources, including meteorological parameters, soil attributes, terrain characteristics, remote sensing data, and heterogeneous wheat phenology, aiming at forecasting the county-level wheat yield from year 2005 to 2020, with an emphasis on different precipitation year-types. The results demonstrated that the LSTM model could explain 71% of yield variations because of its ability to deal with time cumulative effects and high-dimensional data. Its performance outperformed that of machine learning algorithms, even in a normal year (RMSE = 0.26 t/ha) and an extreme year (RMSE = 0.29–0.33 t/ha). Time-series data (meteorological factors and vegetation indices), soil attributes, as well as topographic data that capture spatial and temporal heterogeneity, can achieve accurate wheat prediction approximately three months before harvest. Feature importance analysis showed meteorological factors, especially precipitation, are the most important drivers for yield prediction. This study developed a promising and scalable framework for spatio-temporal yield prediction by integrating available multi-source data to respond to climatic variations and provided insights for policymakers and farmers to adjust agricultural interventions. [ABSTRACT FROM AUTHOR]

• Extending dualex indicators to canopy level by LAI raised N diagnosis accuracy. • Fusing multi-source data helps build dynamic and reliable N diagnosis models. • Independent data and integrated assessment ensured the optimal N diagnosis model. • Advising on the priority of obtaining external data under different targets. Rapid and accurate nitrogen (N) diagnosis plays a crucial role in precise N fertilizer management of wheat. However, most existing N diagnosis models based on proximal fluorescence sensors' indicators are confined to a single growth stage, thereby limiting the accuracy and universality of models. Therefore, this study aimed to construct and evaluate wheat N diagnosis models based on proximal fluorescence sensors across growth stages by fusing multi-source data. Six field experiments conducted over four years, involved diverse planting densities, wheat varieties, and N rates. These experiments were performed to investigate the relationship between optical indicators obtained from Dualex 4 and Multiplex 3, and both the plant N accumulation (PNA) and the N nutrition index (NNI). The dualex indicators were extended to the canopy level (canopy_dualex indicators) by the leaf area index (LAI). In addition to formulating N nutrition diagnostic models based on the three optical indicators solely, we further integrated meteorological factors, soil basic fertility, and cultivation practices to construct dynamic N diagnosis models coupling with machine learning algorithms. Fusing multi-source data significantly improved the R2, resulting in an increase of 0.20 and 0.29 when predicting PNA for dualex and multiplex, respectively. The canopy_dualex indicator consistently performed best in predicting both PNA (R2 = 0.75, RRMSE = 28.71 %) and NNI (R2 = 0.60, RRMSE = 24.62 %) among the three optical indicators. Moreover, the inclusion of LAI effectively addressed the overfitting issue observed in dualex when fusing multi-source data to construct the models. Consistency tests and ROC curve analyses provided robust evidence that canopy_dualex exhibited the highest consistency and the most powerful diagnostic ability. Additionally, multiplex demonstrated superior performance compared to dualex in predicting PNA and NNI, with higher R2 (0.46–0.50) and lower RRMSE (28.57 %-40.84 %). The results underscored that multi-source data fusion significantly improved the accuracy of universal N nutrition models for wheat across growth stages, leveraging proximal fluorescence sensors to cover the entire wheat growth process. This approach allows for the identification of N nutrition status at any given time, facilitating timely adjustments in N fertilizer management. From the two aspects of feature selection results of multi-source data fusion and the difficulty of obtaining data in the application, it is recommended that the first variable to be added in N nutrition diagnosis of is N application amount. If the target is PNA, the accumulated precipitation (APP) data is collected first. In order to obtain NNI, the soil total N content is considered first. Furthermore, this flexibility offers a convenient and promising option for practical agricultural production. [ABSTRACT FROM AUTHOR]

Real-timely monitoring of the crop water status can improve irrigation scheduling to increase water saving and enhance agricultural sustainability, whereas the canopy temperature measured by thermal imaging is an essential indicator for determining the rice water status. The primary goal of this study was to propose a new temperature index based on the temporal variance of the daily temperature and to develop the rice water status estimation model. Two field experiments involving two rice varieties and multi-irrigation regimes were conducted from 2019 to 2020. A thermal imaging camera was used to measure the canopy temperature from 8:00–16:00 at 2-hour intervals across all growth stages. Three plant water parameters, namely plant water content (PWC), canopy water content (CWC), and canopy equivalent water thickness (CEWT), were collected simultaneously. The results showed that canopy temperature and plant water parameters differed obviously among different irrigation treatments. The relative canopy temperature velocity (RCTV) was developed based on the temporal variance of the daily temperature, and the RCTV 8–12 performed well in distinguishing different irrigation treatments and quantifying the rice water status. The coefficient of determination (R2) values of the exponential relationships between the optimal RCTV and plant water parameters reached 0.47 (PWC), 0.39 (CWC) and 0.18 (CEWT). The random forest model, which integrates the multi-temperature indices, achieved a good estimation for PWC (R2 = 0.78), CWC (R2 = 0.77), and CEWT (R2 = 0.64) across all growth stages. In summary, combining the multi-temperature indices derived from the thermal infrared imagery and machine learning algorithm can facilitate the non-destructive estimation of the rice water status and improve the precision irrigation schedule. • Variations of rice water status and canopy temperature with time changes were clarified. • RCTV index successfully assessed the temporal variation of rice canopy temperature and different water treatments. • RF model based on the multi-temperature indices can better estimate the rice water status. [ABSTRACT FROM AUTHOR]

• Soil, meteorological, management, and RS data were combined using ML algorithms. • ML models for in-season prediction of yield and reactive N losses were developed. • N damage costs of reactive N losses were considered in calculating EONR. • In-season AONR and EONR predicted by RFR and SVR well accorded with observed values. • Historical meteorological data improved accuracy for predicting AONR and EONR. The modern in-season crop N recommendation approaches should have high reliability in promoting agricultural sustainability. These approaches are relevant to soil properties, meteorological conditions, management practices, and crop in-season growing status. This study aims to use machine learning (ML) algorithms to incorporate the above variables as well as the field reactive nitrogen (N) losses (i.e., N damage cost) simulated by a DeNitrification–DeComposition (DNDC) model to develop a new strategy for optimizing rice in-season topdressing N (TN) usage. Rice field experiments with multiple N treatments and rice varieties were carried out during 2015–2021 at four study sites in eastern China. Four ML algorithms, namely random forest regression (RFR), support vector regression (SVR), lasso regression (LSR), and partial least square regression (PLSR) were used to develop in-season prediction models of yield and reactive N losses by combining soil, meteorological, and management data with crop remote sensing data. The observed in-season agronomic optimum N rates (AONR) that can maximize rice yield at different sites were in the range of 116.5 to 177.4 kg N ha−1, while the in-season economic optimum N rates (EONR) that can maximize marginal revenue (i.e., yield income minus N fertilizer costs and N damage costs) were in the range of 97.4 to 163.6 kg N ha−1. The developed ML models were further used to simulate yield and marginal revenue responses to a series of assumed TN rates (0–300 kg N ha−1, gradient = 20 kg N ha−1). Comparably, RFR model and SVR model were more suitable for determining optimum TN rates, because their simulated response curves of yield and marginal revenue fit the normal regulation (linear plus plateau or single-peak shapes). Independent validation results showed that the in-season AONR and EONR predicted by RFR and SVR well accorded with the observed values (R2 ≥ 0.64, RRMSE ≤ 18.3 %), and the accuracy of ML models containing both historical and in-season meteorological information is superior to ML models that contain in-season meteorological information only. The proposed ML-based strategy is expected to help the regional rice production systems precisely manage N use, improve net profits, and reduce environmental footprints. [ABSTRACT FROM AUTHOR]

Real-time and non-destructive nitrogen (N) status diagnosis is needed to support in-season N management decision-making for modern wheat production. For this purpose, satellite sensor imaging can act as an effective tool for collecting crop growth information across large areas, but they can be challenging to calibrate with ground reference data. This research aimed to calibrate satellite remote sensing-derived models for crop growth estimation and N status diagnosis based on fine-resolution unmanned aerial vehicle (UAV) images, thus, map wheat growth and N status at the county scale. Seven wheat field experiments involving multi cultivars and different N applications were conducted at four farms of Xinghua county from 2017 to 2021. A fixed-wing UAV sensing system and the Sentinel 2 (S2) satellite were used to collect wheat canopy multispectral images; three growth variables (plant dry matter (PDM), plant N accumulation (PNA) and N nutrition index (NNI)) and weather data, synchronized with spectral imagery, were obtained at the jointing and booting stages. The farm-scale PDM (UAV-PDM) and PNA (UAV-PNA) maps can be derived from the UAV images at the four farms, which were further upscaled to grids to match the S2 image resolution using pixel aggregation method. Then, satellite-based prediction models were constructed by fitting four machine learning algorithms to the relationships between satellite spectral indices, upscaled PDM (PNA) and weather data. Amongst the four methods tested, the random forest (RF) achieved the greatest prediction accuracy for PDM (R 2 = 0.69–0.93) and PNA (R 2 = 0.60–0.77). Meanwhile, an indirect diagnosis method was used to calculate the NNI. The results indicated that the model derived from the S2 imagery performed well for predicting NNI (R 2 = 0.46–0.54) at the jointing and booting stages. Thereby, the NNI was used to map winter wheat N nutrition status at the county scale. In summary, this research demonstrated and evaluated an approach to combine UAV and satellite sensor images to diagnose wheat growth and N status across large areas. • The farm-scale PDM and PNA maps were upscaled to grids to match S2 image resolution. • The satellite-based prediction models were constructed by fitting four ML algorithms. • The NNI diagnosis model and satellite images were used to map winter wheat N nutrition status at the county scale. [ABSTRACT FROM AUTHOR]

Accurate and transferable wheat nitrogen status estimation is very important to plant phenotyping and smart agricultural management. The goal of this study was to establish a wheat nitrogen status estimation model across all growth stages by combining proximal sensing and meteorological data. From 2010–2020, nine multi-nitrogen rates field trials were conducted at five sites involving different wheat varieties. Proximal sensing data were acquired from a Crop Circle sensor at key growth stages and meteorological data were aggregated from planting to the corresponding sensing date. Deep neural network (DNN) and long short-term memory (LSTM) were adopted to estimate above-ground biomass, plant nitrogen uptake, plant nitrogen concentration, and the nitrogen nutrition index. Random forest (RF) was used as a benchmark regression model. Multi-task learning (MTL) based on DNN was conducted to estimate the four nitrogen indicators simultaneously. A genetic algorithm (GA) was tested to optimize the hyperparameters, connection weights, and loss function weights (for MTL) of neural networks separately. The results revealed that DNN (R2 =0.83–0.96) and MTL (R2 =0.81–0.96) achieved an overall comparable high accuracy with RF (R2 =0.83–0.97), whereas LSTM (R2 =0.76–0.93) did not improve the nitrogen status estimation in our dataset. This study presented a concise and efficient framework dedicated to exploring the transferability of phenotypic predictions and provided insights into understanding crop growth and nitrogen dynamics in response to environmental conditions. • A framework combined deep learning and evolutionary algorithm was proposed. • The proposed model performed high transferability in different space and time. • Features with cumulative effect greatly impacted the wheat N status estimation. • Water-related features were essential for wheat nitrogen content estimation. [ABSTRACT FROM AUTHOR]

Water and nitrogen (N) are the most important factors limiting crop productivity. Effective monitoring of the water status of winter wheat under different N treatments is imperative for precision irrigation in improved crop management. Hyperspectral remote sensing is widely used for monitoring the crop water status. However, changes in canopy architecture during ontogeny lead to poor inversion of crop properties and limit the use of spectral indices. This study aimed to improve the water status prediction of winter wheat using multi-source data with machine learning (ML). Two multi-irrigation levels (0, 120, 240, 360 mm) and N rates (75, 225 kg N ha-1) experiments were conducted during the 2019–2021 wheat growing seasons under field conditions using a rainout shelter. Hyperspectral, soil, plant, and climate data were evaluated with two feature selection methods. Selected results were chosen as input variables for prediction models by using three ML algorithms. By constructing the normalized difference spectral index (NDSI), ratio spectral index (RSI), and difference spectral index (DSI), the DSI (2015, 2375) , NDSI (2175, 2245) , and RSI (720, 1200) showed the strongest correlation with canopy water content (CWC), plant water content (PWC), and canopy equivalent water thickness (CEWT), respectively. The best feature selection method and data were delivered by the Pearson correlation coefficient together with the soil, plant, and climate data. The best performing ML algorithm for CWC and PWC prediction was RF, while SVM was the best ML algorithm for CEWT prediction. The R2 of the optimal models ranged from 0.86 to 0.97. These models with multi-source data significantly improved the prediction accuracy of the water status and can thus assist in precision irrigation management of winter wheat. • The best prediction based on spectral indices for CWC, PWC, and CEWT were different. • The optimal ML algorithms were not quite consistent for different water indicators prediction. • The model with multi-source data significantly improved the prediction accuracy of the water status. [ABSTRACT FROM AUTHOR]

Increasing nitrogen (N) diagnosis efficiency and accuracy is crucial for optimizing wheat N management. We aimed to establish a spatially and temporally explicit model for the diagnosis of winter wheat N status on small scale farms using multivariate information. To determine the most accurate approach, seven field experiments involving different cultivars and N treatments were conducted in east China over five years. A fixed-wing unmanned aerial vehicle (UAV) mounted multispectral camera was used to acquire canopy spectral data of winter wheat at the jointing and booting stages, while agronomic indicators of plant dry matter (PDM), plant N accumulation (PNA) and N nutrition index (NNI), as well as agrometeorological (AM) and field management (FM) data, were measured synchronously. Direct and indirect strategies of NNI estimation were applied for N diagnosis at the jointing and booting stages. Four machine learning (ML) algorithms were used to characterize the relationships between agronomic variables and UAV remote sensing, AM and FM data. The results demonstrated the random forest (RF) model that integrated UAV remote sensing, AM and FM data achieved the higher accuracy for predicting NNI (R 2 = 0.82–0.87, RMSE = 0.11–0.12 and RE = 12.94%−15.57%) amongst the four ML models based on the direct strategy at the jointing and booting stages. Similarly, the RF model performed most accurate estimation for PDM (R 2 = 0.69–0.78, RMSE = 0.43–0.61 t ha−1 and RE = 12.74%−24.49%) and PNA (R 2 = 0.83–0.84, RMSE = 13.00–17.53 kg ha−1 and RE = 17.03%−25.44%), then NNI (R 2 = 0.54–0.55, RMSE = 0.09–0.13 and RE = 8.34–12.65%) was further calculated using the indirect diagnosis strategy. Based on the optimal NNI diagnosis interval derived from the relationships between relative yield (RY) and plant NNI at the jointing (0.92–1.04) and booting (0.97–1.15) stages, the two diagnosis strategies obtained similar diagnostic accuracies in the three study farms and performed more accurately at the booting (areal agreement = 0.70–0.90, Kappa = 0.49–0.82) than jointing (areal agreement = 0.54–0.71, Kappa = 0.36–0.53) stages. The combination of fixed-wing UAV remote sensing with AM and FM information using the RF algorithm can significantly increase the accuracy and efficiency of in-season wheat N diagnosis at the farm scale. • Four machine learning algorithms were used to characterize the relationships between agronomic variables and multivariate data. • Direct and indirect strategies of NNI estimation were evaluated based on the optimal NNI diagnosis interval. • Mapping the wheat N status based on fixed-wing UAV imagery at the farm scale. [ABSTRACT FROM AUTHOR]

• Ensemble learning achieved the best performance for field-scale wheat yield prediction. • Normalized Difference Red Edge Index was the most contributory features. • Average temperature, minimum temperature, and relative humidity were key weather data. • The low temperature and dry air in winter and spring are limiting factors in wheat yield. Accurate and timely wheat yield prediction is of great importance to global food security. Early prediction of wheat yield at a field scale is essential for site-specific precision management. This study aimed to develop an in-season wheat yield prediction model at field-scale by integrating proximal sensing and weather data. Nine multi-N rates field experiments were conducted at five sites involving different wheat cultivars from 2010 to 2020. Proximal sensing data were collected from a Crop Circle sensor at the stem elongation stage and weather data were collected from 30 days before planting to the flowering date. Eleven statistical and machine learning (ML) regression algorithms were adopted, along with two aggregation intervals (disaggregated or aggregated data) and two feature selection methods (based on Pearson Correlation Coefficient or Recursive Feature Elimination). The results revealed that the ensemble learning models (Random Forest, eXtreme Gradient Boosting) achieved the best overall performance (R2 = 0.74 ∼ 0.78, RMSE = 0.78 ∼ 0.85 t ha−1). Feature importance analysis showed that Normalized Difference Red Edge Index (NDRE), average temperature, minimum temperature, and relative humidity were the most contributory features, especially from the planting date to the stem elongation date (for weather features). The aggregation approach and feature selection method did not significantly affect the yield prediction performance for the seven ML methods. This study introduced a promising framework that complements county-scale models and provided insights into understanding yield responses to environmental conditions. The best prediction model can be applied for guiding real-time sensor-based precision fertilization. [ABSTRACT FROM AUTHOR]

[Display omitted] • 232 articles from 2001 to 2020 were selected and reviewed. • The important process of machine-learning model development is reviewed. • Three input resources and three feature selection methods are compared. • The machine-learning models are categorized into seven classifications. Global solar radiation information is the basis for many solar energy utilizations as well as for economic and environmental considerations. However, because solar-radiation changes, and measurements are sometimes not available, accurate global solar-radiation data are often difficult or impossible to obtain. Machine-learning models, on the other hand, are capable of conducting highly nonlinear problems. They have many potential applications and are of high interest to researchers worldwide. Based on 232 paper regarding to the machine-learning models for global solar radiation prediction, this paper provides a comprehensive and systematic review of all important aspects surrounding machine-learning models, including input parameters, feature selection and model development. The pros and cons of three input-parameter sources (observation data from a surface meteorological observation station, satellite-based data, numerical weather-predicting re-analyzed data) and three feature selection methods (filter, wrapped, embedded) are reviewed and analyzed in this paper. Using data pre-processing algorithms, output ensemble methods, and model purposes, seven classes of machine-learning models are identified and reviewed. Finally, the state of current and future research on machine-learning models to forecast the global solar radiation are discussed. This paper provides a compact guide of existing model modification and novel model development regarding predicting global solar radiation. [ABSTRACT FROM AUTHOR]