Normalized Root Mean Square Error Research Articles

A meta-learning method for smart manufacturing: Tool wear prediction using hybrid information under various operating conditions

Accurate tool wear prediction during machining is crucial to manufacturing since it will significantly influence tool life, machining efficiency, and workpiece quality. Although existing data-driven methods can achieve competitive performance in tool wear prediction, their main emphasis is on fixed operating conditions with sufficient training samples, which is impractical in engineering practice. This implies that predicting tool wear values under variable working conditions with insufficient data is still a challenge owing to the difference in data distributions in complex tool wear mechanisms. Besides, having no access to samples in new conditions is another challenge for tool wear prediction in engineering practice. To address these issues, we develop a hybrid information model-agnostic domain generalization (H-MADG) method to provide appropriate initial model parameters that can be fast adaptative to the new conditions after fine-tuning. Additionally, we construct hybrid information as model input by fusing process information with temporal properties derived by neural networks, and the hybrid information can offer more useful prior knowledge about the machining process. Experimental results on NASA milling data show that compared with contrastive techniques, the RMSE of the proposed H-MADG method is reduced by an average of 36.81 %, which can achieve a low average RMSE value of 0.0904 with 15 cases under five different network architectures. We also investigate several crucial impact factors of the H-MADG method and summarize corresponding analysis and suggestions.

Prediction of Soil Field Capacity and Permanent Wilting Point Using Accessible Parameters by Machine Learning

The field capacity (FC) and permanent wilting point (PWP) are fundamental hydrological properties critical for assessing water availability within soils, rather than direct measures of soil health. Due to the challenges associated with their field measurement, alternative assessment methods are necessary. In this study, global-scale accessible soil data were retrieved from the world soil database called the World Soil Information Service (WoSIS), and artificial neural network (ANN) and gene-expression programming (GEP) algorithms were used to predict soil FC and PWP based on easily obtainable parameters from the database. The best-fit variable combination for FC (longitude, latitude, altitude, sand content, silt content, clay content, and electrical conductivity) and PWP (best-fit FC combination plus pH) modeling was determined. Both ANN and GEP showed greater accuracy than linear-based models in simulating the FC and PWP from the best-fit variables. The mean absolute error (MAE) was reduced by 51.54% for the FC and 56.38% for the PWP by the ANN model, compared with the linear model used in the previous literature. The normalized root mean square error (NRMSE) evaluation indicated that the ANN model performed best for PWP prediction (NRMSE of 19.9%), while the GEP model was superior for FC prediction (NRMSE of 29.9%). Between the ANN and GEP models, the ANN model showed a slightly higher model of interpretability; however, the GEP model exhibited a similar or better ability to avoid large error, based on the error distribution. Overall, our results demonstrated that machine learning is effective in predicting the FC and PWP from easily accessible data from WoSIS, and the GEP model is more preferable for FC and PWP modeling.

A leaf age-dependent light use efficiency model for remote sensing the gross primary productivity seasonality over pantropical evergreen broadleaved forests.

Tropical and subtropical evergreen broadleaved forests (TEFs) contribute more than one-third of terrestrial gross primary productivity (GPP). However, the continental-scale leaf phenology-photosynthesis nexus over TEFs is still poorly understood to date. This knowledge gap hinders most light use efficiency (LUE) models from accurately simulating the GPP seasonality in TEFs. Leaf age is the crucial plant trait to link the dynamics of leaf phenology with GPP seasonality. Thus, here we incorporated the seasonal leaf area index of different leaf age cohorts into a widely used LUE model (i.e., EC-LUE) and proposed a novel leaf age-dependent LUE model (denoted as LA-LUE model). At the site level, the LA-LUE model (average R2 = .59, average root-mean-square error [RMSE] = 1.23 gC m-2 day-1) performs better than the EC-LUE model in simulating the GPP seasonality across the nine TEFs sites (average R2 = .18; average RMSE = 1.87 gC m-2 day-1). At the continental scale, the monthly GPP estimates from the LA-LUE model are consistent with FLUXCOM GPP data (R2 = .80; average RMSE = 1.74 gC m-2 day-1), and satellite-based GPP data retrieved from the global Orbiting Carbon Observatory-2 (OCO-2) based solar-induced chlorophyll fluorescence (SIF) product (GOSIF) (R2 = .64; average RMSE = 1.90 gC m-2 day-1) and the reconstructed TROPOspheric Monitoring Instrument SIF dataset using machine learning algorithms (RTSIF) (R2 = .78; average RMSE = 1.88 gC m-2 day-1). Typically, the estimated monthly GPP not only successfully represents the unimodal GPP seasonality near the Tropics of Cancer and Capricorn, but also captures well the bimodal GPP seasonality near the Equator. Overall, this study for the first time integrates the leaf age information into the satellite-based LUE model and provides a feasible implementation for mapping the continental-scale GPP seasonality over the entire TEFs.

Koreksi Bias Data Hujan Satelit Persiann Sebagai Alternatif Pos Stasiun Hujan dengan Pendekatan Quantile Mapping di Sub Das Brantas Hulu

The rainfall data is usually obtained from meteorological stations, which many sill observed manually in Indonesia and can lead to an observation error. Remote sensing rainfall data observed by satellite offer interesting alternatives to overcome the limitation of existing meteorological stations in Indonesia. It offers accuracy, spatial coverage, timeliness and cost effectiveness. However it contains bias if compared to land based station. This research will evaluate rainfall data observed in PERSIANN (Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks) in Brantas Hulu Watershed. The rainfall data of 13 years from 12 stations is evaluated to obtain area averaged rainfall using Thiessen Polygon method. Three datasets from PERSIANN, PERSIANN CCS and PERSIANN CDR are compared statistically with the area averaged area for daily, monthly and yearly rainfall. Bias correction is performed using Quantile Mapping to match the cumulative distribution curve from the area averaged rainfall. Verification of bias correction is performed using 2 years data from 2019 to 2020. Finally, the extreme rainfall analysis is performed to test the corrected data for design rainfall calculations. The correlation of daily data shows poor agreement for all three datasets with correlation coefficient below 0,40. The monthly and yearly data gives better correlation with coefficient above 0,80. It shows that for time series event, the satellite data gives poor correlation with the area averaged rainfall. However, quantile mapping correction gives consistent cumulative distribution agreement for correction and model verification. The verification analysis shows an increase in coefficient correlation from 0.387 to 0.43 (daily) and 0.887 to 0.915 (monthly). In addition, there is a decrease in normalized root mean square error (NRMSE) from 31.03% to 16.91% (daily) and 46.03% to 14.03% (monthly). A decrease in normalized mean absolute error (NMAE) occurred from 16.12% to 9.74% (daily) and 31.67% to 10.07% (monthly). Beside that, the results of extreme rainfall analysis produce values that are very close to the rain station. So it is found that the PERSIANN CCS quantile mapping regression model is the best in improving data quality.

Thermodynamic modeling of anticancer drugs solubilities in supercritical CO2 using the PC-SAFT equation of state

In this research, the solubility of anticancer drugs in the supercritical CO2 is modeled using the PC-SAFT equation of state (EoS). Ten anticancer drugs containing Empagliflozin, Sorafenib tosylate, Verapamil, Sodium valproate, Aprepitant, Sunitinib malate, Tamsulosin, Imatinib mesylate, Capecitabine, and Docetaxel are studied to evaluate the model performance. For each component, three temperature-independent model parameters are optimized by the experimental solubility data. The CO2 is modeled as associative molecules with four associating sites and four association sites are considered on anticancer molecules. Therefore, the cross-association between anticancer and CO2 molecules is considered. The average ARD, RMSE, and AAD values of the PC-SAFT EoS for the aforementioned anticancer drugs are obtained 11.45 %, 0.067, and 0.00026 %, respectively. The PC-SAFT EoS results are compared to various types of cubic EoSs and semi-empirical models. The results show that the PC-SAFT EoS with three model parameters can be used as a robust and efficient thermodynamic model to calculate the solubility of complex molecules in supercritical CO2 up to high pressures and temperatures.

A gap filling method for daily evapotranspiration of global flux data sets based on deep learning

In response to irregular data gaps in evapotranspiration (ET) data obtained using eddy covariance (EC) methods, this study seeks to explore a high-precision interpolation method that is not constrained by spatiotemporal limitations. Leveraging daily observations from 241 global flux towers (FLUXNET ET) as a reference, data gaps are artificially introduced at five different proportions: 5 %, 10 %, 20 %, 35 %, and 50 %, this process is repeated 30 times (Take the average of the results). Four ET products, along with environmental factors (ERA5 meteorological data and NOAA Normalized Difference Vegetation Index), are employed as input to assess the performance of the target model, Convolutional Neural Network (CNN)-Long Short-Term Memory (LSTM), as well as five comparative models: CNN, LSTM, Random Forest (RF), Support Vector Machine (SVM), and Extreme Learning Machine (ELM), in filling the data gaps. These results are then compared with the filling outcomes obtained without ET products as input, aiming to investigate whether ET products contributes to improving the accuracy and stability of ET gap filling. The results indicate that, firstly, the filling performance of the target model CNN-LSTM surpasses that of other models, with minimal impact from gap proportions. Moreover, incorporating ET products as input alongside environmental factors significantly enhances the filling accuracy and stability of the model. Compared to scenarios without ET products as input, the CNN-LSTM model shows an average increase in R2 of 5 % to 15 % and a decrease in root mean square error (RMSE) of 15 % to 35 %. Specifically, as the gap proportion increases from 5 % to 50 %, the average Pearson correlation coefficient (R2) decreases from 0.952 to 0.846 (a decline of 11 %), the average RMSE increases from 0.491 mm/d to 0.620 mm/d (a rise of 26 %), and reached the filling limit at the gap ratio of 93 % (R2 below 0.4). Conversely, without ET products as input, as the gap proportion increases, the average R2 decreases from 0.916 to 0.784 (a decline of 14 %), the RMSE increases from 0.552 mm/d to 0.952 mm/d (a rise of 72 %), and reached the filling limit at the gap ratio of 80 %. Secondly, the contribution of ET products to the model estimation far exceeds that of ERA5 meteorological factors and NDVI, with the contribution of ET products increasing with higher gap proportions from 58.0 % to 73.6 %. Finally, different latitudes, land cover types, and climate types influence the model filling performance, with better performance observed in high-latitude regions, Grassland (GRA), and Dry-Cold climates. In summary, incorporating ET products as input for CNN-LSTM yields high-precision gap-filling results, surpassing existing typical methods for gap filling and its application is not constrained by spatiotemporal limitations.

Joint estimation of State of Charge (SOC) and State of Health (SOH) for lithium ion batteries using Support Vector Machine (SVM), Convolutional Neural Network (CNN) and Long Sort Term Memory Network (LSTM) models

This paper proposes a data-driven method for jointly estimating the State of Charge (SOC) and State of Health (SOH) of batteries, addressing the impact of battery aging on SOC estimation. Initially, Support Vector Machine (SVM) is employed to estimate the SOH of the battery, using the constant voltage charging time and constant voltage discharging time of lithium-ion batteries as inputs, and SOH as the output. By training the SVM model, accurate SOH estimation is achieved. Subsequently, the rated capacity of the battery is adjusted based on the estimated SOH to obtain the current maximum available capacity. This adjustment allows for the coupling of SOH and SOC, resulting in SOC estimation that accounts for aging factors. Leveraging the advantages of Convolutional Neural Networks (CNN) in feature extraction and Long Short-Term Memory (LSTM) neural networks in handling long-term sequential data, a CNN-LSTM model is utilized for SOC estimation. The proposed method utilizes the Oxford Battery Dataset (Cells 1–8) and the NASA Battery Dataset (B0005–B0007) for SOH estimation, and the Oxford Battery Dataset (Cell 8) and the NASA Battery Dataset (B0007) for SOC estimation. The SOH estimation results demonstrate that the Root Mean Square Error (RMSE) is less than 0.81 % and the Mean Absolute Error (MAE) is less than 0.65 % for Cells 1–8, while for B0005 to B0007, the RMSE is less than 1.81 % and the MAE is less than 1.29 %. For SOC estimation, the results show that the average RMSE over the entire lifecycle of Cell 8 is 0.3923 % and the average MAE is 0.3339 %, whereas for B0007, the average RMSE over the entire lifecycle is 0.6123 % and the average MAE is 0.4976 %.

Validation of IMU against optical reference and development of open-source pipeline: proof of concept case report in a participant with transfemoral amputation fitted with a Percutaneous Osseointegrated Implant

BackgroundSystems that capture motion under laboratory conditions limit validity in real-world environments. Mobile motion capture solutions such as Inertial Measurement Units (IMUs) can progress our understanding of "real" human movement. IMU data must be validated in each application to interpret with clinical applicability; this is particularly true for diverse populations. Our IMU analysis method builds on the OpenSim IMU Inverse Kinematics toolkit integrating the Versatile Quaternion-based Filter and incorporates realistic constraints to the underlying biomechanical model. We validate our processing method against the reference standard optical motion capture in a case report with participants with transfemoral amputation fitted with a Percutaneous Osseointegrated Implant (POI) and without amputation walking over level ground. We hypothesis that by using this novel pipeline, we can validate IMU motion capture data, to a clinically acceptable degree.ResultsAverage RMSE (across all joints) between the two systems from the participant with a unilateral transfemoral amputation (TFA) on the amputated and the intact sides were 2.35° (IQR = 1.45°) and 3.59° (IQR = 2.00°) respectively. Equivalent results in the non-amputated participant were 2.26° (IQR = 1.08°). Joint level average RMSE between the two systems from the TFA ranged from 1.66° to 3.82° and from 1.21° to 5.46° in the non-amputated participant. In plane average RMSE between the two systems from the TFA ranged from 2.17° (coronal) to 3.91° (sagittal) and from 1.96° (transverse) to 2.32° (sagittal) in the non-amputated participant. Coefficients of Multiple Correlation (CMC) results between the two systems in the TFA ranged from 0.74 to > 0.99 and from 0.72 to > 0.99 in the non-amputated participant and resulted in ‘excellent’ similarity in each data set average, in every plane and at all joint levels. Normalized RMSE between the two systems from the TFA ranged from 3.40% (knee level) to 54.54% (pelvis level) and from 2.18% to 36.01% in the non-amputated participant.ConclusionsWe offer a modular processing pipeline that enables the addition of extra layers, facilitates changes to the underlying biomechanical model, and can accept raw IMU data from any vendor. We successfully validate the pipeline using data, for the first time, from a TFA participant using a POI and have proved our hypothesis.

Reducing data requirement for accurate photovoltaic power prediction using hybrid machine learning-physical model on diverse dataset

This research aims to develop a hybrid model combining machine learning and a physical model and demonstrate its effectiveness across diverse datasets, data scarcity scenarios, and not standard conditions. The hybridization approach involves parallel integration of physical and machine learning models, where the outputs of both models are weighted and yield improved predictions by assigning weight based on error minimization. A deep neural network serves as the machine learning model, while the single-diode model is employed as the physical model. This study utilizes six different solar PV for constructing the diverse dataset. Key findings reveal that the hybrid model can accurately predict power generation from a diverse range of solar PV, during data scarcity scenarios, and in all conditions. The model performs well in predicting solar PV power output in 6 types of solar PV with an average RMSE of 0.177 kW. The hybrid model outperforms machine learning and physical models by 1.5 % and 42 % respectively when trained on the full dataset. Moreover, when 97 % of the data is discarded intentionally, the hybrid model outperforms both machine learning and physical models by 47.7 % and 10.5 % respectively. The model also shows that it can improve the accuracy of the prediction during standard (solar irradiance = 1000 W m−2) and not standard conditions. The successful hybridization of machine learning and physical models addresses previously identified limitations in individual models. In addition, the model indicates a substantial improvement in terms of accuracy compared to prior published studies.

Dynamic-based artificial intelligence model for simulation and optimization of the single chamber anode brush microbial electrolysis cell

Microbial electrolysis cells (MECs) have become more attractive because they can produce hydrogen (H2) from various organic sources with a low input electrical energy requirement. In this study, a dynamic model for the single-chamber anode brush MEC with acetic acid as a carbon source was developed for the first time. The dynamic model was calibrated and validated using both experimental and literature data, with a high coefficient of determination (R2) of 0.87–0.99, a high Pearson correlation coefficient (PCC) of 0.93–0.99, and a low normalized root mean squared error (NRMSE) of 0.221–0.078. For rapid simulation and optimization, the validated dynamic model was reproduced using an artificial neural network (ANN), and particle swarm optimization (PSO) was then applied to optimize the fitness functions derived from this ANN model (i.e., hydrogen production yield, hydrogen production rate, hydrogen fraction in biogas, and total energy recovery). As a result, the proposed method took about 3 s with an absolute error of <3.5 %. The optimization results revealed that an optimal condition (including applied voltage, operating temperature, substrate concentration, and reactor working volume) cannot exist to obtain the maximum values of all performance parameters simultaneously. Under the same importance, the co-optimal condition was determined at an applied voltage of 0.9 V, temperature of 36.2 °C, substrate concentration of 0.87 g/L, and reactor working volume of 0.05 L. At such co-optimal conditions, MEC achieved an H2 yield of 1168.6 ± 13.3 mL/g with a production rate of 185.6 ± 2.7 mL/L/h, an H2 fraction of 63.6 ± 0.2 %, and a total energy recovery (including electrical, substrate, heat, and mixing input energies) of 16.7 ± 0.2 %. Although these performances increased 1.5–2.3 times compared to non-optimal conditions, the low energy efficiency indicates the need for further investigation into combining MEC with sustainable energy sources, such as solar, wind and osmosis.

A multi-model ensemble approach for reservoir dissolved oxygen forecasting based on feature screening and machine learning

Dissolved oxygen (DO) concentration in aquatic systems plays a vital role in water aquaculture. An innovative approach that combines feature selection and ensemble learning to predict DO in aquatic ecosystems was proposed. Feature selection was first performed using Maximum Information Coefficient (MIC). Five machine learning algorithms were then employed to construct five hybrid-MIC models, including K-Nearest Neighbors (KNN), Backpropagation (BP) Neural Network, Long Short-Term Memory (LSTM), Kernel Ridge Regression (KRR), and Support Vector Regression (SVR). Finally, an ensemble-RF prediction model was built using Random Forests(RF). The main findings are as follows: (1) The MIC technique can effectively identify the key factors influencing DO. (2) The MIC significantly improves model performance. (3) The hybrid-MIC model was further improved by the ensemble-RF model, the average R2 and NSE were both as high as 0.99, and the average MAE and RMSE were decreased by 72 % and 64 %, respectively.

Exogenous variable driven deep learning models for improved price forecasting of TOP crops in India

Accurately predicting agricultural commodity prices is crucial for India's economy. Traditional parametric models struggle with stringent assumptions, while machine learning (ML) approaches, though data-driven, lack automatic feature extraction. Deep learning (DL) models, with advanced feature extraction and predictive abilities, offer a promising solution. However, their application to agricultural price data ignored the exogenous factors. Hence, the study explored advanced versions of the well-known univariate models, NBEATSX and TransformerX. The research employed price data for essential crops like Tomato, Onion, and Potato (TOP) from major Indian markets and complemented it with corresponding weather data (precipitation and temperature). To provide a comprehensive analysis, the study also evaluated traditional statistical methods (ARIMAX and MLR) and a suite of ML algorithms (ANN, SVR, RFR, and XGBoost). The performance of these models was rigorously evaluated using error metrics like RMSE, MAE, sMAPE, MASE and QL. The findings were significant indicating DL models, particularly when augmented with exogenous variables, consistently outshone other methods with NBEATSX and TransformerX showing an average RMSE of 110.33 and 135.33, MAE of 60.08 and 74.92, sMAPE of 22.14 and 24.00, MASE of 1.02 and 1.32 and QL of 30.04 and 34.07, respectively. They exhibited lower error metrics, as compare to the statistical and ML models underscoring their effectiveness and potential in agricultural crop price forecasting. This study not only bridged a crucial research gap but also highlighted the robust potential of DL models in enhancing the accuracy of agricultural commodity price predictions in India.

Smartphone videos-driven musculoskeletal multibody dynamics modelling workflow to estimate the lower limb joint contact forces and ground reaction forces.

The estimation of joint contact forces in musculoskeletal multibody dynamics models typically requires the use of expensive and time-consuming technologies, such as reflective marker-based motion capture (Mocap) system. In this study, we aim to propose a more accessible and cost-effective solution that utilizes the dual smartphone videos (SPV)-driven musculoskeletal multibody dynamics modeling workflow to estimate the lower limb mechanics. Twelve participants were recruited to collect marker trajectory data, force plate data, and motion videos during walking and running. The smartphone videos were initially analyzed using the OpenCap platform to identify key joint points and anatomical markers. The markers were used as inputs for the musculoskeletal multibody dynamics model to calculate the lower limb joint kinematics, joint contact forces, and ground reaction forces, which were then evaluated by the Mocap-based workflow. The root mean square error (RMSE), mean absolute deviation (MAD), and Pearson correlation coefficient (ρ) were adopted to evaluate the results. Excellent or strong Pearson correlations were observed in most lower limb joint angles (ρ = 0.74 ~ 0.94). The averaged MADs and RMSEs for the joint angles were 1.93 ~ 6.56° and 2.14 ~ 7.08°, respectively. Excellent or strong Pearson correlations were observed in most lower limb joint contact forces and ground reaction forces (ρ = 0.78 ~ 0.92). The averaged MADs and RMSEs for the joint lower limb joint contact forces were 0.18 ~ 1.07 bodyweight (BW) and 0.28 ~ 1.32 BW, respectively. Overall, the proposed smartphone video-driven musculoskeletal multibody dynamics simulation workflow demonstrated reliable accuracy in predicting lower limb mechanics and ground reaction forces, which has the potential to expedite gait dynamics analysis in a clinical setting.

Recurrent neural network using LSTM for prediction of atmospheric pollutants in the State of Veracruz, Mexico

This article shows the use of a recurrent neural network LSTM whose objective was to make a 24-hour ozone forecast with data obtained from the INECC (National Institute of Ecology and Climate Change) and the validated database of the National Information System of Air Quality (SINAICA spanish acronym). The problem of air quality in the world is essential due to health problems. Currently, the techniques used to generate forecasts in monitoring stations are not accurate and do not allow generating alerts to avoid exposure to poor air quality. The state of Veracruz in Mexico has 7 air quality monitoring stations, showing that it is feasible to alternatively have a betterquality pollutant concentration forecasting system. The results obtained on the correlations of variables, although weak, positive or negative, allowed us to recognize their usefulness in the development of training in the RNN. The use of neural networks was demonstrated for the forecast of ozone concentration values in the metropolitan area of Poza Rica, using the LSTM using Python, Tensor Flow and Keras, with a good fit to the normalized pattern with an average RMSE of 0.0053 ppm of the days chosen for testing between the period from June 2019 to July 2020. Future work will consider PM10 and PM2.5.

Evaluation and development of pedotransfer functions of saturated hydraulic conductivity for subtropical soils

The determination of soil saturated hydraulic conductivity (Ks) is crucial in environmental and engineering fields. However, the current direct measurement methods of Ks are time-consuming and labor-intensive. As an alternative method, many researchers have developed a series of pedotransfer functions (PTFs) that estimate Ks based on easily accessible soil properties. Unfortunately, most existing PTFs of Ks focus on temperate and tropical regions in the United States and Europe. There is a lack of research discussing the applicability of Ks models in subtropical areas. To resolve this issue, we conduct a study using 515 subtropical soil samples to test the performance of existing PTFs of Ks. The Ks values of investigated soils range from 1.4 × 10-4 to 290 cm h−1. Among the affecting factors, soil bulk density (ρb) and effective porosity (φe) are found to be the most important variables. Current PTFs considering soil pore structural property (i.e., φe) or soil texture solely are not effective in assessing Ks of subtropical soils. To address this limitation, we integrate soil pore structural and textural properties and develop a new PTF based on the Kozeny-Carman equation. The performance and reliability of the new PTF are evaluated using independent Ks datasets from various regions. The results show a significant positive correlation between the measured Ks and the predicted Ks from the new PTF for subtropical soils with an R2 of 0.69, an average RMSE in log10Ks (cm h−1) of 0.496, and a mean bias value of 0.070. Besides, the new PTF is found to perform as well as widely used machine learning tools.

Influence of Grid Resolution and Assimilation Window Size on Simulating Storm Surge Levels

Grid resolution and assimilation window size play significant roles in storm surge models. In the Bohai Sea, Yellow Sea, and East China Sea, the influence of grid resolution and assimilation window size on simulating storm surge levels was investigated during Typhoon 7203. In order to employ a more realistic wind stress drag coefficient that varies with time and space, we corrected the storm surge model using the spatial distribution of the wind stress drag coefficient, which was inverted using the data assimilation method based on the linear expression Cd = (a + b × U10) × 10−3. Initially, two grid resolutions of 5′ × 5′ and 10′ × 10′ were applied to the numerical storm surge model and adjoint assimilation model. It was found that the influence of different grid resolutions on the numerical model is almost negligible. But in the adjoint assimilation model, the root mean square (RMS) errors between the simulated and observed storm surge levels under 5′ × 5′ and 10′ × 10′ grid resolutions were 11.6 cm and 15.6 cm, and the average PCC and WSS values for 10 tidal stations changed from 89% and 92% in E3 to 93% and 96% in E4, respectively. The results indicate that the finer grid resolution can yield a closer consistency between the simulation and observations. Subsequently, the effects of assimilation window sizes of 6 h, 3 h, 2 h, and 1 h on simulated storm surge levels were evaluated in an adjoint assimilation model with a 5′ × 5′ grid resolution. The results show that the average RMS errors were 11.6 cm, 10.6 cm, 9.6 cm, and 9.3 cm under four assimilation window sizes. In particular, the RMS errors for the assimilation window sizes of 1 h and 6 h at RuShan station were 3.9 cm and 10.2 cm, a reduction of 61.76%. The PCC and WSS values from RuShan station in E4 and E7 separately showed significant increases, from 85% to 98% and from 92% to 99%. These results demonstrate that when the assimilation window size is smaller, the simulated storm surge level is closer to the observation. Further, the results show that the simulated storm surge levels are closer to the observation when using the wind stress drag coefficient with a finer grid resolution and smaller temporal resolution.

Digital Mapping and Scenario Prediction of Soil Salinity in Coastal Lands Based on Multi-Source Data Combined with Machine Learning Algorithms

Salinization is a major soil degradation process threatening ecosystems and posing a great challenge to sustainable agriculture and food security worldwide. This study aimed to evaluate the potential of state-of-the-art machine learning algorithms in soil salinity (EC1:5) mapping. Further, we predicted the distribution patterns of soil salinity under different future scenarios in the Yellow River Delta. A geodatabase comprising 201 soil samples and 19 conditioning factors (containing data based on remote sensing images such as Landsat, SPOT/VEGETATION PROBA-V, SRTMDEMUTM, Sentinel-1, and Sentinel-2) was used to compare the predictive performance of empirical bayesian kriging regression, random forest, and CatBoost models. The CatBoost model exhibited the highest performance with both training and testing datasets, with an average MAE of 1.86, an average RMSE of 3.11, and an average R2 of 0.59 in the testing datasets. Among explanatory factors, soil Na was the most important for predicting EC1:5, followed by the normalized difference vegetation index and soil organic carbon. Soil EC1:5 predictions suggested that the Yellow River Delta region faces severe salinization, particularly in coastal zones. Among three scenarios with increases in soil organic carbon content (1, 2, and 3 g/kg), the 2 g/kg scenario resulted in the best improvement effect on saline–alkali soils with EC1:5 &gt; 2 ds/m. Our results provide valuable insights for policymakers to improve saline–alkali land quality and plan regional agricultural development.

Generating surface soil moisture at the 30 m resolution in grape-growing areas based on stacked ensemble learning

ABSTRACT Accurate and timely monitoring of drought conditions in grape-producing regions is crucial for achieving healthy growth of grapes. Current soil moisture (SM) products are primarily available at coarse resolutions (e.g. several to tens of kilometres), constraining its applications at fine scales. Here, we trained a weighted stacking ensemble model including three tree-based models (categorical boosting, random forest, and gradient boosting decision tree), using seven forcing parameters related to spectral reflectance (SR), land surface temperature (LST), and evapotranspiration (ET), in conjunction with the digital elevation model (DEM) feature. The weighted stacking ensemble model exhibited an average R2 of 0.86 and an average RMSE of 0.021 m3/m3 in simulating SM in the vegetive stage and the mid-ripening stage of grape. Then we generated high spatiotemporal downscaled SM (HSM) data at a grape growing area at high spatiotemporal resolutions (30 m, 8-day) from 2009 to 2018. Our HSM dataset demonstrated strong spatial, seasonal and interannual dynamics that align with 500 m SM dataset derived from single MODIS data, and the HSM dataset shows more details in SM distribution. Additionally, the SM time series in the HSM is consistently correlated with drought events, offering intricate spatiotemporal information for drought monitoring. The application of downscaled SM results identified a concentration of drought events in the eastern foothills of the Helan Mountains, particularly severe drought conditions were observed in the Hongsipu production area. Drought occurrences in the Hongsipu production area ranged from 90% to 91% during May and June, decreasing to 73% and 41% in July and August, respectively. These findings significantly contribute to enhancing high spatiotemporal SM monitoring capabilities, offering valuable guidance for timely water management in grape-growing regions.

Streamflow Prediction Using Complex Networks.

The reliable prediction of streamflow is crucial for various water resources, environmental, and ecosystem applications. The current study employs a complex networks-based approach for the prediction of streamflow. The approach consists of three major steps: (1) the formation of a network using streamflow time series; (2) the calculation of the clustering coefficient (CC) as a network measure; and (3) the use of a clustering coefficient-based nearest neighbor search procedure for streamflow prediction. For network construction, each timestep is considered as a node and the existence of link between any node pair is identified based on the difference (distance) between the streamflow values of the nodes. Different distance threshold values are used to identify the critical distance threshold to form the network. The complex networks-based approach is implemented for the prediction of daily streamflow at 142 stations in the contiguous United States. The prediction accuracy is quantified using three statistical measures: correlation coefficient (R), normalized root mean square error (NRMSE), and Nash-Sutcliffe efficiency (NSE). The influence of the number of neighbors on the prediction accuracy is also investigated. The results, obtained with the critical distance threshold, reveal that the clustering coefficients for the 142 stations range from 0.799 to 0.999. Overall, the prediction approach yields reasonably good results for all 142 stations, with R values ranging from 0.05 to 0.99, NRMSE values ranging from 0.1 to 12.3, and the NSE values ranging from -0.89 to 0.99. An attempt is also made to examine the relationship between prediction accuracy and the catchment characteristics/streamflow statistical properties (drainage area, mean flow, coefficient of variation of flow). The results suggest that the prediction accuracy does not have much of a relationship with the drainage area and the mean streamflow values, but with the coefficient of variation of flow. The outcomes from this study are certainly promising regarding the application of complex networks-based concepts for the prediction of streamflow (and other hydrologic) time series.

Spatial impact on inflation of Java Island prediction using Autoregressive Integrated Moving Average (ARIMA) and Generalized Space-Time ARIMA (GSTARIMA)

Inflation is one of macroeconomic issues in Indonesia that needs to be controlled. Inflation could happen because of widespread increases in the cost of goods and services. Annual inflation rate in Indonesia on 2008 to 2023 are quite fluctuating and several periods are not achieved inflation target yet. One of the ways to control inflation is by making predictions for the upcoming period. Java Island is the biggest contributor on economy and Gross Domestic Product (GDP) in Indonesia so it can be considered as general indicator to measure overall inflation rate of Indonesia. Thus, data used in this study is monthly inflation at each province in Java Island from January 2008 to December 2023. This study using two methods, Autoregressive Integrated Moving Average (ARIMA) for univariate time series prediction and Generalized Space-Time ARIMA (GSTARIMA) for multivariate time series prediction with a spatial factor. The results of both models will be compared to determine which model has better accuracy. Based on RMSE value, GSTARIMA model has least average RMSE value, which is 0.113 compared with ARIMA model which has average RMSE value 0.319 thus it can conclude that spatial factors addition could increase accuracy on inflation prediction in Java Island.•This paper purposes to get Java Island's inflation rate prediction to determine better policy on controlling cost of goods and services.•Best model using GSTARIMA methods is GSTARMA(1,1) with distance invese matrix that indicate that coordinate point of each location increase performance of inflation rate prediction.•The result indicate GSTARIMA has better accuracy than ARIMA for inflation prediction in Java Island based on RMSE value.