Relative Root Mean Square Error Research Articles

External Evaluation of Population Pharmacokinetic Models of Methotrexate for Model-Informed Precision Dosing in Pediatric Patients with Acute Lymphoid Leukemia.

Methotrexate (MTX) is a key immunosuppressant for children with acute lymphoid leukemia (ALL), and it has a narrow therapeutic window and relatively high pharmacokinetic variability. Several population pharmacokinetic (PopPK) models of MTX in ALL children have been reported, but the validity of these models for model-informed precision dosing in clinical practice is unclear. This study set out to evaluate the predictive performance of published pediatric PopPK models of MTX using an independent patient cohort. A PubMed literature search was performed to identify suitable models for evaluation. Demographics and measurements of the validation dataset were retrospectively collected from the medical records of ALL children who had received intravenous MTX. Predictive performance for each model was assessed by visual comparison of predictions to observations, median and mean predicted error (PE), and relative root mean squared error (RMSE). Six models were identified for external evaluation, carried out on a dataset containing 354 concentrations from 51 pediatrics. Model performance varied considerably from one model to another. Different models had the median PE for population and individual predictions at -33.23% to 442.04% and -25.20% to 6.52%, mean PE for population and individual predictions at -25.51% to 780.87% and 1.33% to 64.44%, and RMSE for population and individual predictions at 62.88% to 1182.24% and 63.39% to 152.25%. All models showed relatively high RMSE. Some of the published models showed reasonably low levels of bias but had some problems with imprecision, and extensive evaluation is needed before model application in clinical practice.

Construction of a computational MDCT model for simulations of the detector signals

ObjectiveTo develop a computational model of a multi-detector CT scanner (MDCT), which could be used to simulate the signal of each detector element in the MDCT by using the Monte Carlo method. MethodsThe CT scanner was modelled, including the X-ray source, the bowtie filter, the collimator, the couch and the detector panel. Under a general scanning condition, the signal in each detector element was simulated based on the model by using the MCNPX code. Both the energy spectra at different tube voltages and energy deposition in the detector panel at different collimations were simulated to test the robustness of the MDCT model built in this study. Furthermore, the simulated signals in each detector element were compared with their recorded signals. The accuracies were evaluated by the relative root mean square error (RRMSE) and the structural similarity (SSIM) for each detector element and the whole detector panel, respectively. ResultsThe simulated energy spectra before and after passing through the phantom and simulated energy deposition in the detector panel were rational. In the scan range from the apex of lung to pubic symphysis, the RRMSE of the 18 axial projections ranged from 0.02 to 0.17, with an average of 0.08. And the SSIMs were calculated to be 0.979 and 0.976 for projections with the largest peak signal and the smallest peak signal, respectively. ConclusionsThe computational model of the MDCT developed in this study is accurate and successful, it is helpful for further accurate simulations of the radiation dose and image quality of the MDCT.

Tillage practices influence winter wheat grain yield prediction using seasonal precipitation

Making the best use of limited precipitation in semi-arid dryland cropping systems is important for crop production. Tillage practices may influence how this precipitation is utilized to predict winter wheat grain yield (Triticum aestivum L.). This study examined how tillage practices influence winter wheat grain yield prediction accuracy using precipitation received at three different periods of the season. Data were obtained from the period of 1972 to 2010 from a long-term tillage experiment. The study was designed as a winter wheat-fallow experiment. Each phase of the winter wheat-fallow rotation was present each year. The trial was set up as a randomized complete block design with three replications. Tillage treatments included no-till (NT), stubble mulch (SM), and moldboard plow (MP). Feed-forward neural network and multiple linear regression (ordinary least squares) were used to fit models under each tillage practice. No-till had the highest yield prediction accuracy with a root mean square error (RMSE) of 0.53 Mg ha-1 and accounted for 81% of the variability in grain yield. Stubble mulch had an RMSE of 0.55 Mg ha-1 and explained 73% of the variability in yield. Stubble mulch and NT were more accurate in yield prediction than MP which had an RMSE of 0.77 Mg ha-1 and accounted for 53% of the variability in yield. The multiple linear regression model was less accurate than the feed-forward neural network model since it had at least 0.30 Mg ha-1 more RMSE and accounted for only 5-8% of the variability in yield. Relative RMSE classified all neural network models as fair (21.6-27.3%) while linear regression models for the different tillage practices was classified as poor (33.3-43.6%), an illustration that the neural network models improve yield prediction accuracy. This study demonstrated that a large proportion of the variability in grain yield may be accounted for under NT and SM systems when using precipitation as predictors with neural networks.

Natural Convection Heat Transfer from an Isothermal Plate

Using boundary-layer theory, natural convection heat transfer formulas that are accurate over a wide range of Rayleigh numbers (Ra) were developed in the 1970s and 1980s for vertical and downward-facing plates. A comprehensive formula for upward-facing plates remained unsolved because they do not form conventional boundary-layers. From the thermodynamic constraints on heat-engine efficiency, the novel approach presented here derives formulas for natural convection heat transfer from isothermal plates. The union of four peer-reviewed data-sets spanning 1<Ra<1012 has 5.4% root-mean-squared relative error (RMSRE) from the new upward-facing heat transfer formula. Applied to downward-facing plates, this novel approach outperforms the Schulenberg (1985) formula’s 4.6% RMSRE with 3.8% on four peer-reviewed data-sets spanning 106<Ra<1012. The introduction of the harmonic mean as the characteristic length metric for vertical and downward-facing plates extends those rectangular plate formulas to other convex shapes, achieving 3.8% RMSRE on vertical disk convection from Hassani and Hollands (1987) and 3.2% from Kobus and Wedekind (1995).

Investigation of Data-Driven Rating Curve (DDRC) Approach

Flooding is a recurring natural disaster worldwide; developing countries are particularly affected due to poor mitigation and management strategies. Often discharge is used to inform the flood forecast. The discharge is usually inferred from the water level via the rating curve because the latter is relatively easy to measure compared to the former. This research focuses on Cambodia, where data scarcity is prevalent, as in many developing countries. Thus, the rating curve has not been updated, making it difficult to effectively evaluate the performance of the global streamflow services, such as the Global Flood Awareness System (GloFAS) and Streamflow Prediction Tool (SPT), whose longer lead time can benefit the country in taking early action. In this study, we used time series of water level and discharge data to understand the changes in the flood plain to generate a data-derived rating curve for fifteen stations in Cambodia. We deployed several statistical and data-driven techniques to derive a generalized, scalable, and region-agnostic method. We further validated the process by applying it to ten stations in the US and found similar performance. In Cambodia, we obtained an average Kling Gupta Efficiency (KGE) of ∼99% & an average Relative Root Mean Squared Error (RRMSE) of 12% with an average Mean Absolute Error (MAE) of 200 m3/s. In the US, overall KGE was 97%, with an average RRMSE of 17% and an average MAE of 32 m3/s. The results indicated that the distribution of the dataset was key in deriving a good rating curve and that the stations with a low flow stations generally had higher errors than the high flow stations. The time series approach was shown to have more probability in capturing the high-end and low-end events compared to traditional method, where usually fewer data points are used. The study demonstrates that time series of data has valuable information to update the rating curve, especially in a data-scarce country.

In-Stream Laser Diffraction for Measuring Suspended Sediment Concentration and Particle Size Distribution in Rivers: Insights from Field Campaigns

This study evaluates the laser in situ scattering and transmissometry (LISST) instrument LISST-SL2, a laser diffraction instrument for suspended sediment sampling in rivers, with concurrent physical measurements of suspended sediment concentration (SSC) and particle size distribution (PSD) as well as velocity measurements by an acoustic Doppler current profiler (ADCP). We collected 136 LISST-SL2 samples along with 61 physical samples for SSC measurement, of which 24 physical samples included PSD measurement during 2018–2020 from 11 sites in Washington state and Virginia. An effective density is required to convert the measured volumetric SSC by the LISST-SL2 into a reported mass SSC, and by default the LISST-SL2 assumes a value of 2.65 g/mL. From our data set, we computed effective densities (mass SSC/volumetric SSC) that ranged from 0.5 to 5.4 g/mL, with a best-fit value of 2.05 g/mL. Additionally, the LISST-SL2 was not able to measure the finest sediment sizes in suspension, which affects the resulting PSD. Therefore, we propose some adjustments of the LISST-SL2 data with a supporting physical sample to account for these effective density and PSD issues. When doing so, we were able to reduce the root-mean square relative error (RMSRE) to 18% from 117% for SSC, and to 26% from 78% for PSD. LISST-SL2 velocities were generally higher than ADCP velocities with a 21% RMSRE. Our results and guidance will allow for more accurate sampling by the LISST-SL2, which has potential for studying spatial and temporal variation of suspended sediment characteristics in rivers.

Can We Accurately Quantify a Lateral Diffusivity from a Single Tracer Release?

Abstract Mixing along sloping isopycnals plays a key role in the transport and uptake of heat and carbon by the ocean. This mixing is quantified by a lateral diffusivity, which can be measured by tracking the lateral spreading of point release tracer patches. We present a definition for the area of a tracer patch, the time derivative of which provides the lateral diffusivity. To accurately estimate the diffusivity, an ensemble mean concentration field of many tracer release experiments is required. We use numerical experiments to quantify how accurately the “true” lateral diffusivity (obtained from the ensemble mean concentration field) can be estimated from a single tracer release experiment (one ensemble member). To simulate observational campaigns, we also estimate the diffusivity from a single tracer release that is spatially and/or temporally subsampled, quantifying how the error between the estimated diffusivity and the true diffusivity grows as this sampling resolution worsens. We perform these numerical experiments in a two-layer quasigeostrophic model of turbulent flow on a β plane, using an ensemble of 50 passive tracer release experiments, each initialized as a 2D Gaussian but with differing realizations of the turbulent flow. We find that the diffusivity estimates from the single tracer releases have a relative root-mean-square error (RMSE) of 1.43% from the true diffusivity. Subsampling a single tracer release experiment every 956 km increases the relative RMSE from the true diffusivity to 3.1%; also subsampling every 277 days raises this figure to 6.5%.

Simulating the soil phosphorus dynamics of four long‐term field experiments with a novel phosphorus model

AbstractPhosphorus is a nonrenewable resource, which is required for crop growth and to maintain high yields. The soil P cycle is very complex, and new model approaches can lead to a better understanding of those processes and further guide to research gaps. The objective of this study was to present a P‐submodel, which has been integrated in the existing Carbon Candy Balance (CCB) model that already comprises a C and N module. The P‐module is linked to the C mineralization and the associated C‐pools via the C/P ratio of fresh organic material. Besides the organic P cycling, the module implies a plant‐available P‐pool (Pav), which is in a dynamic equilibrium with the nonavailable P‐pool (Pna) that comprises the strongly sorbed and occluded P fraction. The model performance was tested and evaluated on four long‐term field experiments with mineral P fertilization, farmyard manure as organic fertilizer and control plots without fertilization. The C dynamics and the Pav dynamics were modelled with overall good results. The relative RMSE for the C was below 10% for all treatments, while the relative RMSE for Pav was below 15% for most treatments. To accommodate for the rather small variety of available P‐models, the presented CNP‐model is designed for agricultural field sites with a relatively low data input, namely air temperature, precipitation, soil properties, yields and management practices. The CNP‐model offers a low entry threshold model approach to predict the C‐N and now the P dynamics of agricultural soils.

Performance of methods for SARS-CoV-2 variant detection and abundance estimation within mixed population samples.

The accurate identification of SARS-CoV-2 (SC2) variants and estimation of their abundance in mixed population samples (e.g., air or wastewater) is imperative for successful surveillance of community level trends. Assessing the performance of SC2 variant composition estimators (VCEs) should improve our confidence in public health decision making. Here, we introduce a linear regression based VCE and compare its performance to four other VCEs: two re-purposed DNA sequence read classifiers (Kallisto and Kraken2), a maximum-likelihood based method (Lineage deComposition for Sars-Cov-2 pooled samples (LCS)), and a regression based method (Freyja). We simulated DNA sequence datasets of known variant composition from both Illumina and Oxford Nanopore Technologies (ONT) platforms and assessed the performance of each VCE. We also evaluated VCEs performance using publicly available empirical wastewater samples collected for SC2 surveillance efforts. Bioinformatic analyses were performed with a custom NextFlow workflow (C-WAP, CFSAN Wastewater Analysis Pipeline). Relative root mean squared error (RRMSE) was used as a measure of performance with respect to the known abundance and concordance correlation coefficient (CCC) was used to measure agreement between pairs of estimators. Based on our results from simulated data, Kallisto was the most accurate estimator as it had the lowest RRMSE, followed by Freyja. Kallisto and Freyja had the most similar predictions, reflected by the highest CCC metrics. We also found that accuracy was platform and amplicon panel dependent. For example, the accuracy of Freyja was significantly higher with Illumina data compared to ONT data; performance of Kallisto was best with ARTICv4. However, when analyzing empirical data there was poor agreement among methods and variations in the number of variants detected (e.g., Freyja ARTICv4 had a mean of 2.2 variants while Kallisto ARTICv4 had a mean of 10.1 variants). This work provides an understanding of the differences in performance of a number of VCEs and how accurate they are in capturing the relative abundance of SC2 variants within a mixed sample (e.g., wastewater). Such information should help officials gauge the confidence they can have in such data for informing public health decisions.

Mapping and dynamic monitoring of military training-induced vegetation cover loss using Sentinel-2 images and method comparison.

Sustainable management of the US Army installations is critical for military training and readiness of forces. However, monitoring military training-induced vegetation cover disturbances using remote sensing data is challenging due to the lack of methodology for optimizing the selection of spectral variables or predictors and spatial modeling methods. This study aimed to propose and demonstrate a methodological solution for this purpose. The study was conducted in the Fort Riley installation in which three training areas were selected to map and monitor the training-induced vegetation cover loss. Sentinel-2 images and field observations of percentage vegetation cover (PVC) were combined at a spatial resolution of 10m by 10m to map PVC and its dynamics by comparison of two predictor selection methods and five spatial modeling algorithms based on a total of 304 spectral variables from the images before and after the training. Results showed that overall, the correlation-based predictor selection method reduced the relative root mean square error (RRMSE) of PVC predictions by 4.44% than the random forest (RF)-based predictor selection. Machine learning methods including RF, neural network, and support vector machine overall reduced the RRMSE of PVC predictions by 42.83% compared with multiple linear regression and k-nearest neighbors. Combining correlation-based predictor selection and RF modeling, coupled with leave one out cross validation, provided the greatest potential of increasing the accuracy of monitoring the vegetation cover loss. The findings provided useful implications to develop a near real-time system of monitoring military training-induced vegetation cover loss.

Potassium Simulation Using HYDRUS-1D with Satellite-Derived Meteorological Data under Boro Rice Cultivation

Potassium (K) is a critical nutrient for crops, as it is a major constituent in fertilizer formulations. With increasing concentrations of K in agricultural soil, it is necessary to understand its movement and retention in the soil. Sub-surface modeling is an alternative method to overcome the exhausting and uneconomical methods to study and determine the actual concentration of K in soil. HYDRUS-1D is considered an effective finite-element model which is suitable for sub-surface modeling. This model requires the input of ground-station meteorological (GM) data taken at a daily timestep for the simulation period. It can be a limiting factor in the absence of ground stations. The study compares K predictions in surface and sub-surface soil layers under Boro rice cultivation obtained with the usage of different meteorological datasets. Thus, the main hypothesis of the study was to validate that, in the absence of GM data, satellite-based meteorological data could be utilized for simulating the K concentration in soil. The two meteorological datasets that are considered in the study included the GM and satellite-derived NASA-Power (NP) meteorological datasets. The usage of a satellite meteorological product at a field scale may help in applying the method to other regions where GM data is not available. The numerical model results were validated with field experiments from four experimental fields which included varied K doses. The concentration in soil was assessed at the regular depths (0–5, 5–10, 10–15, 15–30, 30–45 and 45–60 cm), and at various stages of crop growth, from bare soil and sowing, to the tillering stages. The concentration of K was measured in the laboratory and also simulated through the optimized model. The modeled values were compared with measured values statistically using relative root mean square error (RMSER) and Nash–Sutcliffe modeling efficiency (E) for simulating K concentration in the soil for the Boro rice cropping pattern with both GM data and NP data. The model was found most suitable for the 0–30 cm depth on all days and for all treatment variations.

Circulant Singular Spectrum Analysis and Discrete Wavelet Transform for Automated Removal of EOG Artifacts from EEG Signals.

Background: Portable electroencephalogram (EEG) systems are often used in health care applications to record brain signals because their ease of use. An electrooculogram (EOG) is a common, low frequency, high amplitude artifact of the eye blink signal that might confuse disease diagnosis. As a result, artifact removal approaches in single EEG portable devices are in high demand. Materials: Dataset 2a from the BCI Competition IV was employed. It contains the EEG data from nine subjects. To determine the EOG effect, each session starts with 5 min of EEG data. This recording lasted for two minutes with the eyes open, one minute with the eyes closed, and one minute with eye movements. Methodology: This article presents the automated removal of EOG artifacts from EEG signals. Circulant Singular Spectrum Analysis (CiSSA) was used to decompose the EOG contaminated EEG signals into intrinsic mode functions (IMFs). Next, we identified the artifact signal components using kurtosis and energy values and removed them using 4-level discrete wavelet transform (DWT). Results: The proposed approach was evaluated on synthetic and real EEG data and found to be effective in eliminating EOG artifacts while maintaining low frequency EEG information. CiSSA-DWT achieved the best signal to artifact ratio (SAR), mean absolute error (MAE), relative root mean square error (RRMSE), and correlation coefficient (CC) of 1.4525, 0.0801, 18.274, and 0.9883, respectively. Comparison: The developed technique outperforms existing artifact suppression techniques according to performance measures. Conclusions: This advancement is important for brain science and can contribute as an initial pre-processing step for research related to EEG signals.

Integrating environmental and satellite data to estimate county-level cotton yield in Xinjiang Province.

Accurate and timely estimation of cotton yield over large areas is essential for precision agriculture, facilitating the operation of commodity markets and guiding agronomic management practices. Remote sensing (RS) and crop models are effective means to predict cotton yield in the field. The satellite vegetation indices (VIs) can describe crop yield variations over large areas but can't take the exact environmental impact into consideration. Climate variables (CVs), the result of the influence of spatial heterogeneity in large regions, can provide environmental information for better estimation of cotton yield. In this study, the most important VIs and CVs for estimating county-level cotton yield across Xinjiang Province were screened out. We found that the VIs of canopy structure and chlorophyll contents, and the CVs of moisture, were the most significant factors for cotton growth. For yield estimation, we utilized four approaches: least absolute shrinkage and selection operator regression (LASSO), support vector regression (SVR), random forest regression (RFR) and long short-term memory (LSTM). Due to its ability to capture temporal features over the long term, LSTM performed best, with an R2 of 0.76, root mean square error (RMSE) of 150 kg/ha and relative RMSE (rRMSE) of 8.67%; moreover, an additional 10% of the variance could be explained by adding CVs to the VIs. For the within-season yield estimation using LSTM, predictions made 2 months before harvest were the most accurate (R2 = 0.65, RMSE = 220 kg/ha, rRMSE = 15.97%). Our study demonstrated the feasibility of yield estimation and early prediction at the county level over large cotton cultivation areas by integrating satellite and environmental data.

Global photosynthetic capacity of C3 biomes retrieved from solar-induced chlorophyll fluorescence and leaf chlorophyll content

The maximum Ribulose-1,5-bisphosphate carboxylase/oxygenase (Rubisco) carboxylation rate (Vcmax) has been widely used to help determine the photosynthetic capacity for carbon cycle modelling. Field measurements of Vcmax are very limited, so fixed Vcmax values over growing seasons are often used, causing considerable uncertainties in modelled results. Also, the lack of a global Vcmax dataset with seasonal variations limits our understanding of the spatial and temporal patterns of Vcmax, and hence photosynthesis. This study proposes a refined data assimilation scheme to retrieve daily Vcmax using remotely sensed solar-induced chlorophyll fluorescence (SIF) and leaf chlorophyll content (LCC) data. Using the refined method, we produced a global Vcmax dataset with daily 0.1° × 0.1° resolutions. The seasonal variations in Vcmax have been successfully retrieved. Vcmax retrieved in this study was compared with field-measured Vcmax from past literature. Good agreements were achieved in relation to Vcmax measured at 201 sites (R2 is 0.78, and relative RMSE is 34%) and the seasonal variation in Vcmax measured at 9 sites (R2 ranges from 0.57 to 0.99, and the relative RMSE ranged from 6% to 28%). Our method effectively utilizes the information from both SIF and LCC, showing advantages over previous approaches that estimate Vcmax based solely on using SIF or LCC. Our results suggest that the spatial distribution of Vcmax depends on the spatial distribution of biomes, along with air temperature and absorbed photosynthetically active radiation (APAR). Higher seasonal mean air temperature and APAR tend to lead to higher seasonal mean Vcmax. The seasonal variation in Vcmax is driven by the seasonal variations in air temperature and APAR, rather than their seasonal means. We also derived and analyzed Vcmax25, which is Vcmax normalized to 25 °C (Vcmax25) and has been found to be more closely related to the Rubisco content. Vcmax25 also has considerable spatiotemporal variations, and the variations differ to that in Vcmax. Different biomes have different ratios of Vcmax25 to LCC. Air temperature and APAR exert weak but positive effects on the seasonal variation in Vcmax25. Vcmax retrieved in this study enhances our understanding of the spatiotemporal pattern of leaf photosynthetic capacity on the global scale and is potentially useful to terrestrial carbon cycle modelling.

Radiometric cross-calibration of Sentinel-2B MSI with HY-1C SCS based on the near simultaneous imaging of common ground targets.

To simplify the cross-calibration process and improve calibration frequency and accuracy, this paper proposes a cross-calibration method for the multispectral remote sensor Multi-Spectral Instrument (MSI) carried by Sentinel-2B using the hyperspectral remote sensor, that is, the satellite calibration spectrometer (SCS) carried by Hai Yang (HY)-1C, as the reference sensor and establishes the calibration process. Precise spectral response matching between SCS and MSI spectral channels is performed by the interpolation and iteration of hyperspectral data to eliminate the difference in band settings and significantly improve the accuracy of cross-calibration coefficients. The SNO-x inherited from the simultaneous nadir overpass (SNO) method is used as a prediction method to carry out cross-calibration imaging in mid- and low-latitude regions, which improves the cross-calibration frequency and broadens the dynamic range of calibration. The cross-calibration coefficients and offsets of MSI B1∼B7 and B8a were obtained by processing the earth observation images of the MSI and SCS on January 24, 2019. Then, the cross-calibration coefficients and offsets are applied to the ocean, farmland and other ground objects with different reflectance, and the reliability and accuracy of the cross-calibration results are evaluated with the Moderate-resolution Imaging Spectroradiometer (MODIS) carried by Terra as a reference. To improve the accuracy of the evaluation, the spectral band adjustment factor between the corresponding channels of MSI and MODIS is used to correct the measured reflectance of MODIS based on the satellite calibration coefficient. The reflectance directly obtained by processing the MSI image is used as the MSI-measured reflectance, the reflectance obtained based on the cross-calibration coefficient is used as the MSI-calculated reflectance, and the reflectance corrected by the spectral band adjustment factor (SBAF) is used as the MODIS-calculated reflectance. The results show that the mean root-mean-square relative error (RMSRE) between the MODIS-calculated reflectance and the MSI-calculated reflectance is 2.16% and that the mean RMSRE between the MODIS-calculated reflectance and the MSI-measured reflectance is 3.05%, indicating that the reflectance corrected based on calibration coefficients is closer to the MODIS-calculated reflectance. Finally, each uncertainty source in the cross-calibration is analyzed, and the comprehensive uncertainty is found to be 4.03%, indicating that SCS can be used as a reference for MSI cross-calibration.

Evaluating how lodging affects maize yield estimation based on UAV observations.

Timely and accurate pre-harvest estimates of maize yield are vital for agricultural management. Although many remote sensing approaches have been developed to estimate maize yields, few have been tested under lodging conditions. Thus, the feasibility of existing approaches under lodging conditions and the influence of lodging on maize yield estimates both remain unclear. To address this situation, this study develops a lodging index to quantify the degree of lodging. The index is based on RGB and multispectral images obtained from a low-altitude unmanned aerial vehicle and proves to be an important predictor variable in a random forest regression (RFR) model for accurately estimating maize yield after lodging. The results show that (1) the lodging index accurately describes the degree of lodging of each maize plot, (2) the yield-estimation model that incorporates the lodging index provides slightly more accurate yield estimates than without the lodging index at three important growth stages of maize (tasseling, milking, denting), and (3) the RFR model with lodging index applied at the denting (R5) stage yields the best performance of the three growth stages, with R2 = 0.859, a root mean square error (RMSE) of 1086.412 kg/ha, and a relative RMSE of 13.1%. This study thus provides valuable insight into the precise estimation of crop yield and demonstra\tes that incorporating a lodging stress-related variable into the model leads to accurate and robust estimates of crop grain yield.

Integrating random forest and crop modeling improves the crop yield prediction of winter wheat and oil seed rape

The fast and accurate yield estimates with the increasing availability and variety of global satellite products and the rapid development of new algorithms remain a goal for precision agriculture and food security. However, the consistency and reliability of suitable methodologies that provide accurate crop yield outcomes still need to be explored. The study investigates the coupling of crop modeling and machine learning (ML) to improve the yield prediction of winter wheat (WW) and oil seed rape (OSR) and provides examples for the Free State of Bavaria (70,550 km2), Germany, in 2019. The main objectives are to find whether a coupling approach [Light Use Efficiency (LUE) + Random Forest (RF)] would result in better and more accurate yield predictions compared to results provided with other models not using the LUE. Four different RF models [RF1 (input: Normalized Difference Vegetation Index (NDVI)), RF2 (input: climate variables), RF3 (input: NDVI + climate variables), RF4 (input: LUE generated biomass + climate variables)], and one semi-empiric LUE model were designed with different input requirements to find the best predictors of crop monitoring. The results indicate that the individual use of the NDVI (in RF1) and the climate variables (in RF2) could not be the most accurate, reliable, and precise solution for crop monitoring; however, their combined use (in RF3) resulted in higher accuracies. Notably, the study suggested the coupling of the LUE model variables to the RF4 model can reduce the relative root mean square error (RRMSE) from −8% (WW) and −1.6% (OSR) and increase the R2 by 14.3% (for both WW and OSR), compared to results just relying on LUE. Moreover, the research compares models yield outputs by inputting three different spatial inputs: Sentinel-2(S)-MOD13Q1 (10 m), Landsat (L)-MOD13Q1 (30 m), and MOD13Q1 (MODIS) (250 m). The S-MOD13Q1 data has relatively improved the performance of models with higher mean R2 [0.80 (WW), 0.69 (OSR)], and lower RRMSE (%) (9.18, 10.21) compared to L-MOD13Q1 (30 m) and MOD13Q1 (250 m). Satellite-based crop biomass, solar radiation, and temperature are found to be the most influential variables in the yield prediction of both crops.

Evaluation of ERA5 and ERA5-Land reanalysis precipitation datasets over Spain (1951–2020)

Reanalysis precipitation estimates are widely used in the fields of meteorology and hydrology because they can provide physical, spatial, and temporal coherent long time series at a global scale. Nevertheless, as a pre-requisite for many applications their performance needs to be assessed. The objective of this study was to evaluate the European Centre for Medium-Range Weather Forecasts (ECMWF) latest fifth-generation reanalysis precipitation products, i.e., ERA5 and ERA5-Land, at country scale in Spain. For doing so, we compared it against a high-resolution precipitation product of the Spanish Meteorological Agency which spans approximately 70 years (1951–2020). A comprehensive assessment (continuous, categorical, probability distribution function (pdf), spatial pattern, and temporal trend) was performed in order to ascertain the quality of the reanalysis products. Results of the analysis revealed a general agreement between observations and ERA5-Land/ERA5 estimates: spearman correlation values between 0.5 and 0.9, Root Mean Square Error (RMSE) mostly between 2 and 8 mm/d and Kling Gupta Efficiency (KGE) values >0.4. Categorical assessment additionally indicated a good performance (Heiken Skill score (HSS) score, also known as kappa, between 0.4 and 0.8). Nevertheless, performance was found to be dependent on the climatic region, precipitation intensity and orography. Correlation revealed a north-west (higher values) south-east (lower values) spatial gradient while relative bias (RBIAS) and RMSE spatial patterns were positively correlated with slope (ρ = 0.41/0.35, 0.69/0.70, respectively). In addition, as indicated by the categorical analysis, along the Mediterranean coast a wet bias (i.e., overestimation of days with precipitation) was found. Reanalysis detection capacity (kappa) shown a negative correlation with the slope (ρ = −0.29/−0.34). Worst model performance is obtained during summer months, with a generalized overestimation. The pdf assessment revealed that the ERA5-Land/ERA5 tended to overestimate light (≥1 and < 5 mm/day), and moderate (≥5 and < 20 mm/day) precipitation categories while underestimating the heavy (≥20 and < 40 mm/day) and violent (≥40 mm/day) categories. Moderate precipitation provided the best detection capacity, as indicated by the precipitation-intensity analysis. ERA5-Land/ERA5 showed a good capacity to reproduce the spatial patterns and temporal trends of the observations. ERA5-Land and ERA5, with a different spatial resolution, performed very similar in all the analysis considered. Mediterranean and northern coast were highlighted as the most critical for reanalysis modelling purposes because of its performance.

Forecasting COVID-19 Infection Using Encoder-Decoder LSTM and Attention LSTM Algorithms

The COVID-19 epidemic has in fact placed the whole community in a dire predicament that has led to numerous tragedies, including an economic downturn, political unrest, and job losses. Forecasting and identifying COVID-19 infection cases is crucial for the government at all levels because the pandemic grows exponentially and results in fatalities. Hence, by giving information about the spread of the epidemic, the government can move quickly at multiple levels to establish new policies and modalities in order to minimize the trajectory of the COVID-19 pandemic's effects on both public health and the economic sectors. Forecasting models for COVID-19 infection cases in the Ural region in Russia were developed using two deep Long Short-Term Memory (LSTM) learning-based approaches namely Encoder–Decoder LSTM and Attention LSTM algorithms. The models were evaluated based on five standard performance evaluation metrics which include Mean Square Error (MSE), Mean Absolute Error (MAE), Root MSE (RMSE), Relative RMSE (RRMSE), and coefficient of determination (R2). However, the Encoder–Decoder LSTM deep learning-based forecasting model achieved the best performance results (MSE=32794.09, MAE=168.56, RMSE=181.09, RRMSE=13.46, and R2=0.87) compared to the model developed with Attention LSTM models.

Prediction of methyl orange dye (MO) adsorption using activated carbon with an artificial neural network optimization modeling

In this study, methyl orange (MO) dye removal by adsorption utilizing activated carbon made from date seeds (DPAC) was modeled using an artificial neural network (ANN) technique. Instrumental investigations such as X-ray diffraction (XRD), scanning electron microscopy (SEM), and Brunauer-Emmett-Teller (BET) analysis were used to assess the physicochemical parameters of adsorbent. By changing operational parameters including adsorbent dosage (0.01–0.03 g), solution pH 3–8, initial dye concentration (5–20 mg/L), and contact time (2–60 min), the viability of date seeds for the adsorptive removal of methyl orange dye from aqueous solution was assessed in a batch procedure. The system followed the pseudo 2nd order kinetic model for DPAC adsorbent, according to the kinetic study (R2 = 0.9973). The mean square error (MSE), relative root mean square error (RRMSE), root mean square error (RMSE), mean absolute percentage error (MAPE), relative error (RE), and correlation coefficient (R2) were used to measure the ANN model performance. The maximum RE was 8.24% for the ANN model. Two isotherm models, Langmuir and Freundlich, were studied to fit the equilibrium data. Compared with the Freundlich isotherm model (R2 = 0.72), the Langmuir model functioned better as an adsorption isotherm with R2 of 0.9902.Thus, this study demonstrates that the dye removal process can be predicted using an ANN technique, and it also suggests that adsorption onto DPAC may be employed as a main treatment for dye removal from wastewater.