Search

Estimating the spatiotemporal variations in natural grassland carrying capacity is crucial for maintaining the balance between grasslands and livestock. However, accurately assessing this capacity presents significant challenges due to the high costs of biomass measurement and the impact of human activities. In this study, we propose a novel method to estimate grassland carrying capacity based on potential net primary productivity (NPP), applied to the source area of the Nujiang River and Selinco Lake on the Tibetan Plateau. Initially, we utilize multisource remote sensing data—including soil, topography, and climate information—and employ the random forest regression algorithm to model potential NPP in areas where grazing is banned. The construction of the random forest model involves rigorous feature selection and hyperparameter optimization, enhancing the model's accuracy. Next, we apply this trained model to areas with grazing, ensuring a more accurate estimation of grassland carrying capacity. Finally, we analyze the spatiotemporal variations in grassland carrying capacity. The main results showed that the model achieved a high level of precision, with a root mean square error (RMSE) of 4.89, indicating reliable predictions of grassland carrying capacity. From 2001 to 2020, the average carrying capacity was estimated at 9.44 SU/km2, demonstrating a spatial distribution that decreases from southeast to northwest. A slight overall increase in carrying capacity was observed, with 65.7% of the area exhibiting an increasing trend, suggesting that climate change has a modest positive effect on the recovery of grassland carrying capacity. Most of the grassland carrying capacity is found in areas below 5000 m in altitude, with alpine meadows and alpine meadow steppes below 4750 m being particularly suitable for grazing. Given that the overall grassland carrying capacity remains low, it is crucial to strictly control local grazing intensity to mitigate the adverse impacts of human activities. This study provides a solid scientific foundation for developing targeted grassland management and protection policies. [ABSTRACT FROM AUTHOR]

Precision glass molding (PGM) is an effective approach to manufacturing infrared chalcogenide glass (ChG) aspherical lens with complex shapes. However, infrared ChG aspherical lens often experiences form error in the designed profile and the final profile obtained by PGM. To reduce the form error of infrared ChG aspherical lens in the PGM process, a form error compensation model based on the random forest (RF) algorithm is proposed. The infrared ChG aspherical lens profile was first machined on an electroless nickel-phosphorus (Ni–P) plating to serve as the mold for PGM. After molding, the profile data of the lens was extracted, and a compensation model based on RF was constructed to optimize the model parameters using the evaluation parameters such as root mean square error (RMSE), coefficient of determination (R2), and out-of-bag (OOB). Finally, the generated compensated profile based on the compensation model was used for the compensation machining of the mold. Through this compensation approach, we have demonstrated a substantial 63.5 % reduction in the form error of the fabricated infrared ChG aspherical lens, decreasing the Peak-to-Valley (PV) value from 1.04 μm to 0.38 μm. • We propose a novel aspherical mold compensation method based on the random forest algorithm, which is applied to precision molding of Infrared Chalcogenide Glass. • Aspherical mold machining is performed on the electroless nickel-phosphorus (Ni–P) surface for precision molding, and the profile data of lens are extracted for construct the mold compensation machining model. • Through this compensation approach, we have demonstrated a substantial 63.5 % reduction in the form error of the fabricated infrared Chalcogenide Glass aspherical lens, decreasing the Peak-to-Valley (PV) value from 1.04 μm to 0.38 μm. [ABSTRACT FROM AUTHOR]

Traffic prediction is essential for transportation planning, resource allocation, congestion management and enhancing travel experiences. This study optimizes data preprocessing techniques to improve machine learning-based traffic prediction models. Data preprocessing is critical in preparing the data for machine learning models. This study proposes an approach that optimizes data preprocessing techniques, focusing on flow-based analysis and optimization, to enhance traffic prediction models. The proposed approach explores fixed and variable orders of data preprocessing using a genetic algorithm across five diverse datasets. Evaluation metrics such as root mean squared error (RMSE), mean absolute error (MAE) and R-squared error assess model performance. The results indicate that the genetic algorithm’s variable order achieves the best performance for the ArcGIS Hub and Frementon Bridge Cycle datasets, fixed order one preprocessing for the Traffic Prediction dataset and variable order using the genetic algorithm for the PeMS08 dataset. Fixed order 2 preprocessing yields the best performance for the XI AN Traffic dataset. These findings highlight the importance of selecting the appropriate data preprocessing flow order for each dataset, improving traffic prediction accuracy and reliability. The proposed approach advances traffic prediction methodologies, enabling more precise and reliable traffic forecasts for transportation planning and management applications. [ABSTRACT FROM AUTHOR]

Machine learning-based prediction of the output characteristics of a negative capacitance (NC) double gate (DG) Tunnel FET (TFET) (NC DG TFET) has been demonstrated in this paper. On current (ION), OFF current (IOFF) and sub-threshold slope (SS) have been predicted for different values of the body thickness (TSi), oxide thickness (TOX) and channel length (LCh) and the work function (WF). Random Forest Regression (RFR) model is used in the ML framework to explore the device characteristics. A total of 5444 simulated data has been generated by using the SILVACO ATLAS tool. After pre-processing the data, RFR-ML has been applied to obtain maximum accuracy. Root means square error (RMSE) (~0.34), R2 score (~0.965) and accuracy (~98.51%) parameter have been used as the figures of merit (FOM) to judge the efficiency of the proposed model with respect to the state-of-the-art literature. Besides, with higher accuracy, the model is capable of predicting the data within a few seconds compared to the few hundred hours using TCAD tools. This fast and accurate design strategy will surely be efficacious for the present VLSI industries. [ABSTRACT FROM AUTHOR]

Conventional concrete causes significant environmental problems, including resource depletion, high CO2 emissions, and high energy consumption. Steel slag aggregate (SSA), a by-product of the steelmaking industry, offers a sustainable alternative due to its environmental benefits and improved mechanical properties. This study examined the predictive power of four modeling techniques—Gene Expression Programming (GEP), an Artificial Neural Network (ANN), Random Forest Regression (RFR), and Gradient Boosting (GB)—to predict the compressive strength (CS) of SSA concrete. Using 367 datasets from the literature, six input variables (cement, water, granulated furnace slag, superplasticizer, coarse aggregate, fine aggregate, and age) were utilized to predict compressive strength. The models' performance was evaluated using statistical measures such as the mean absolute error (MAE), root mean squared error (RMSE), mean values, and coefficient of determination (R2). Results indicated that the GB model consistently outperformed RFR, GEP, and the ANN, achieving the highest R2 values of 0.99 and 0.96 for the training and testing dataset, respectively, followed by RFR with R2 values of 0.97 (training) and 0.93 (testing), GEP with R2 values of 0.85 (training) and 0.87 (testing), and ANN with R2 values of 0.61 (training) and 0.82 (testing). Additionally, the GB model had the lowest MAE values of 0.79 MPa (training) and 2.61 MPa (testing) and RMSE values of 1.90 MPa (training) and 3.95 MPa (testing). This research aims to advance predictive modeling in sustainable construction through analysis and well-defined conclusions. [ABSTRACT FROM AUTHOR]

Crop yield prediction plays a key role in modern agriculture, it enables farmers to make decisions about resource distribution, crop production management, and marketing business strategies. Regression models are extensively used for crop yield prediction. The performance of different regression techniques may vary depending on various factors such as the dataset, features, and modeling assumptions. In this paper, a comparative study was conducted to evaluate and compare the performance of different regression models for agriculture crop yield prediction. Collected a comprehensive dataset encompassing historical crop yield data, weather parameters and pesticides data features from various agricultural regions, then applied and compared various regression models, including Linear Regression (LR), K Nearest neighbor Regression (KNR), Support Vector Regression (SVR), Decision Tree Regression (DTR), Random Forest Regression (RFR), Gradient Boosting Regression (GBR),Linear Model Lasso Regression, Elasticnet Regression, Ridge Regression to predict crop yields for various crops. This study involved evaluating the performance of these regression models based on several performance metrics, including R² score, Root Mean Squared Error(RMSE), Mean Squared Error(MSE), Mean Absolute Error(MAE), Median Absolute Error(Median AE), Explain variance score and computing time. The results of our study provide insights into the comparative performance of different regression models for crop yield prediction in agriculture. Determined that the performance of the regression models varies crop type, area, and dataset used. Overall, the random forest regression model demonstrated the best performance in terms of R2, followed by K Nearest neighbor with hyper parameter tuning and decision tree regression. However, the choice of the most suitable regression model may also depend on other factors such as the interpretability and computational efficiency requirements of the application. Our research findings contribute to the existing literature on crop yield prediction in agriculture and afford treasured information for farmers, policymakers, and researchers to make conversant conclusions about the selection of appropriate regression models for crop yield prediction in their specific contexts. Further research could explore the combination of different regression models or the integration of other ML techniques to better the R2and robustness of crop yield prediction models in agriculture. [ABSTRACT FROM AUTHOR]

Soil moisture (SM) is a crucial criterion for agronomics and the management of water resources, particularly in areas where the socio-economic status and significant source of income depend upon agriculture and related sectors. This paper intends to estimate SM over the vegetative area using a generalized regression neural network (GRNN) and ground scatterometer and compare the results with SM retrieved using Sentinel-1 data. At the same time, random forest regression (RFR) and support vector regression (SVR) models are used for SM estimation. Correlation analysis results concluded that L-band HV-polarization at 300 incidence angle showed the highest correlation with the measured field parameters. This study investigated backscattering coefficients, VV/VH polarization ratio and polarization phase difference over wheat's entire growth phase to estimate SM. The results indicate that the GRNN with backscattering coefficients and polarization ratio provided the highest accuracy compared to the random forest (RF) and SVR with the root mean square error of 0.093 over the Yavatmal District, Maharashtra, India. [ABSTRACT FROM AUTHOR]

Land Surface Temperature (LST) products obtained by thermal infrared (TIR) remote sensing contain considerable blank areas due to the frequent occurrence of cloud coverage. The studies on the all-time reconstruction of the cloud-covered LST of geostationary meteorological satellite LST products are relatively few. To accurately fill the blank area, a hybrid method for reconstructing hourly FY-4A AGRI LST under cloud-covered conditions was proposed using a random forest (RF) regression algorithm and Savitzky-Golay (S-G) filtering. The ERA5-Land surface cumulative net radiation flux (SNR) reanalysis data was first introduced to represent the change in surface energy arising from cloud coverage. The RF regression method was used to estimate the LST correlation model based on clear-sky LST and the corresponding predictor variables, including the normalized difference vegetation index (NDVI), the normalized difference water index (NDWI), surface elevation and slope. The fitted model was then applied to reconstruct the cloud-covered LST. The S–G filtering method was used to smooth the outliers of reconstructed LST in the temporal dimension. The accuracy evaluation was performed using the measured LST of the representative meteorological stations after scale correction. The coefficients of determination derived with the reference LST were all above 0.73 on the three examined days, with a bias of −1.13–0.39 K, mean absolute errors (MAE) of 1.46–2.4 K, and root mean square errors (RMSE) of 1.77–3.2 K. These results indicate that the proposed method has strong potential for accurately restoring the spatial and temporal continuity of LST and can provide a solution for the production and research of gap-free LST products with high temporal resolution. [ABSTRACT FROM AUTHOR]

The use of fly ash (FA)-based geopolymer concrete as a low-carbon and eco-friendly substitute to Portland cement concrete has gained attention in recent years. However, accurately predicting its compressive strength remains a challenge due to the complex chemical and physical interactions involved in the geopolymerization process. In this research, three machine learning models, namely backpropagation neural network (BPNN), random forest regression (RFR), and k-nearest neighbors (KNN), were employed to predict the compressive strength of FA-based geopolymer concrete. The models were trained, validated, and tested using a dataset that considered the chemical composition, mix proportions, and pre-curing conditions of the concrete. The performance of each model was assessed utilizing various metrics, including the coefficient of determination (R2), mean square error (MSE), root mean square error (RMSE), and mean absolute error (MAE). The results indicated that the BPNN model gave the best results with an R2 value of 0.948 relative to RFR and KNN with R2 values of 0.927 and 0.911, respectively. The permutation feature importance (PFI) index revealed that the coarse aggregate content, SiO2 content in FA, and NaOH concentration were found to have the greatest impact on the compressive strength of the FA-based geopolymer concrete. [ABSTRACT FROM AUTHOR]

In the context of food insecurity in resource-poor settings, agroecology (AE) has emerged as an important approach promoted for improving crop productivity, yet few studies have demonstrated how a combination of agroecological methods can improve crop health and thereby crop productivity. Using a geospatial approach, this study investigated whether agroecological practices can improve crop health in smallholder contexts. We compared leaf area indexes (LAIs) of crops on AE and non-AE farms and prospectively predicted the impact of AE using vegetation indexes (VIs). We found that crops on AE farms produced higher average growing season LAIs for maize and pigeon peas (1.28 m2/m2) and maize and beans (1.29 m2/m2) farms compared to 0.97 m2/m2 and 0.80 m2/m2, respectively, for the same crops on the non-AE farms. The higher LAIs suggest that the combination of farming strategies practiced on the AE farms produced healthier crops on AE farms. Random forest regression prospective predictions generated statistically significant higher LAIs for maize and beans (R2 = 0.90, root mean square error [RMSE] = 0.32 m2/m2) and maize and pigeon peas (R2 = 0.88 m2/m2, RMSE = 0.42 m2/m2) on the AE farms, but predictions for the non-AE farms were not statistically significant. The findings demonstrate that combining AE strategies can potentially improve crop productivity to enhance household food security and income in smallholder contexts. [ABSTRACT FROM AUTHOR]

Oceanic primary production (OPP) is crucial for ecosystem services and global carbon cycle. However, sensitivity to geographic and environmental characteristics limits the application of semi‐empirical OPP estimate models, such as the vertically generalized productivity model (VGPM) and its modified version, particularly in coastal regions. In addition, the difficulty in collecting necessary parameters also hampers long‐term OPP estimates. Data‐driven machine learning (ML) methods can automatically capture the relationships between the input parameters and the objective; hence, they may become new methods for global OPP estimates. In this study, the effectiveness of ML methods to estimate OPP and the key attributes influencing ML performance in different regions and seasons are discussed. First, the ML models obtain a lower root mean square error than the VGPM and Eppley‐VGPM. In addition, the random forest (RF) model achieves the best performance among the four selected ML models. The enhancement in the accuracy of OPP estimates based on the RF model is more obvious in coastal regions than in the open ocean. In the four seasons, the RF model obtains better estimates of OPP than the Eppley‐VGPM, especially for summer. Moreover, input attributes, including sea surface temperature (SST), photosynthetic active radiation (PAR), and chlorophyll‐a concentration (Chlor‐a), achieve the best performance. The suitable alternative input attributes are SST/Chlor‐a in the coastal regions, and single Chlor‐a, SST/Chlor‐a, and PAR/Chlor‐a in the open ocean. Except for the SST/PAR/Chlor‐a combination, input with Chlor‐a, that is, SST/Chlor‐a and PAR/Chlor‐a, result a relatively acceptable performance in the four seasons. Plain Language Summary: Oceanic primary production (OPP) is often estimated by traditional semi‐empirical models which are sensitive to geographic and environmental characteristics. In addition, the difficulty in collecting necessary parameters of OPP estimate algorithms also hampers long‐term OPP estimates. With the development of big data techniques, data‐driven methods, such as machine learning (ML), have been increasingly applied in oceanic feature estimates. However, the effectiveness of ML to estimate OPP and the key attributes influencing ML performance are not fully discussed. Hence, in this study, the superiority of ML methods to estimate OPP is presented, especially in coastal regions and in summer. In addition, suitable alternative input attributes for ML methods for different regions and seasons are analyzed when necessary oceanic parameters cannot be acquired simultaneously. Key Points: The effectiveness of machine learning (ML) methods for oceanic primary production estimate in different regions and seasons is analyzedThe influences of input attributes on ML methods for different regions and seasons are analyzedThe corresponding alternative input attributes are determined [ABSTRACT FROM AUTHOR]

Rainfall forecasting is critical for the economy, but it has proven difficult due to the uncertainties, complexities, and interdependencies that exist in climatic systems. An efficient rainfall forecasting model will be beneficial in implementing suitable measures against natural disasters such as floods and landslides. In this paper, a novel hybrid model of empirical mode decomposition (EMD) and random forest (RF) was developed to enhance the accuracy of annual rainfall prediction. The EMD technique was utilized to decompose the rainfall signal into six intrinsic mode functions (IMFs) to extract underlying patterns, while the RF algorithm was employed to make predictions based on the IMFs. The hybrid RF–IMF model was trained and tested using a dataset of annual rainfall in Kerala from 1871 to 2020, and its performance was compared to traditional models such as RF regression and the autoregressive moving average (ARMA) model. Mean absolute error (MAE), mean absolute percentage error (MAPE), mean squared error (MSE), root mean squared error (RMSE), and coefficient of determination or R-squared (R2) were used to compare the performances of these three models. Model evaluation metrics show that the RF–IMF model outperformed both the RF model and ARMA model. [ABSTRACT FROM AUTHOR]

Electrohydrodynamic jet (E-jet) printing has broad application prospects in the preparation of flexible electronics and optical devices. Ejection cycle time and droplet size are two key factors affecting E-jet-printing quality, but due to the complex process of E-jet printing, it remains a challenge to establish accurate relationships among ejection cycle time and droplet diameter and printing parameters. This paper develops a model based on random forest regression (RFR) for E-jet-printing prediction. Trained with 72 groups of experimental data obtained under four printing parameters (voltage, nozzle-to-substrate distance, liquid viscosity, and liquid conductivity), the RFR model achieved a MAPE (mean absolute percent error) of 4.35% and an RMSE (root mean square error) of 0.04 ms for eject cycle prediction, as well as a MAPE of 2.89% and an RMSE of 0.96 μm for droplet diameter prediction. With limited training data, the RFR model achieved the best prediction accuracy among several machine-learning models (RFR, CART, SVR, and ANN). The proposed prediction model provides an efficient and effective way to simultaneously predict the ejection cycle time and droplet diameter, advancing E-jet printing toward the goal of accurate, drop-on-demand printing. [ABSTRACT FROM AUTHOR]

[Display omitted] • Reviewed recent trend in non-fullerene organic solar cells using statistical method. • The energy loss in devices was predicted using different machine learning models. • Random Forest model showed the best performance in predicting the energy loss. • Correlation of energy level and offset with device performance was investigated. The efficiency of organic solar cells (OSCs) has been improved more than 19% recently with the development of non-fullerene acceptor materials. Further improvement is still attainable with the optimal combinations of donors and acceptors that provide minimal energy losses. In this work, a data-enabled machine-learning (ML) framework was employed to predict the energy losses in the polymer:non-fullerene acceptor based devices. Based on the collected experimental dataset, the prediction accuracies of various machine learning models were systematically compared by estimating mean absolute percentage errors (MAPE), root mean squared errors (RMSE), and person's r coefficient. The Random Forest regression model showed the best performance in predicting the energy losses with a correlation coefficient of 0.83 and relative error in the range of 0 – 20%. The predictive ability of this model was further validated using the different parameters of devices with power conversion efficiency range of 6 – 18%. Three different donor–acceptor combinations were chosen for fabricating the photovoltaic devices to fit this model into practical devices and experimentally obtained energy loss values were compared with the predicted values. In addition, the device parameters with the molecular descriptors to understand the correlation and energy loss is highly correlated with the HOMO offset. This study demonstrates that the ML approach provide an effective method to predict and virtual screen of promising donor–acceptor pairs with minimal energy loss and would be useful for developing next-generation high performance solar cell materials. [ABSTRACT FROM AUTHOR]

Europium titanate (EuTiO3) is a quantum para-electric magnetocaloric compound with potential application in green magnetic cooling technology. Super-exchange and indirect interactions existing between the magnetic moments of this compound are responsible for the observed magnetic properties such as the refrigeration capacity which falls among the key factors that characterize the usefulness of magnetic system of cooling in addressing energy crisis. Crystallographic substitutions and ionic replacement potentially shift magnetic ordering between anti-ferromagnetism to ferromagnetism and ultimately alters the refrigeration (magnetic) capacity of EuTiO 3 compound. This work proposes extreme learning machine (ELM) and random forest regression (RFR) computational approaches to model refrigeration capacity of europium titanate magnetic refrigerant with the aid of applied magnetic field, ionic concentrations and ionic radii predictors. Sigmoid activation based extreme learning machine algorithm(SGA-ELM) demonstrates superior performance over sine function activation based extreme learning machine algorithm(SIA-ELM) with improvement of 0.03 %, 16.04 % and 35.03 % on the basis of Pearson correlation coefficient (CC), root mean square error (RMSE) and mean absolute error (MAE) performance yardsticks, respectively on testing set of EuTiO 3 based magnetocaloric compounds. SGA-ELM performs better than SIA-ELM and RFR model with improvement of 35.03 % and 66.84 %, respectively while SIA-ELM model performs better than RFR model with improvement of 59.97 %. SGA-ELM model developed in this work outperforms the existing swarm based support vector regression (SW-SVR) models with performance improvement of 15.38 % (for SW-SVR-Gas with Gaussian function) and 42. 71 % % (for SW-SVR-Pol with polynomial function) using mean absolute deviation (MAD) performance yardstick. Significance of magnetic field and dopants concentrations on refrigeration capacity of different EuTiO 3 perovskite was investigated with SGA-ELM model. The accuracy of developed models in determining refrigeration capacity of EuTiO 3 compounds would ultimately circumvents experimental challenges and explores future potentials of these classes of compound for addressing present and future energy crisis in cooling industries. [Display omitted] • SIA-ELM and SGA-ELM models are developed for estimating RC of EuTiO 3 compounds. • Developed models outperform the existing models using different assessment metrics. • Effect of field on RC of EuTi 1-x Mn x O 3 magnetocaloric compounds was investigated. • Effect of dopant on EuTi 1-x Mn x O 3 magnetocaloric compound was established. [ABSTRACT FROM AUTHOR]

Ischemic stroke (IS) is one of the top risk factors for death and disability. Meteorological conditions have an effect on IS attack. In this study, we try to develop models of medical meteorological forecast for IS attack based on machine learning and deep learning algorithms. The medical meteorological forecast would be beneficial to public health in IS events prevention and treatment. We collected data on IS attacks and climatology in each day from 18th September 2016 to 31th December 2020 in Haikou. Data on IS attacks were from the number of hospital admissions due to IS attack among general population. The random forest (RF) regression and long short-term memory (LSTM) algorithms were respectively used to develop the predictive model based on meteorological data. Performance of the model was assessed by mean squared error (MSE) and root mean squared error (RMSE). A total of 42849 IS attacks was included in this study. IS attacks were significantly decreased in winter. The pattern of climatological data was observed the regularity in seasons. For the performance of RF regression model, the MSE is 243, and the RMSE is 15.6. For LSTM model, the MSE is 36, and the RMSE is 6. In conclusion, LSTM model is more accurate than RF regression model to predict IS attacks in general population based on meteorological data. LSTM model showed acceptable accuracy for the prediction and could be used as medical meteorological forecast to predict IS attack among population according to local climate. [ABSTRACT FROM AUTHOR]

Predicting rainfall is essential for many applications, including agriculture, hydrology, and disaster management. In this work, we undertake a comparison examination of various machine learning models to forecast rainfall based on meteorological data. The target variable in this study is rainfall, and the dataset used includes characteristics like temperature, relative humidity, wind speed, and wind direction. The following seven machine learning models were assessed: Support Vector Regression (SVR), Multivariate adaptive regression splines (MARS), Random Forest Regression, and Deep Neural Network with Historical Data (DWFH), Haar Wavelet Function, Decision Tree and Discrete wavelet Transform (DWT). Data preprocessing, which includes standardisation and lagging to capture temporal dependencies, comes first in the analysis phase. A wavelet transformation is also used to capture complex patterns in the data. Each model is tested on a different test set after being trained on a subset of the dataset. The results are assessed using the Root Mean Squared Error (RMSE) and Mean Squared Error (MSE), focusing on the RMSE and MSE values for better comparison across models. Our findings reveal that the DWFH model achieved an RMSE of 0.0138807 mm and MSE of 0.000193 mm2, demonstrating their effectiveness in predicting rainfall. The Random Forest and SVR models also provided competitive results. This study highlights the importance of selecting an appropriate machine learning model for rainfall prediction and the significance of preprocessing techniques in improving model performance. These insights can aid decision-makers in choosing the most suitable model for their specific application, contributing to more accurate rainfall predictions and enhanced decision support systems. • ML models are more efficient at capturing rainfall dynamics than conventional methods. • This study proved the ability of four ML models to forecast daily rainfall using data. • The results are specific to the Subarnarekha Basin, Odisha, but could have broader impacts. • With an RMSE of 0.0139 mm, the DWFH model outperformed others in rainfall forecasting. [ABSTRACT FROM AUTHOR]

Traditional crop cutting experiment-based yield estimation method captures the regional yield variability but lacks field-level information. Satellite images hold enormous crop information at finer spatial resolution. Crop yield mapping with optical images is particularly challenging if cloud-free images are unavailable during the crucial crop developmental stages. All-weather availability and sensitivity to crop structure, dielectric properties make synthetic aperture radar (SAR) images an excellent resource for yield estimation. Both types of data provide complementary information about crop conditions. A random forest regression model with genetic algorithm-based feature selection is developed to exploit the Sentinel-2 optical and Sentinel-1 SAR images for yield estimation. We utilized the crop harvest and quality survey (BEE) yield data set collected by the Hessisches Statistisches Landesamt (HSL), Wiesbaden, Germany, over 490 fields. We prepared 20m resolution yield maps for winter wheat, winter barley, winter rye and winter rapeseed. Input features for the yield estimation model are selected based on the prior knowledge of remote sensing of vegetation. Baseline random forest regression models are developed for all the four crop types with optical and SAR input features. An optimized random forest regression model with genetic algorithm-based feature selection results in performance improvement. Dissimilarity in genetic algorithm selected image features highlights the significance of crop-specific feature selection for yield estimation. The optimized models reliably estimate yield by achieving correlation coefficient (r) of 0.65-0.86, mean absolute error 0.93-1.16 t ha-1 and root mean square error 1.12-1.56 t ha-1 with BEE yield on testing data set. The proposed models could estimate the intra-field yield variation when winter wheat, winter barley, winter rye were in the shooting phase to the beginning of ear-shifting, and winter rapeseed began to flower or was already flowering. These results demonstrate the merits of our model for early-stage crop yield estimation at the field level with mono-temporal image and adaptability for the cropping season with high cloud cover. [ABSTRACT FROM AUTHOR]

Monitoring and management of grapevine water status (GWS) over the critical period between flowering and veraison plays a significant role in producing grapes of premium quality. Although unmanned aerial vehicles (UAVs) can provide efficient mapping across the entire vineyard, most commercial UAV-based multispectral sensors do not contain a shortwave infrared band, which makes the monitoring of GWS problematic. The goal of this study is to explore whether and which of the ancillary variables (vegetation characteristics, temporal trends, weather conditions, and soil/terrain data) may improve the accuracy of GWS estimation using multispectral UAV and provide insights into the contribution, in terms of direction and intensity, for each variable contributing to GWS variation. UAV-derived vegetation indices, slope, elevation, apparent electrical conductivity (ECa), weekly or daily weather parameters, and day of the year (DOY) were tested and regressed against stem water potential (Ψstem), measured by a pressure bomb, and used as a proxy for GWS using three machine learning algorithms (elastic net, random forest regression, and support vector regression). Shapley Additive exPlanations (SHAP) analysis was used to assess the relationship between selected variables and Ψstem. The results indicate that the root mean square error (RMSE) of the transformed chlorophyll absorption reflectance index-based model improved from 213 to 146 kPa when DOY and elevation were included as ancillary inputs. RMSE of the excess green index-based model improved from 221 to 138 kPa when DOY, elevation, slope, ECa, and daily average windspeed were included as ancillary inputs. The support vector regression best described the relationship between Ψstem and selected predictors. This study has provided proof of the concept for developing GWS estimation models that potentially enhance the monitoring capacities of UAVs for GWS, as well as providing individual GWS mapping at the vineyard scale. This may enable growers to improve irrigation management, leading to controlled vegetative growth and optimized berry quality. [ABSTRACT FROM AUTHOR]

Soil salinity and alkalinity seriously threaten crop production, soil productivity, and sustainable agriculture, especially in arid and semi-arid areas, leading to land degradation. Therefore, the spatial distribution of these parameters is really important for the successful management of such areas. The salinity and sodium adsorption ratio (SAR) of soil surface have been modeled in this article. Auxiliary data were terrain attributes derived from the digital elevation model (DEM), remote sensing spectral bands, and indices of vegetation and salinity derived from the Landsat 8 OLI satellite. In total, 118 soil samples were collected from a depth of 0–15 cm in homogenous units at Doviraj plain in the southern part of Ilam province, western Iran. Saturated electrical conductivity (ECe), SAR, and other soil properties were analyzed and calculated. To model ECe and SAR parameters with the auxiliary data, stepwise multiple linear regression (SMLR) and random forest (RF) regression were applied. The highest accuracy was obtained through the RF model with validation coefficient of determination (R2) = 0.82 and 0.83 and validation root mean square error (RMSE) = 7.40 dS/m and 11.20 for ECe and SAR, respectively. Furthermore, results indicated that the strongest influence on the prediction of soil salinity was obtained with Band10, principal component analysis (PC3), vertical distance to channel network (VDCN), and analytical hill-shading (AH). Also, Band10, Band11, flow accumulation (FA), and topographic wetness index (TWI) were the important covariates in alkalinity prediction through the RF model. Finally, it is suggested that similar techniques can be used to map and monitor soil salinity and alkalinity in other parts of arid regions. [ABSTRACT FROM AUTHOR]

Crop yield is related to household food security and community resilience, especially in smallholder agricultural systems. As such, it is crucial to accurately estimate within-season yield in order to provide critical information for farm management and decision making. Therefore, the primary objective of this paper is to assess the most appropriate method, indices, and growth stage for predicting the groundnut yield in smallholder agricultural systems in northern Malawi. We have estimated the yield of groundnut in two smallholder farms using the observed yield and vegetation indices (VIs), which were derived from multitemporal PlanetScope satellite data. Simple linear, multiple linear (MLR), and random forest (RF) regressions were applied for the prediction. The leave-one-out cross-validation method was used to validate the models. The results showed that (i) of the modelling approaches, the RF model using the five most important variables (RF5) was the best approach for predicting the groundnut yield, with a coefficient of determination (R2) of 0.96 and a root mean square error (RMSE) of 0.29 kg/ha, followed by the MLR model (R2 = 0.84, RMSE = 0.84 kg/ha); in addition, (ii) the best within-season stage to accurately predict groundnut yield is during the R5/beginning seed stage. The RF5 model was used to estimate the yield for four different farms. The estimated yields were compared with the total reported yields from the farms. The results revealed that the RF5 model generally accurately estimated the groundnut yields, with the margins of error ranging between 0.85% and 11%. The errors are within the post-harvest loss margins in Malawi. The results indicate that the observed yield and VIs, which were derived from open-source remote sensing data, can be applied to estimate yield in order to facilitate farming and food security planning. [ABSTRACT FROM AUTHOR]

Major fields such as military applications, medical fields, weather forecasting, and environmental applications use wireless sensor networks for major computing processes. Sensors play a vital role in emerging technologies of the 20th century. Localization of sensors in needed locations is a very serious problem. The environment is home to every living being in the world. The growth of industries after the industrial revolution increased pollution across the environment. Owing to recent uncontrolled growth and development, sensors to measure pollution levels across industries and surroundings are needed. An interesting and challenging task is choosing the place to fit the sensors. Many meta-heuristic techniques have been introduced in node localization. Swarm intelligent algorithms have proven their efficiency in many studies on localization problems. In this article, we introduce an industrial-centric approach to solve the problem of node localization in the sensor network. First, our work aims at selecting industrial areas in the sensed location. We use random forest regression methodology to select the polluted area. Then, the elephant herding algorithm is used in sensor node localization. These two algorithms are combined to produce the best standard result in localizing the sensor nodes. To check the proposed performance, experiments are conducted with data from the KDD Cup 2018, which contain the name of 35 stations with concentrations of air pollutants such as PM, SO2, CO, NO2, and O3. These data are normalized and tested with algorithms. The results are comparatively analyzed with other swarm intelligence algorithms such as the elephant herding algorithm, particle swarm optimization, and machine learning algorithms such as decision tree regression and multi-layer perceptron. Results can indicate our proposed algorithm can suggest more meaningful locations for localizing the sensors in the topology. Our proposed method achieves a lower root mean square value with 0.06 to 0.08 for localizing with Stations 1 to 5. [ABSTRACT FROM AUTHOR]

Water temperature in rivers is the key property determining the biotic and abiotic processes occurring in these ecosystems. In many regions of the world, the significant lack of measurement data is a serious problem. This paper presents reconstruction of water temperature for selected Polish rivers with monitoring discontinued in the period 2015–2020. Information regarding air temperature and water temperature in lakes provided the basis for the comparison of three models: multiple linear regression, random forest regression, and multilayer perceptron network. The results show that the best reconstruction results were obtained with a multilayer perceptron network model based on water temperatures in the lake and air temperatures from three meteorological stations. The average values of mean error, root mean square error and standard error were for the rivers in Poland: 1.52 °C, 5.03%, and 0.47 °C. The course of mean yearly water temperature in the years 1987–2020 showed a statistically significant increase from 0.18 to 0.49 °C per decade. The results show that the largest increases occurred in June, August, September, November, and December. [ABSTRACT FROM AUTHOR]

To overcome limitations of currently used blood pressure measurement devices in accuracy, continuity and comfort, we propose an approach for blood pressure estimation from electrocardiogram (ECG) signals only. Thereby, statistical signal features are extracted from the ECG which, eventually, serve as input to a random forest regression. The method is trained and tested on MIMIC III waveform data with a large range of blood pressure values. It obtains a mean absolute error ± standard deviation of 3.73 ± 5.19 mmHg for diastolic blood pressure (DBP) and 5.92 ± 7.23 mmHg for systolic blood pressure (SBP), with Pearson coefficients ΥDBP=0.92 and ΥSBP=0.91 respectively. [ABSTRACT FROM AUTHOR]

The Electrochemical impedance spectroscopy (EIS) of three novel cathode channels designs square, boot, and triangular for open cathode polymer electrolyte cathode fuel cell stacks are analyzed. Fundamental mechanisms such as oxygen reduction reaction, proton conductivity, bulk resistance of stack, capacitance and structure features of the porous carbon electrode etc. are investigated with the help of Nyquist/Bode plots and Randles circuit. Operational parameters including airflow rate, current, H 2 gas condition (humidified and dry), and stack temperature are studied focused on understanding the trend of high and low frequency resistance. The minimum resistance and temperature mapping for cross-section designs at different operating conditions is utilized for parametric optimization. Overall, it is recommended to operate the stacks at minimum resistance obtained just before the peak power density and its corresponding temperature ∼40 ° C. Finally, a data-driven hybrid model integrating EIS and artificial intelligence is presented with root mean square error of ∼0.98. • EIS analysis of different cross-section cathode channels is carried out. • OC-PEMFC stack thermal mapping is studied to understand interdependent parameters. • Lowest frequency resistance is utilized for parametric optimization. • Hybrid EIS-AI random forest regression model is used to understand stack operation. [ABSTRACT FROM AUTHOR]

Simulating anaerobic digestion (AD) using a machine learning (ML) is important for guiding methane production in practice. But the performance of the ML is affected by operational conditions and microbial types and quantities, which lead to low accuracy of the traditional model. In this study, a combined ML algorithm SMOTER-GA-RF was constructed by expanding the number of genomic data, introducing oversampling technique (SMOTER) and genetic algorithm (GA) to increase prediction accuracy, which the test set R2 increased 36.22%. The normalized root mean square error and percent bias of the validation set (D2) were 8.92% (<10%) and 4.13% (<15%), indicating the accuracy of prediction. Simultaneously, the predictive accuracy of the validation sets (V1–V5) in the presence of Gaussian white noise remained high, confirming the robustness of the SMOTER-GA-RF. Furthermore, similar ML models were collected and compared, and it was found that SMOTER-GA-RF constructed in this paper had best effect. Finally, important factors affecting the AD performance of straw were analyzed using SHAP. The hydraulic retention time was controlled at approximately 40 d, volatile fatty acid should be controlled within 1000 mg/L, and the relative abundances of the important archaea Methanoculleus , Methanobacterium and Methanosarcina in the second stage of AD were controlled at 5–10%, <18%, and <5%. [Display omitted] • GA determined the hyper-parameters of the RF to increase the prediction performance. • SMOTER generated new samples strengthened the learning capacity of extreme data. • SMOTER-GA-RF model had good predictive ability and robustness in monitoring and controlling the AD process. • SHAP quantified the range of important features promoting the AD of straw lignocellulose. [ABSTRACT FROM AUTHOR]

Rice is the primary staple food source for Indonesian people, with consumption increasing so that rice production needs to be increased. Rice drought is one of the problems that can hamper rice production. This research aims to determine the best extraction feature between the normalized difference vegetation index (NDVI) and the normalized difference water index (NDWI) in describing rice fields' dryness. Moreover, using the random forest regression algorithm. This research compares NDVI with NDWI using data originating from Sentinel-2A and retrieved via the google earth engine. Regression algorithms are used in research to predict drought in paddy fields. This research shows that NDVI is better than NDWI in predicting drought using random forest regression algorithms and logistic regression algorithms. The random forest regression algorithm based on the results obtained shows that the average root mean square error (RMSE) on NDVI is 0.018, and NDWI is 0.012. Based on the logistic regression algorithm results, it was found that the average value of RMSE on NDVI was 0.346, and NDWI was 0.336. Based on the results of the RMSE, it shows that the forecasting ability of the random forest regression algorithm is better than the logistic regression. [ABSTRACT FROM AUTHOR]

Next-generation IEEE 802.11 WLANs when deployed in dense environments and complex situations, the throughput achieved is much lower than the estimated values due to interference, overlapping of channel bandwidths and contention. Throughput estimation through simulators is tedious and needs elaborate information regarding the deployment details related to overlapping BSS scenarios. With large accurate datasets of BSS deployments, it is found to be possible to approach the problem of prediction of throughput of each BSS by using well-crafted machine learning (ML) models. In this paper, we proposed three ML approaches to predict the throughput viz artificial neural networks (ANN), k-Nearest Neighbours (kNN) regression and random forest regression. The root mean square error and the mean absolute error thus calculated in each of these approaches in the given setting are promising enough to further pursue the probe for more accurate prediction models based on machine learning. [ABSTRACT FROM AUTHOR]

Accurate water quality predicting has an essential role in improving water management and pollution control. The machine learning models have been successfully implemented for modelling total dissolved solids (TDS), sodium absorption ratio (SAR) and total hardness (TH) content in aquatic ecosystems with insufficient data. However, due to multiple pollution sources and complex behaviours of pollutants, these models' effect in predicting TDS, SAR, and TH levels in the Karun River system is still unclear. Given this problem, multiple linear regression (MLR), M5P model tree, support vector regression (SVR) and random forest regression (RFR) models were used to predict TDS, SAR and TH variables in the four stations in the Karun River for 1999–2019 period. Initially, to reduce the number of input variables, the principal component analysis (PCA) technique was used. The developed models are valued in terms of the coefficient of determination (R2) and the root mean square error (RMSE). Base on the PCA, it was found that sodium (Na), chloride (Cl) and TH and Na and Cl are the most influential inputs on TDS and SAR, respectively, while calcium (Ca) and magnesium (Mg) are the most effective on TH. The results indicated that RFR, SVR and MLR models had the lowest error in predicting TDS, SAR and TH, respectively, in all stations. RFR model had the highest performance for predicting TDS (R2= 0.98, RMSE= 70.50 mg l−1), SVR model for predicting SAR (R2= 0.99, RMSE= 0.04) and MLR model for predicting TH (R2= 0.99, RMSE= 1.54 mg l−1) in Darkhovin station. The comparison of the results indicated that the machine learning models could satisfactorily estimate the TDS, SAR and TH for all stations. [ABSTRACT FROM AUTHOR]

A sugar beet is one of the most important cash crops in northern China. It is a high demand for the rapid, accurate, and high-throughput acquisition of the fresh weight of aboveground and root, the sugar content of root, and the chlorophyll content of aboveground in the production of sugar beet. An Unmanned Aerial Vehicle (UAV) can serve as a significant approach, due to its flexibility, low cost, and high spatiotemporal resolution. In this study, a UAV equipped with digital and multispectral cameras was utilized to capture the images of sugar beet during the leaf clusters, root tuber, sugar growth, and accumulation period, thereby extracting the structural and spectral characteristics of the canopy. The estimation models were also established for the various indexes using the Random Forest Regression (RFR) and Partial Least Squares Regression (PLSR), including the fresh weight of shoot and root tuber, the sugar content of root tuber, and Soil Plant Analysis Development (SPAD) value during the whole period of sugar beet. The results showed that the RFR and PLSR model performed well to predict the fresh weight and sugar content of shoot and root tuber, with the coefficient of determination R² ranging from 0.9 to 0.94 and from 0.88 to 0.9, respectively, while the relative Root Mean Square Error (rRMSE) ranging from 7.6% to 17% and from 8.8% to 20%, respectively. Both models presented weak predictions for the SPAD values, where the R² values were only 0.66 and 0.67, respectively. Furthermore, a Permutation Importance (PIMP) was used to screen the more sensitive variables with the dominated impacts on the prediction, in order to reduce the size of the input variable set for the less cost and complexity of data collection. As such, the optimal prediction models of RFR and PLSR were achieved for the growth monitoring of sugar roots. It was found that excellent predictions were achieved on the fresh weight and sugar content of shoot and root tuber, with the R² value ranging from 0.89 to 0.94, and from 0.74 to 0.91, respectively, and the rRMSE value ranging from 7.3% to 19% and from 7.6% to 19%, respectively. Nevertheless, the RFR and PLSR model presented weak predictions for the SPAD values, where the R² values were only 0.65 and 0.68, respectively. Correspondingly, the accuracy of the RFR model was slightly better than that of the PLSR model. More importantly, the PIMP variable screening can be widely expected to reduce the complexity of data collection with optimal accuracy. Consequently, the canopy structure and spectral features obtained by UAVs can be utilized to quickly and accurately monitor the growth and sugar content of sugar beet. The finding can provide a strong reference to estimate the root active substances of tubers crops using UAV proximity. [ABSTRACT FROM AUTHOR]

The present study aims to evaluate the effect of vegetation on land surface temperature (LST) in different land uses and covers in Vilnius district in 1999 and 2019. To that end, in addition to mono-window and split-window algorithms that help estimate the LST, the variables digital elevation model (DEM), slope, heat load index (HLI), distances from the road and the water, the normalized difference vegetation index (NDVI), and the normalized difference water index (NDWI) affecting the surface temperature were used. Furthermore, the random forest regression (RFR) method was applied to evaluate the effect of the mentioned variables on the LST. The performance model was also assessed by using the mean absolute (MAE), mean squared (MSE), and root mean square error (RMSE). Based on the results, NDVI and NDWI indexes had the greatest impact on the temperature of Vilnius city, respectively. The study area images were categorized as built-up area, cropland, semi-forest land, dense forest land, water bodies, pastures, and green urban areas. It was found that the pastures in 1999 and the built-up class in 2019 received the highest temperature from the land surface and that the classes characterized by natural land cover such as forest land and agricultural and water bodies had a relatively low surface temperature. NDVI response curves in both 1999 and 2019 indicated that the higher the density of vegetation on the land surface, the lower the surface temperature. A lower rate of urbanization, a higher density of vegetation and consequently, a lower the temperature of the land surface were recorded for 1999 in comparison with 2019. Therefore, urbanization was demonstrated to play a significant role in changes in LULC and the increase in LST. [ABSTRACT FROM AUTHOR]

Harvest time is an important factor affecting grain yield and postharvest quality. The estimation of crop maturity greatly supports farmers in the optimization of the harvest time. This study proposed a method for predicting paddy rice maturity (J816 and 5Y4 varieties) using color features and the random forest (RF) regression algorithm. Paddy panicle images were obtained using a flatbed scanner during the maturation period. To estimate paddy rice maturity, 22 color features representing the greenness of crop leaves were extracted from the paddy panicle images. Stepwise regression was used to select superior features as inputs to the RF regression model. The coefficient of determination (R2) and root mean square error (RMSE) values of the model were 0.93 and 1.18% for J816 and 0.94 and 1.60% for 5Y4, respectively. The results indicate that the proposed method in this study is a promising technique for the estimation of paddy rice maturity. [ABSTRACT FROM AUTHOR]

The major goal of this research was to assess the ability of visible-near-infrared (VIS-NIR) spectroscopy combined with multivariate regressions based on four modeling algorithms, namely partial least squares regression (PLSR), principal component regression (PCR), support vector regression(SVR), and random forest (RF) to quantify some quality soil parameters. Saturated electrical conductivity (ECe), organic carbon (OC), and texture (including sand, silt, and clay) were measured. Overall, 114 soil samples were taken from 30 profiles (0–100 cm) in Doviraj plain, western Iran. For improving spectral data, various preprocessing techniques including Savitzky-Golay first deviation (SGD), Normalization (N), and Standard Normal Variate transformation (SNV) were employed. A leave-one-out cross-validation procedure was used. Results revealed that RF model produced much more precise predictions compared with PCR, SVR, and PLSR models and specifying the most important variables for the model. Also, the highest accuracy for RF model was achieved by SGD preprocessing for ECe, OC, sand, and clay with validation coefficient of determination (R2val) equal to 0.57, 0.75, 0.70, and 0.73, respectively; also, validation root mean square error (RMSEval) was equal to 0.74 dS/m, 0.13, 12.97, and 0.14%, respectively, However, SNV preprocessing with RF model caused to best result with R2val = 0.65 and RMSE val = 9.21% for silt. This methodology displays the capability of the VIS-NIR for estimation of ECe, OC, and texture. [ABSTRACT FROM AUTHOR]

Nitrogen is a large number of elements required by crop life and is closely related to crop yield and quality. Accurate diagnosis and dynamic monitoring of crops are important contents of precision agriculture research. With the advantages of high timeliness, high resolution, and low cost, Unmanned Aerial Vehicle (UAV) remote sensing is developing rapidly in the diagnosis and monitoring of nitrogen nutrition in winter wheat. In this study, Nitrogen Nutrition Index (NNI) was selected as an indicator of the nitrogen nutritional status of winter wheat. In order to quantitatively estimate the nitrogen nutrition index of winter wheat at a regional scale, hyperspectral images of the winter wheat experimental plots in Qianxian, Shaanxi province during the year 2016-2017 was obtained by the Cubert UHD185 imaging spectrometer boarded on the UAV. Meanwhile, the above-ground biomass and plant nitrogen concentration data were collected during the flying and were used to calculate the NNI according to the local critical nitrogen concentration dilution model of wheat. Three types of spectral parameters were selected, including trilateral parameters, any two bands spectral indices, and vegetation indices. According to the correlation analysis between spectral parameters and plant nitrogen content, aboveground biomass, and nitrogen nutrition index, ten spectral parameters significantly related to each of the three indicators were screened out, including red-edge area, green peak reflectance maximum, red valley reflectivity minimum, difference spectral index, normalized spectral index, ratio spectral index, red-edge position by linear interpolation, red-edge normalized difference vegetation index, Vogelmann red-edge index, and photochemical reflectance index. Finally, nitrogen nutrition index estimation models of winter wheat at the heading stage were established based on the ten spectral parameters above with simple regression, Multiple Linear Stepwise Regression (MLSR), Partial Least Squares Regression (PLSR), and Random Forest Regression (RFR) algorithms respectively. The coefficient of determination, Root Mean Square Error (RMSE), and Relative Prediction Deviation (RPD) were used to compare the accuracy of each model. The results showed that any two bands’ spectral indices were sensitive to the nitrogen nutrition index. The correlation coefficients between any two bands spectral indices and nitrogen nutrition index were dramatically superior to trilateral parameters and typical vegetation indices. The optimal any two bands spectral index was the ratio spectral index made up of the reflectance at 718 and 738 nm. Of all the nitrogen nutrition index estimation models by simple regression, Difference Spectral Index (DSI) and Red-edge Normalized Difference Vegetation Index (RNDVI) had rough estimation capabilities of nitrogen nutrition index, the relative prediction deviation of which were 1.53 and 1.56 respectively. Among the nitrogen nutrition index estimation models of winter wheat established through ten spectral parameters based on multiple linear stepwise regression, partial least squares regression and random forest regression, the model based on the random forest algorithm had the highest precision, strong reliability, and excellent prediction ability. The validation results showed that the coefficient of determination of the model was 0.79, the root mean square error was 0.13, and the relative prediction deviation was 2.25. The second was the model through the partial least squares regression method, the relative prediction deviation of which was1.54. The model by multiple linear stepwise regression methods could not be used to estimate the nitrogen nutrition index correctly. Overall, the model with the random forest algorithm was considered as the best estimation model of the nitrogen nutrition index, which could be used for the remote sensing mapping of the winter wheat nitrogen nutrition index at a regional scale. It would provide a scientific basis for the diagnosis of nitrogen nutrition, monitoring of yield and quality, and late field management of winter wheat. [ABSTRACT FROM AUTHOR]

Aim: To evaluate the transferability between fire recurrence scenarios of post‐fire vegetation cover models calibrated with satellite imagery data at different spatial resolutions within two Mediterranean pine forest sites affected by large wildfires in 2012. Location: The northwest and east of the Iberian Peninsula. Methods: In each study site, we defined three fire recurrence scenarios for a reference period of 35 years. We used image texture derived from the surface reflectance channels of WorldView‐2 and Sentinel‐2 (at a spatial resolution of 2 m × 2 m and 20 m × 20 m, respectively) as predictors of post‐fire vegetation cover in Random Forest regression analysies. Percentage vegetation cover was sampled in two sets of field plots with a size roughly equivalent to the spatial resolution of the imagery. The plots were distributed following a stratified design according to fire recurrence scenarios. Model transferability was assessed within each study site by applying the vegetation cover model developed for a given fire recurrence scenario to predict vegetation cover in other scenarios, iteratively. Results: For both wildfires, the highest model transferability between fire recurrence scenarios was achieved for those holding the most similar vegetation community composition regarding the balance of species abundance according to their plant‐regenerative traits (root mean square error [RMSE] around or lower than 15%). Model transferability performance was highly improved by fine‐grained remote‐sensing data. Conclusions: Fire recurrence is a major driver of community structure and composition so the framework proposed in this study would allow land managers to reduce efforts in the context of post‐fire decision‐making to assess vegetation recovery within large burned landscapes with fire regime variability. [ABSTRACT FROM AUTHOR]

Soil moisture content (SMC) plays an important role in the hydrological functioning of wetlands. Remote sensing shows potential for the quantification and monitoring of the SMC of palustrine wetlands; however, this technique remains to be assessed across a wetland-terrestrial gradient in South Africa. The ability of the Sentinel Synthetic Aperture Radar (SAR) and optical sensors, which are freely available from the European Space Agency, were evaluated to predict SMC for a palustrine wetland and surrounding terrestrial areas in the grassland biome of South Africa. The percentage of volumetric water content (%VWC) was measured across the wetland and terrestrial areas of the Colbyn Wetland Nature Reserve, located in the City of Tshwane Metropolitan Municipality of the Gauteng Province, using a handheld SMT-100 soil moisture meter at a depth of 5 cm during the peak and end of the hydroperiod in 2018. The %VWC was regressed against the Sentinel imagery, using random forest, simple linear and support vector machine regression models. Random forest yielded the highest prediction accuracies in comparison to the other models. The results indicate that the Sentinel images have the potential to be used to predict SMC with a high coefficient of determination (Sentinel-1 SAR = /P>0.9; Sentinel-2 optical = >0.9) and a relatively low root mean square error (Sentinel-1 RMSE =<17%; Sentinel-2 optical = RMSE <21%). Predicted maps show higher ranges of SMC for wetlands (> 50%VWC; p<0.05) compared to terrestrial areas, and therefore SMC monitoring may benefit the inventorying of wetlands, as well as monitoring of their extent and ecological condition. Significance: • The freely available and space-borne Sentinel sensors show potential for the quantification of surface soil moisture across a wetland-terrestrial gradient • Significant differences between the surface soil moisture of palustrine wetlands and terrestrial areas, imply that inventorying and monitoring of the extent and hydroperiod of palustrine wetlands can potentially be done. [ABSTRACT FROM AUTHOR]

Copyright of Forest Research is the property of Forest Research Editorial Office and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Sea Level Rise (SLR) is one of the most difficult elements to predict in the hydrological cycle. 12% of the area of Peninsular Malaysia, where the western low plains of muddy sediment are home to 2.5 million people, is vulnerable to flooding. In this study, two Artificial Intelligence (AI) techniques were used to predict SLR, namely, the Multi-Layer Perceptron Neural Network (MLP-NN) and Random Forest Regression (RFR) techniques. This studied, two cases were presented. The first case (Case 1) was to establish the prediction model for SLR by a monthly data set, while the second case (Case 2) was by means of a cyclical data set. From sensitivity analysis result, it was found that the most effective meteorological input parameters were rainfall (mm) and wind direction (degree). The performance of the models was evaluated according to three statistical indices in terms of the correlation coefficient (R), root mean square error (RMSE) and scatter index (SI). A comparison of the results of the MLP-NN and RFR showed that the MLP-NN performed better than the latter as the R obtained in Case 2 of the MLP-NN was 0.733 with 65.652 and 2.735 for RMSE and SI respectively. Meanwhile, accuracy improvement percentage (%AI) was 8%. [ABSTRACT FROM AUTHOR]

Downscaling satellite‐based precipitation to fine scales is crucial for deepening our understanding of global hydrologic cycles and water‐related issues. In this study, a novel approach that integrates precipitation zoning with random forest regression is proposed for the spatial downscaling of satellite‐based precipitation. Precipitation zoning is delineated through iterative rotated empirical orthogonal function (REOF) analyses of ground‐ and satellite‐based precipitation observations. Random forest regression is applied to link the satellite‐based precipitation to 1‐km‐resolution predictors such as latitude (Lat), longitude (Lon), elevation, aspect, slope, and the normalized difference vegetation index (NDVI). The accuracy of the resultant downscaled precipitation is evaluated based on five statistical evaluation indices. The performance of the proposed approach is exemplified in the Lancang–Mekong River basin, taking Tropical Rainfall Measuring Mission (TRMM) 3B43 and Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks–Climate Data Record (PERSIANN–CDR) products acquired in 2001 (wet year), 2005 (normal year), and 2009 (dry year) as the databases. The results show that seven precipitation subregions can be roughly distinguished in the study basin. Zoning‐based downscaling outperforms non‐zoning‐based downscaling in terms of accuracy, resulting in statistically significant reductions in root mean square error (RMSE) and mean absolute error (MAE) of 1–17% across the entire basin. Among the selected predictors, the variables Lat and Lon are the most important for precipitation estimation, whereas the remaining variables have lesser and subregion‐dependent importance. The proposed approach is promising for generating high‐spatial‐resolution precipitation data in regions with sparse ground‐based observations and differentiated climatic regimes. [ABSTRACT FROM AUTHOR]

• Propose a fatigue strength prediction method based on Bys-RFR model. • Highly accurate prediction of fatigue strength with easily measured parameters. • Ferrous alloys prediction applications validate the feasibility and superiority of Bys-RFR. Fatigue is a prevalent failure mode in mechanical structures, and the significance of fatigue strength as a process parameter for predicting the life of a material or structure cannot be emphasized enough. Given the high cost and time consumption associated with traditional experimental approaches for determining fatigue strength, which leads to prolonged analysis and design cycles, it is crucial to investigate efficient and precise methods for predicting fatigue strength. In this paper, we present a fatigue strength prediction method that utilizes Bayesian optimization for the random forest regression (RFR) model. During the training of the RFR model, a Bayesian optimization algorithm is employed to optimize the five main parameters of the model, mitigating the issue of overfitting or underfitting. The optimized RFR model demonstrates superior performance and accuracy in comparison to linear regression (LR), artificial neural network (ANN), decision tree regression (DTR), and support vector regression (SVR). This is determined by evaluating the mean absolute error (MAE), root mean square error (RMSE), R-squared (R2), and mean absolute percentage error (MAPE) scores of the feature parameter-fatigue strength dataset. The fatigue strength of unknown ferrous alloys can be predicted with the proposed model, which largely reduces the need to obtain fatigue strength data through traditional fatigue tests and saves fatigue analysis time and test costs. [ABSTRACT FROM AUTHOR]

The water uptake rate is a key parameter for improving irrigation efficiency and represents an indicator of tree health and yield. Here, we investigate the potential of C-band Synthetic Aperture Radar (SAR) data acquired by Sentinel-1 to estimate and map the water uptake rate across a very high-density olive orchard (5.3 km2 of olive plots) in the hot and arid desert climate of Saudi Arabia. The SAR predictor variables for water uptake rate estimation were the difference between the SAR backscattering of a given image acquired during the year (t i) and the average SAR backscattering in the second-half of January (t 0), which corresponds to the point when the uptake rate was closest to zero. Random forest regression and multi-linear regression models were used to explore the relationship between the SAR predictor variables and in situ water uptake rate, inferred from twelve sap flow meters that were installed across six plots and collected over two years (2020 and 2021). The trained random forest regression model provided an improved estimate of water uptake rate with correlation coefficient (R2) of 0.87 and root mean square error (RMSE) of 0.13 L.h−1 compared with the multi-linear regression model (R2 = 0.74, RMSE = 0.18 L.h−1). Using the random forest regression, the water uptake rate was mapped at the plot level for three years (2019, 2020, and 2021) at a temporal resolution of 6 days. Consistent with expectations, the results show that the average uptake rate over the mapped area co-varied with vapor pressure deficit (R2 = 0.82). The analysis provides a first exploration of exploiting SAR data to infer the water uptake rate across commercial scale orchards, offering important insights that can be useful for improved water management and irrigation control. • Variation in SAR backscatter time-series is a predictor of olive tree water uptake. • Random forest provides an estimation of tree water uptake with a RMSE = 0.13 L.h−1. • SAR backscattering time-series allow mapping the water uptake rate at plot level. • The mapped water uptake rate values co-varied with the vapor pressure deficit. [ABSTRACT FROM AUTHOR]

Highlights • Random forest regression is proposed for on-line battery capacity estimation. • The estimation is developed from partial charging voltage-capacity data. • Two features indicative of battery capacity fade are extracted from charging curves. • An incremental capacity analysis is used for assisting battery feature selection. Abstract Machine-learning based methods have been widely used for battery health state monitoring. However, the existing studies require sophisticated data processing for feature extraction, thereby complicating the implementation in battery management systems. This paper proposes a machine-learning technique, random forest regression, for battery capacity estimation. The proposed technique is able to learn the dependency of the battery capacity on the features that are extracted from the charging voltage and capacity measurements. The random forest regression is solely based on signals, such as the measured current, voltage and time, that are available onboard during typical battery operation. The collected raw data can be directly fed into the trained model without any pre-processing, leading to a low computational cost. The incremental capacity analysis is employed for the feature selection. The developed method is applied and validated on lithium nickel manganese cobalt oxide batteries with different ageing patterns. Experimental results show that the proposed technique is able to evaluate the health states of different batteries under varied cycling conditions with a root-mean-square error of less than 1.3% and a low computational requirement. Therefore, the proposed method is promising for online battery capacity estimation. [ABSTRACT FROM AUTHOR]

• Biodiesel production prepared by transesterification reaction. • Machine learning present a versatile approach to predicting biodiesel yield. • Four machine learning models was applied to predict biodiesel yield. • Four models compared by Correlation Coefficient and Root Mean Square Error. • Random Forest model can be more suitable to accurately predict biodiesel yield. Biodiesel prepared by transesterification reaction was a potential fuel to address the global energy issues due to obtaining from biomass resources (waste oils, micro-algae, and plant oils) and environmentally friendly. However, biodiesel production achieved by means of the transesterification process was affected by various factors such as feedstock type, reaction time, reaction temperature, and catalyst. Recently, machine learning (ML) presents a versatile approach to predicting biodiesel yield which avoids a number of experiments. Herein, we collected 13 cases with 381 individuals experimentally data and used four ML algorithms containing k-nearest neighbor algorithm (kNN), Support Vector Machine (SVM), Random Forest regression (RF), and AdaBoost regression to predict the biodiesel yield using transesterification reaction. The Random Forest regression can be more suitable to accurately predict biodiesel yield than other three ML models due to presenting a lower RMSE values for both Training (2.778) and Validation (5.178), and a higher correlation coefficient. [ABSTRACT FROM AUTHOR]

• Forage quality evaluation is essential in breeding and selecting forage crops. • In-field hyperspectral spectroscopy was utilized to evaluate 12 biomass quality traits. • Four spectral feature datasets were extracted to access prediction of quality traits. • Features were selected based on normalized difference and ratio spectral indices datasets. • High prediction accuracy was achieved with random forest after feature selection. Forage quality evaluation is essential in breeding and selecting forage crops to improve livestock health and reduce greenhouse gas emissions. A non-destructive and high-throughput process to assess forage quality can advance cultivar development research and commercial feed analysis. The overall objective of this study was to estimate 12 biomass quality traits of field peas in the 2019 and 2020 field seasons using in-field hyperspectral spectroscopy (350–2500 nm) at the leaf level. Six machine-learning models were utilized to develop the relationships between each quality trait and four feature extraction datasets (first derivative reflectance data, vegetation indices, normalized difference spectral indices, and ratio spectral indices) of leaf reflectance spectra from the hyperspectral data. The high and consistent performance of all quality traits was produced from only ridge regression, elastic net regression, and random forest regression models in first derivative reflectance and vegetation indices datasets. In addition, higher prediction performance (0.81 < R2 < 0. 93; 0.05 < root mean square error (%) < 1.80; 0.03 < mean absolute error (%) < 1.32) was found in the random forest model using normalized difference spectral indices and ratio spectral indices datasets after utilizing a feature selection technique (leave one feature out). The results suggest that the proposed evaluation approach of spectroscopy may be applied to predict field pea quality traits in a timely fashion to assist breeders and farmers in their decision-making. [ABSTRACT FROM AUTHOR]

The main purpose of this work was to develop different machine learning-based regression methods referred to as decision tree regression (DTR), generalized additive model regression (GAMR) and random forest regression (RFR) to predict bacterial population on beef. For this purpose, 2654 bacterial data points of Listeria monocytogenes, Escherichia coli and Pseudomonas spp. Which are the most investigated bacterial genera in beef were collected from the ComBase database (www.combase.cc). Temperature, salt concentration, water activity and acidity were used as the main predictor variables to estimate the growth or survival behaviour of the microorganisms in beef. The hyperparameters are optimized for proposed machine learning-based regression methods with nested cross-validation. The fitting capabilities of the proposed machine learning algorithms were compared considering their statistical indices (coefficient of determination " R 2 " and root mean square error " RMSE). Each regression method provided satisfactory predictions with being 0.931 < R 2 < 0.949 and 0.597 < RMSE < 0.692 considering each of the microorganism populations. However, the RFR yielded the best prediction capability and therefore its prediction capability was further assessed. The RFR in the external validation process provided statistical indices being 1.017 < Bias factor <1.151 and 1.137 < Accuracy factor <1.370, indicating that the random forest regression can be reliably employed as an alternative way of describing simultaneously survival and growth behaviour of microorganisms in beef and has a significant potential to be used as an alternative simulation method by skipping a secondary model step in two-step modelling approach, traditionally utilized in the predictive microbiology field. • Machine learning-based regression methods are used to describe microbial behaviour. • Machine learning methodology provides simultaneously growth and survival modelling. • Hyperparameters are optimized with nested cross-validation method. • Random forest regression method yields the best prediction capability. • Machine learning approach can be an alternative way of modelling in predictive microbiology. [ABSTRACT FROM AUTHOR]

[Display omitted] • A new normalized differential soil organic carbon spectral index was proposed. • Some measured environmental auxiliary data can be replaced by spectral indices. • NDVI, LST, SI, and CI contribute the most to the prediction of SOC. • The PSO-RFR model can better predict the SOC of coastal wetlands. To study the dynamic changes of soil organic carbon (SOC) content in coastal wetlands over the past 23 years, we use the third core area of the Dafeng Elk Nature Reserve in Jiangsu Province, China as the study area. A normalized difference soil spectral index (NDSOC), which is directly related to SOC, was constructed by using the spectral reflectance of Landsat images, and six auxiliary environmental variables related to SOC were extracted: surface temperature, surface moisture, soil salinity, the normalized difference vegetation index (NDVI), surface brightness, and soil texture information. The coastal wetland soil surface SOC inversion model was constructed by multiple linear regression, support vector machine, and particle swarm optimization-based random forest regression (PSO-RFR). By using the best inversion model, we obtain the dynamic distribution of soil surface SOC in the third core area of the Dafeng Elk Reserve for the past 23 years. The results indicate (1) that the NDSOC index constructed based on the short-wave infrared (SWIR) bands 1 and 2 of the Landsat images has a strong indication of SOC. Among the extracted auxiliary environmental variables, the NDVI, soil clay content, surface temperature, and soil salinity contribute more strongly to the SOC content of coastal wetlands, which averaged 27.9 %, 24.4 %, 22.1 %, and 18 %, respectively. (2) The PSO-RFR algorithm provided the best simulation of the surface SOC content of coastal wetlands: the model determination coefficient R 2 is 0.731, the average fitting root mean square error RMSE is 0.917 g/kg, and the fitting uncertainty standard deviation was 0.362 g/kg. (3) From 2000 to 2022, the organic carbon storage and organic carbon density in the third core area of the Dafeng Elk Reserve gradually increased, the highest value was reached in 2016, they were 4.017 (104 t) and 1.718 kg/cm2, respectively. [ABSTRACT FROM AUTHOR]

Actual evapotranspiration (ET a) is an important component of the global energy exchange process and hydrological cycle. ET a data with high accuracy are of great significance to the understanding of hydrological processes and the regulation of local climate. In this paper, an ET a product with hourly temporal resolution for the entire Tibetan Plateau (TP) was proposed by using Fengyun-4A geostationary satellite data and a random forest (RF) model. Validation between RF-based ET a and in situ observations from six eddy-covariance stations over the TP showed a root mean square error (RMSE) of 0.991 mm day−1. Compared with the ET a results from the surface energy balance system (SEBS), maximum entropy production (MEP) and the European Centre for Medium-Range Weather Forecasts Reanalysis Fifth Generation (ERA5) with RMSE values of 1.997 mm day−1, 2.182 mm day−1 and 1.555 mm day−1, respectively, the RF model showed the best performance. The annual RF-based ET a over the entire TP was approximately 365.60 mm. Thus, the total water amount evapotranspired from the TP surface was approximately 9811.01×108 t yr−1. Additionally, diurnal, monthly and seasonal ET a variations in different land cover types and climate zones over the TP and their contributions to the total ET a were clearly quantified. • Hourly actual ET over the entire Tibetan Plateau (TP) was firstly obtained by the random forest model (RF). • The RF ET a showed the best performance in estimation of hourly ET a data, including RF, SEBS, MEP and ERA5. • ETa variations of different land cover types and climate zones over the TP were clearly identified. [ABSTRACT FROM AUTHOR]

With the gradual declining streamflow gauging stations in many world-rivers, emphasis is given nowadays to develop remote sensing (RS)-based approaches as the next-generation hydrometry for estimating riverine ecological flow regimes (EFR). For constructing EFR based on daily-streamflow data in scantily-gauged reaches, use of RS techniques in narrow flow-width tropical rain-fed rivers is constrained with the non-availability of finer spatial satellite data at daily scale. To address these limitations, this study proposes a novel framework that integrates the enhanced spatiotemporal adaptive reflectance fusion (FUS) of the 250 m × 1-day resolution Aqua-MODIS and 30 m × 1-day resolution Landsat satellite-based remote sensing images in the near-infrared region with the machine learning algorithms. These developed frameworks are named as Artificial Neural Network-based ANNFUS, Random Forest Regression-based RFRFUS, and Support Vector Regression-based SVRFUS models, which were tested for daily-scale streamflow estimation in a typical Brahmani River Basin, India. The results reveal that by addressing the linear and nonlinear dynamism between the streamflow and satellite signals, all the developed models could simulate the streamflow very well with the Nash-Sutcliffe efficiency>0.8, Kling-Gupta efficiency>0.8, relative root mean square error (rRMSE) of 0.051–0.12, and normalized RMSE of 0.23–0.36. However, for reproducing the high, median, and low streamflow regimes, the SVRFUS model was found to be the best with the NSE>0.85 and KGE>0.8. Conclusively, the proposed approach is found to have the potential to be replicated in other world-river basins to estimate ecological flow regimes at defunct gauging stations facilitating the basin-scale aquatic environmental management. [Display omitted] • To estimate daily streamflow, advocated remote sensing-based machine learning tools. • The models are ANN, Random Forest Regression and Support Vector Regression. • For river/near-river environments, MODIS and Landsat pixels in NIR-band are fused. • Tested the proposed techniques in a narrow-width river constrained with mixels. • For streamflow regimes, Support Vector Regression models are the best. [ABSTRACT FROM AUTHOR]

As the main solid waste produced by coal combustion, the large accumulation of coal fly ash (CFA) causes serious environmental pollution and resource waste. Whether CFA can be recycled depends on its reactivity, which in turns can be represented by its amorphous content. This paper established random forest regression models optimized by artificial bee colony (ABC) for rapid screening of CFA, based on the correlation between chemical composition and amorphous phase. The study evaluated the model using correlation coefficient, r-square, root mean square error, and mean absolute error, giving results of testing set of 0.773, 0.477, 6.542, and 5.279. Feature importance and permutation importance were used to measure feature contribution. Partial dependence plots, Shapley additive explanations, and local interpretable model-agnostic explanations were also used to give global and local interpretation of the model performance. The overall results showed that Al 2 O 3 , MgO and CaO had the greatest influence on the amorphous content within CFA. When the mass fractions of Al 2 O 3 , MgO and CaO varied from 17.5% to 22.5%, 2%–4% and 10–15%, the mass fraction of amorphous phase reached the highest. For different CFA samples, chemical composition played a different role in determining the amorphous content. The post-analysis of the model provided some reference for promoting CFA recycling. The above results proved that the established model had good robustness and generalization capability, which can effectively determine the potential of CFA as supplementary cementitious materials to promote the cleaner production of cement or geopolymer resources. [Display omitted] • A hybrid machine learning model (RF-ABC) was established. • The correlation between chemical composition and amorphous phase was analyzed. • This model framework enables rapid screening of high-reactive CFA. • Multiple post-analysis of the model provides a reference for recycling of CFA. • The recycling of CFA can achieve cleaner production of building materials. [ABSTRACT FROM AUTHOR]

• Predicting retention times for multiple chromatographic conditions. • Predicting retention times for four retention mechanisms. • Predictions are projectable to different chromatographic gradients. Structural elucidation of compounds detected with liquid chromatography coupled to high resolution mass spectrometry is a challenging and time-consuming step in the workflow of non-targeted analysis and often requires manual validation of the results. Retention time, alongside exact mass, isotope pattern, fragmentation spectra, and collision cross-section, is valuable information for ruling out unlikely structures and increasing the confidence in others. Different approaches to predict retention times have been used previously for reversed phase chromatography and hydrophilic interaction liquid chromatography (HILIC), but application is limited to a small set of mobile phases and gradient profiles. Here, we expand the toolbox available for retention time predictions by developing a random forest regression model for predicting retention times for four column types and twenty mobile phase systems. MultiConditionRT was built using a dataset containing 78 compounds analyzed with C18 reversed phase, mixed mode, HILIC, and biphenyl columns. In addition, different eluent compositions were used: both methanol and acetonitrile were combined with different aqueous phases with pH from 2.1 to 10.0 (formic acid, acetic acid, trifluoroacetic acid, formate, acetate, bicarbonate, and ammonia). The root mean square error (RMSE) of the test set predictions was 1.55 min for C18 reversed phase, 1.79 min for mixed-mode, 1.93 min for HILIC, and 1.56 min for biphenyl column. Additionally, MultiConditionRT can be applied to different gradient profiles with a general additive model-based calibration approach. The approach of MultiConditionRT was validated externally and internally with 356 and 151 compounds respectively, yielding an RMSE of 2.68 and 2.32 min. 324 and 84 of these compounds were not in the dataset used in the model development. [ABSTRACT FROM AUTHOR]