Terms Of Root Mean Square Error Research Articles

A novel ionospheric TEC mapping function with azimuth parameters and its application to the Chinese region

The ionospheric mapping function (MF) for Global Navigation Satellite System (GNSS), a mutual projection method for the slant total electron content (STEC) and vertical total electron content, is one of the significant factors affecting the performance of ionospheric models. The commonly used MF assumes isotropic TEC variations and takes into account only the satellite elevation angle, which may result in significant ionospheric projection errors, especially at low elevation angles. Based on the single-layer model, we propose an additional azimuth parameter mapping function (APMF). The APMF was estimated and evaluated by the NeQuick model during the periods of January 2014 and January 2022 from the aspect of simulation and measured STEC during the periods of 2014 and 2022 from the aspect of actual measurements over China, respectively. Compared to the modified single-layer model mapping function (MSLM-MF), the experimental results indicate that (1) The APMF can significantly reduce the ionospheric projection error, and the fluctuation in errors with different azimuth angles is small. (2) According to the evaluation based on the NeQuick simulation during the TEC peak time, when the ionosphere is quite active, the upper and lower quartiles of the absolute projection error boxplot of the APMF relative to the MSLM-MF in January 2014 are reduced by 56.1% and 60.0%, respectively, and in January 2022, they are reduced by 67.7% and 65.2%, respectively. Similarly, the upper whiskers in the boxplot are reduced by 54.7% and 67.5% in January 2014 and January 2022, respectively; the APMF performance in terms of the root mean square error (RMSE) is improved by 47.0% in January 2014 and 58.3% in January 2022. (3) According to the evaluation based on the measured STEC from GNSS raw data during the TEC peak time, the upper and lower quartiles of the absolute mapping error boxplot of the APMF relative to the MSLM-MF in 2014 are reduced by 48.9% and 46.9%, respectively, while in 2022, they are reduced by 48.3% and 41.2%, respectively. The upper whiskers in the boxplot are reduced by 41.8% and 35.2% in 2014 and 2022, respectively; the APMF performance in terms of RMSE is improved by 44.6% in 2014 and 39.2% in 2022.

Comparison of three-dimensional imaging of the nose using three different 3D-photography systems: an observational study

BackgroundNew 3D technologies for superficial soft tissue changes, especially in plastic and reconstructive surgical procedures, can improve the planning and documentation of facial surgeries. The purpose of this study was to compare and determine the applicability and feasibility of three different 3D-photography systems in clinical practice imaging the nose.MethodsA total of 16 healthy non-operated noses were included in this prospective study. A plaster model of each nose was produced, digitized, and converted to a .stl mesh (= ground truth model). Three-dimensional images of each nose were then taken using Artec Space Spider (gold standard), Planmeca ProFace®, and the Bellus3D Dental Pro application. All resulting .stl files were aligned to the ground truth model using MeshLab software, and the root mean square error (RMSE), mean surface distance (MSD), and Hausdorff distance (HD) were calculated.ResultsThe Artec Space Spider 3D-photography system showed significantly better results compared to the two other systems in regard to RMSE, MSD, and HD (each p < 0.001). There was no significant difference between Planmeca ProFace® and Bellus3D Dental Pro in terms of RMSE, MSD, and HD. Overall, all three camera systems showed a clinically acceptable deviation to the reference model (range: -1.23–1.57 mm).ConclusionsThe three evaluated 3D-photography systems were suitable for nose imaging in the clinical routine. While Artec Space Spider showed the highest accuracy, the Bellus3D Dental Pro app may be the most feasible option for everyday clinical use due to its portability, ease of use, and low cost. This study presents three different systems, allowing readers to extrapolate to other systems when planning to introduce 3D photography in the clinical routine.

A hybrid model of CNN and LSTM autoencoder-based short-term PV power generation forecasting

AbstractSolar energy is one of the main renewable energies available to fulfill global clean energy targets. The main issue of solar energy like other renewable energies is its randomness and intermittency which affects power grids stability. As a solution for this issue, energy storage units could be used to store surplus energy and reuse it during low solar generation intervals. Also, in order to sustain stable power grid and better grid operation and energy storage management, photovoltaic (PV) power forecasting is inevitable. In this paper, new hybrid model based on deep learning techniques is proposed to predict short-term PV power generation. The proposed model incorporates convolutional neural network (CNN) and long short-term memory (LSTM) autoencoder network. The new model differentiates itself in accomplishing high prediction accuracy by extracting spatial features in time series via CNN layers and temporal features between the time series data through LSTM. The introduced model is tested on dataset of power generation from southern UK solar farm and the weather data corresponding to same location and time intervals; the forecasting performance of the suggested model is evaluated in metrics of root-mean-square error (RMSE) and mean absolute error (MAE). The used model is compared with different models from the literature either of pure type of network such as LSTM and gated recurrent unit (GRU) or hybrid combination of different networks like CNN-LSTM and CNN-GRU. The results show that proposed model provides enhanced results and reduces training time significantly compared to other competitive models, where the performance of the proposed model improved averagely by 5% to 25% in terms of RMSE and MAE performance metrics, and the execution time of training significantly reduced with almost 70% less compared to other models.

Enhancing sediment transport predictions through machine learning-based multi-scenario regression models

Machine learning is one effective way of increasing the accuracy of sediment transport models. Machine learning captures patterns in the sequence of both structured and unstructured data and uses it for forecasting. In this research, the different regression models were train to forecast sediment data using 8 years of measured sediment data collected in Sg. Linggui suspended sediment station. Data from different scenarios were used where each scenario indicates the number of lags. Seven regression models, namely, Linear Regression, Regression Trees, Support Vector Machines, Gaussian Process Regression, Kernel Approximation, Ensemble of Trees, and Neural Network were trained using the data and compared. The trained models were evaluated using Root Mean Square Error (RMSE) and Coefficient of Determination (R2). The best-performing models from two different types of regression models were chosen and they were tested using the test data to find the Relative Percentage Error (RPE) of the predicted data. The Exponential Gaussian Process Regression model performs much better than the other models in terms of RMSE and R2 values. When the exponential models from all 3 scenarios are compared, scenario 3 seems to have a better-performing model but only by a very small margin, after using testing data, the result shows scenario 3 has less RPE compared to other models. Hence, it can be deduced that the exponential gaussian process regression model from scenario 3 is the best-performing model overall in terms of RSME, R2, and RPE.

Forecasting Ionospheric foF2 Using Bidirectional LSTM and Attention Mechanism

AbstractThe critical frequency of ionospheric F2 layer (foF2) is an important ionospheric characteristic parameter. In this paper, a deep learning model based on Bidirectional long short‐term memory (BiLSTM) and attention mechanism is implemented for predicting the foF2 parameter. The inputs of models are the foF2 of globally available ionospheric ionosonde stations, geographic longitude and latitude, world time (UT), geomagnetic activity index, and solar activity index from 2015 to 2017. The superiority of the model is analyzed from different latitudes, seasons, and geomagnetic conditions. The results show that the prediction performance of the Bidirectional long short‐term memory model based on attention mechanism (BiLSTM‐Attention) is better than other models. The performance of the prediction model is optimal at high latitudes. The root mean square error (RMSE) and correlation coefficient (R) of the BiLSTM‐Attention model are 0.539 MHZ and 0.908 MHz at high latitudes, respectively. In terms of RMSE, it is 25.243%, 18.209%, and 11.203% lower than those of the international reference ionosphere (IRI), LSTM, and BiLSTM models, respectively. The prediction results of the four seasons show that the models are more applicable in winter. Compared with the IRI model, the RMSE of the BiLSTM‐Attention model in spring, summer, autumn, and winter is reduced by 24.344%, 21.181%, 25.058%, and 30.948%, respectively. The prediction effect of the BiLSTM‐Attention model is improved in the magnetic quiet period, the magnetic moderate period and the magnetic storm period. Also, the improvement effect is more obvious in the magnetostatic day, and the RMSE is reduced by 27.462% compared with the IRI model.

A technical evaluation on the mathematical attitudes and fitting accuracy of soil moisture retention curve models

Numerous mathematical equations have been formulated in the literature of different researchers for describing soil moisture retention curve (SMRC), which can be applied to simulate and solve soil hydraulic modeling problems. The primary concern lies in selecting an efficient model to simulate accurately the S-shaped curve or sigmoid-type of the SMRC for soils with different textures. Therefore, the objective of this study was a comprehensive and technical evaluation of 50 developed models of the SMRC based on the influence of parameters on the behavior of the curve and the ability of their fitting accuracy on 728 soil samples of the UNSODA dataset, that has not been investigated so far. Statistical criteria including corrected Akaike’s information criterion (AICc), root mean square error (RMSE) and coefficient of determination (R2) together with Duncan's multiple range test and cluster analysis were employed to assess the fitting accuracy of SMRC models to measured data. Results from fitting accuracy on the UNSODA dataset indicated that Brutsaert model provided the best fit to the measured data compared to other models in 14.6% of the soil samples with RMSE = 0.0125 and AICc = -94.18. This model was classified in the same cluster with Groenevelt and Grant (GG1, GG2 and GG3), Dexter, Mualem, and Fredlund and Xing models and did not have a significant difference in terms of RMSE. Also, Brutsaert model had the highest fitting accuracy in 67% of different soil textural classes compared to other models. Finally, the technical evaluation in terms of accuracy, flexibility and simplicity of the fitting process showed that the Brutsaert, Mualem, Dexter and GG3 models can be selected for better simulation of the SMRC in water and soil research.

Prediction of Feed Efficiency and Performance-Based Traits in Fish via Integration of Multiple Omics and Clinical Covariates.

Fish aquaculture is a rapidly expanding global industry, set to support growing demands for sources of marine protein. Enhancing feed efficiency (FE) in farmed fish is required to reduce production costs and improve sector sustainability. Recognising that organisms are complex systems whose emerging phenotypes are the product of multiple interacting molecular processes, systems-based approaches are expected to deliver new biological insights into FE and growth performance. Here, we establish 14 diverse layers of multi-omics and clinical covariates to assess their capacities to predict FE and associated performance traits in a fish model (Oncorhynchus tshawytscha) and uncover the influential variables. Inter-omic relatedness between the different layers revealed several significant concordances, particularly between datasets originating from similar material/tissue and between blood indicators and some of the proteomic (liver), metabolomic (liver), and microbiomic layers. Single- and multi-layer random forest (RF) regression models showed that integration of all data layers provide greater FE prediction power than any single-layer model alone. Although FE was among the most challenging of the traits we attempted to predict, the mean accuracy of 40 different FE models in terms of root-mean square errors normalized to percentage was 30.4%, supporting RF as a feature selection tool and approach for complex trait prediction. Major contributions to the integrated FE models were derived from layers of proteomic and metabolomic data, with substantial influence also provided by the lipid composition layer. A correlation matrix of the top 27 variables in the models highlighted FE trait-associations with faecal bacteria (Serratia spp.), palmitic and nervonic acid moieties in whole body lipids, levels of free glycerol in muscle, and N-acetylglutamic acid content in liver. In summary, we identified subsets of molecular characteristics for the assessment of commercially relevant performance-based metrics in farmed Chinook salmon.

A Machine Learning-Aided Network Contention-Aware Link Lifetime- and Delay-Based Hybrid Routing Framework for Software-Defined Vehicular Networks

The functionality of Vehicular Ad Hoc Networks (VANETs) is improved by the Software-Defined Vehicular Network (SDVN) paradigm. Routing is challenging in vehicular networks due to the dynamic network topology resulting from the high mobility of nodes. Existing approaches for routing in SDVN do not exploit both link lifetimes and link delays in finding routes, nor do they exploit the heterogeneity that exists in links in the vehicular network. Furthermore, most of the existing approaches compute parameters at the controller entirely using heuristic approaches, which are computationally inefficient and can increase the latency of SDVN as the network size grows. In this paper, we propose a novel hybrid algorithm for routing in SDVNs with two modes: the highest stable least delay mode and the highest stable shortest path mode, in which the mode is selected by estimating the network contention. We distinctly identify two communication channels in the vehicular network as wired and wireless, where network link entropy is formulated accordingly and is used in combination with pending transmissions to estimate collision probability and average network contention. We use the prospect of machine learning to predict the wireless link lifetimes and one-hop channel delays, which yield very low Root Mean Square Errors (RMSEs), depicting their very high accuracy, and the wireless link lifetime prediction using deep learning yields a much lower average computational time compared to an optimization-based approach. The proposed novel algorithm selects only stable links by comparing them with a link lifetime threshold whose optimum value is decided experimentally. We propose this routing framework to be compatible with the OpenFlow protocol, where we modify the flow table architecture to incorporate a route valid time and send a packet_in message to the controller when the route’s lifetime expires, requesting new flow rules. We further propose a flow table update algorithm to map computed routes to flow table entries, where we propose to incorporate an adaptive approach for route finding and flow rule updating upon reception of a packet_in message in order to minimize the computational burden at the controller and minimize communication overhead associated with control plane communication. This research contributes a novel hybrid routing framework for the existing SDVN paradigm, scrutinizing machine learning to predict the lifetime and delay of heterogeneity links, which can be readily integrated with the OpenFlow protocol for better routing applications, improving the performance of the SDVN. We performed realistic vehicular network simulations using the network simulator 3 by obtaining vehicular mobility traces using the Simulation of Urban Mobility (SUMO) tool, where we collected data sets for training the machine learning models using the simulated environment in order to test models in terms of RMSE and computational complexity. The proposed routing framework was comparatively assessed against existing routing techniques by evaluating the communication cost, latency, channel utilization, and packet delivery ratio. According to the results, the proposed routing framework results in the lowest communication cost, the highest packet delivery ratio, the least latency, and moderate channel utilization, on average, compared to routing in VANET using Ad Hoc On-demand Distance Vector (AODV) and routing in SDVN using Dijkstra; thus, the proposed routing framework improves routing in SDVN. Furthermore, results show that the proposed routing framework is enhanced with increasing routing frequency and network size, as well as at low vehicular speeds.

S2MetNet: A novel dataset and deep learning benchmark for methane point source quantification using Sentinel-2 satellite imagery

Methane, as a crucial greenhouse gas, plays a significant role in global warming, contributing to approximately one-quarter of the observed climate change since pre-industrial times. Consequently, the detection and quantification of major methane emitters are vital in addressing this issue effectively. Satellite sensors with shortwave infrared (SWIR) spectral bands provide valuable information for monitoring methane emissions. For example, Sentinel-2 multispectral data have the capability to detect methane plumes of large point sources. As such, a wide range of quantification approaches have been developed to quantify methane source rates based on this dataset. Most of the existing methods, however, require auxiliary data, such as wind speed, and have large uncertainties. In this study, we introduce a novel approach based on deep learning models to enhance the precision of methane quantification using Sentinel-2 data without the reliance on external data sources. To train the proposed deep learning model, a comprehensive benchmark dataset has been generated, using Sentinel-2 data. This dataset is created by integrating simulated plumes and background noise extracted from real Sentinel-2 images. This approach ensures the integration of realistic environmental conditions within the simulated data, enhancing the robustness and reliability of our proposed model. The generated benchmark dataset is utilized in different deep learning architectures, namely VGG-19, ResNet-50, Inception-v3, DenseNet-121, Swin-T, and EfficientNet-V2L, to estimate methane source rate. The performance of deep models has been evaluated in three learning strategies, namely from scratch, transfer-learning, and fine-tuning. The fine-tuned EfficientNet-V2L achieves the highest accuracy with root-mean-square error (RMSE), mean absolute percentage error (MAPE), and Pearson R of 2101 kg h−1, 10.05%, and 95.70%, respectively. More importantly, the proposed model demonstrates superior performance compared to conventional physical-based quantification methods (e.g., integrated mass enhancement) and recently developed deep learning model techniques (e.g., MethaNet). In particular, the proposed model exhibits an improvement of approximately 1287 kg h−1 in terms of RMSE, a 3.92% reduction in MAPE, and a 5.01% enhancement in R compared to the IME method. These results highlight the advancements achieved by the proposed approach in accurately quantifying methane emissions using Sentinel-2 imagery. The generated benchmark dataset and the developed deep learning model presented in this study serve as a fundamental resource and constructive framework for future research, promoting extensive implementation across various methane monitoring scenarios on different satellites and in distinct geographic regions, which delivering greater effectiveness to global methane emission monitoring initiatives.

East Asia Reanalysis System (EARS)

Abstract. Reanalysis data play a vital role in weather and climate study as well as meteorological resource development and application. In this work, the East Asia Reanalysis System (EARS) was developed using the Weather Research and Forecasting (WRF) model and the Gridpoint Statistical Interpolations (GSI) data assimilation system. The regional reanalysis system is forced by the European Centre for Medium-Range Weather Forecasts (ECMWF) global reanalysis ERA-Interim data at 6 h intervals. Hourly surface observations are assimilated by the Four-Dimension Data Assimilation (FDDA) scheme during the WRF model integration; upper observations are assimilated in three-dimensional variational data assimilation (3D-VAR) mode at the analysis moment. It should be highlighted that many of the assimilated observations have not been used in other reanalysis systems. The reanalysis runs from 1980 to 2018, producing a regional reanalysis dataset covering East Asia and surrounding areas at 12 km horizontal resolution, 74 sigma levels, and 3 h intervals. Finally, an evaluation of EARS has been performed with respect to the root mean square error (RMSE), based on the 10-year (2008–2017) observational data. Compared to the global reanalysis data of ERA-Interim, the regional reanalysis data of EARS are closer to the observations in terms of RMSE in both surface and upper-level fields. The present study provides evidence for substantial improvements seen in EARS compared to the ERA-Interim reanalysis fields over East Asia. The study also demonstrates the potential use of the EARS data for applications over East Asia and proposes further plans to provide the latest reanalysis in real-time operation mode. Simple data and updated information are available on Zenodo at https://doi.org/10.5281/zenodo.7404918 (Yin et al., 2022), and the full datasets are publicly accessible on the Data-as-a-Service platform of the China Meteorological Administration (CMA) at http://data.cma.cn (last access: 19 May 2023).

Improvement of transpiration estimation based on a two-leaf conductance-photosynthesis model with seasonal parameters for temperate deciduous forests.

Conductance-photosynthesis (Gs-A) models, accompanying with light use efficiency (LUE) models for calculating carbon assimilation, are widely used for estimating canopy stomatal conductance (Gs) and transpiration (Tc) under the two-leaf (TL) scheme. However, the key parameters of photosynthetic rate sensitivity (gsu and gsh) and maximum LUE (ϵmsu and ϵmsh) are typically set to temporally constant values for sunlit and shaded leaves, respectively. This may result in Tc estimation errors, as it contradicts field observations. In this study, the measured flux data from three temperate deciduous broadleaved forests (DBF) FLUXNET sites were adopted, and the key parameters of LUE and Ball-Berry models for sunlit and shaded leaves were calibrated within the entire growing season and each season, respectively. Then, the estimations of gross primary production (GPP) and Tc were compared between the two schemes of parameterization: (1) entire growing season-based fixed parameters (EGS) and (2) season-specific dynamic parameters (SEA). Our results show a cyclical variability of ϵmsu across the sites, with the highest value during the summer and the lowest during the spring. A similar pattern was found for gsu and gsh, which showed a decrease in summer and a slight increase in both spring and autumn. Furthermore, the SEA model (i.e., the dynamic parameterization) better simulated GPP, with a reduction in root mean square error (RMSE) of about 8.0 ± 1.1% and an improvement in correlation coefficient (r) of 3.7 ± 1.5%, relative to the EGS model. Meanwhile, the SEA scheme reduced Tc simulation errors in terms of RMSE by 3.7 ± 4.4%. These findings provide a greater understanding of the seasonality of plant functional traits, and help to improve simulations of seasonal carbon and water fluxes in temperate forests.

Improved learning performance for small datasets in high dimensions by new dual-net model for non-linear interpolation virtual sample generation

The number of reliable samples obtained in early decision-making activity is usually relatively small. Due to variable distribution and incomplete structure of tiny datasets, it is challenging to create reliable and robust predictive modeling using classic statistical and machine learning models in small sample settings. The virtual sample generation (VSG) technique improves model learning accuracies for minimal datasets across diverse applications. Virtual samples on independent variables were generated using established VSG methods predicated on the assumption of a probability distribution or a membership function to fill data gaps. However, in the actual world, non-linear function interactions between variables are common. To address this issue, this paper developed a novel VSG method called Dual-VSG, which generates non-linear interpolation virtual samples using a self-supervised learning (SSL) framework to improve learning performance on small datasets. We generated non-linear interpolation virtual samples without labels by estimating non-linear functions and transforming them into a high-dimensional space using the proposed dual-net model. The weights of the dual-net model are transferred to a downstream task to generate virtual sample labels. To demonstrate the effectiveness of the suggested strategy, this research employed five datasets. On the Backpropagation Neural Networks (BPNN) predictive model, we compared the suggested method's prediction performance to two state-of-the-art VSG approaches. To assess prediction performance on a regression dataset, the Mean Absolute Percentage Error (MAPE) and the Root Mean Square Error (RMSE) are used. Furthermore, the classification accuracy (ACC) and the Fl measure are used to assess classification capability on classification datasets. In addition, the paired t-test was utilized to see if the suggested Dual-VSG approach differed significantly from the other VSG methods in terms of RMSE, MAPE, accuracy (ACC), or F1 score. For short datasets, the suggested Dual-VSG method outperforms those VSG methods, according to our experimental results.

Biogeography-Based Teaching Learning-Based Optimization Algorithm for Identifying One-Diode, Two-Diode and Three-Diode Models of Photovoltaic Cell and Module

This article handles the challenging problem of identifying the unknown parameters of solar cell three models on one hand and of photovoltaic module three models on the other hand. This challenge serves as the basis for fault detection, control, and modelling of PV systems. An accurate model of PV is essential for the simulation research of PV systems, where it has a significant role in the dynamic study of these systems. The mathematical models of the PV cell and module have nonlinear I-V and P-V characteristics with many undefined parameters. In this paper, this identification problem is solved as an optimization problem based on metaheuristic optimization algorithms. These algorithms use root mean square error (RMSE) between the calculated and the measured current as an objective function. A new metaheuristic amalgamation algorithm, namely biogeography-based teaching learning-based optimization (BB-TLBO) is proposed. This algorithm is a hybridization of two algorithms, the first one is called BBO (biogeography-based optimization) and the second is TLBO (teaching learning-based optimization). The BB-TLBO is proposed to identify the unknown parameters of one, two and three-diode models of the RTC France silicon solar cell and of the commercial photovoltaic solar module monocrystalline STM6-40/36, taking into account the performance indices: high precision, more reliability, short execution time and high convergence speed. This identification is carried out using experimental data from the RTC France silicon solar cell and the STM6-40/36 photovoltaic module. The efficiency of BB-TLBO is checked by comparing its identification results with its own single algorithm BBO, TLBO and newly introduced hybrid algorithms such as DOLADE, LAPSO and others. The results reveal that the suggested approach surpasses all compared algorithms in terms of RMSE (RMSE min, RMSE mean and RMSE max), standard deviation of RMSE values (STD), CPU (execution time), and convergence speed.

Adaptive-weighted high order TV algorithm for sparse-view CT reconstruction.

With the development of low-dose computed tomography (CT), incomplete data reconstruction has been widely concerned. The total variation (TV) minimization algorithm can accurately reconstruct images from sparse or noisy data. However, the traditional TV algorithm ignores the direction of structures in images, leading to the loss of edge information and block artifacts when the object is not piecewise constant. Since the anisotropic information can facilitate preserving the edge and detail information in images, we aim to improve the TV algorithm in terms of reconstruction accuracy via this approach. In this paper, we propose an adaptive-weighted high order total variation (awHOTV) algorithm. We construct the second order TV-norm using the second order gradient, adapt the anisotropic edge property between neighboring image pixels, adjust the local image-intensity gradient to keep edge information, and design the corresponding Chambolle-Pock (CP) solving algorithm. Implementing the proposed algorithm, comprehensive studies are conducted in the ideal projection data experiment where the Structural similarity (SSIM), Root Mean Square Error (RMSE), Contrast to noise ratio (CNR), and modulation transform function (MTF) curves are utilized to evaluate the quality of reconstructed images in statism, structure, spatial resolution, and contrast, respectively. In the noisy data experiment, we further use the noise power spectrum (NPS) curve to evaluate the reconstructed images and compare it with other three algorithms. We use the 2D slice in the XCAT phantom, 2D slice in TCIA Challenge data and FORBILD phantom as simulation phantoms and use real bird data for real verification. The results show that, compared with the traditional TV and FBP algorithms, the awHOTV has better performance in terms of RMSE, SSIM, and Pearson correlation coefficient (PCC) under the projected data with different sparsity. In addition, the awHOTV algorithm is robust against the noise of different intensities. The proposed awHOTV method can reconstruct the images with high accuracy under sparse or noisy data. The awHOTV solves the strip artifacts caused by sparse data in the FBP method. Compared with the TV method, the awHOTV can effectively suppress block artifacts and has good detail protectionability.

Assessing and comparing the subseasonal variations of summer soil moisture of satellite products over eastern China

AbstractUsing the 43‐station soil moisture (SM) at a depth of 0–10 cm over eastern China, the near‐surface (about 0–5 cm) SM products from satellite missions including the level‐2 product of Chinese FengYun 3C (FY3C), level‐2 neural network product of the European Soil Moisture and Ocean Salinity (SMOS) and level‐4 product of the US Soil Moisture Active Passive (SMAP) were evaluated and compared during summers of 2015–2018. Thus, features of those products can be identified for their further application of climate study over eastern China. Large diversity is found among the satellite and field SM in terms of spatial distribution. This disagreement may cause the poor performance of the relationship between spatial coefficients of variation (CV) and mean SM. Compared with the field SM, the FY3C SM has smaller bias than the other satellite products. FY3C also performs better than SMOS in terms of the root‐mean‐square error (RMSE) and unbiased RMSE (ubRMSE) but has the smallest correlation coefficient (R). The SMAP product is generally the best among the three products in terms of RMSE, ubRMSE and R. However, a good performance in those metrics does not guarantee the same results on various time scales. On subseasonal time scales, the R in FY3C is the smallest among the three products, and the SMAP product has the largest R, but its amplitudes of the subseasonal variations are much smaller than the field observations. This indicates that when the SMAP products are applied for the analysis on subseasonal SM–atmosphere interactions, the effects of SM may be underestimated. On 10–30 days and above 60 days, dry period tends to have large spatial CV but this phenomenon is weak in all the satellite products. On the other hand, dry area tends to have large temporal CV on the four time scales, and this relationship is the strongest in SMAP but the weakest in FY3C. Therefore, there are large uncertainties of variability among different satellite SM products on subseasonal time scales over eastern China. Besides seasonal and overall performance, more attention is called to the variations on different time scales.

A comparative study of heterogeneous and homogeneous ensemble approaches for landslide susceptibility assessment in the Djebahia region, Algeria.

This study aims to compare the performance of ensembles according to their inherent diversity in the context of landslide susceptibility assessment. Heterogeneous and homogeneous ensemble types can be distinguished; four ensembles of each approach were implemented in the Djebahia region. The heterogeneous ensembles include stacking (ST), voting (VO), weighting (WE), and a new approach in landslide assessment called meta-dynamic ensemble selection (DES), while the homogeneous ensembles include AdaBoost (ADA), bagging (BG), random forest (RF), and random subspace (RSS). To ensure a consistent comparison, each ensemble was implemented using individual base learners. The heterogeneous ensembles were generated by combining eight different machine learning algorithms, while the homogeneous ensembles only used a single base learner, with diversity achieved through resampling the training dataset. The spatial dataset used in this study consisted of 115 landslide events and 12 conditioning factors, which were randomly divided into training and testing datasets. The models were evaluated through various aspects, including receiver operating characteristic (ROC) curves, root mean squared error (RMSE), landslide density distribution (LDD), threshold-dependent metrics (Kappa index, accuracy, and recall scores), and a global visual representation using the Taylor diagram. Additionally, a sensitivity analysis (SA) was conducted for the best performing models to assess the importance of the factors and the resilience of the ensembles. The results revealed that homogeneous ensembles outperformed heterogeneous ensembles in terms of AUC and threshold-dependent metrics, with AUC ranging from 0.962 to 0.971 for the test dataset. ADA was the best performing model for these metrics and the least in terms of RMSE (0.366). However, the heterogeneous ensemble ST provided a finer RMSE (0.272), and DES showed the best LDD, indicating a stronger potential to generalize the phenomenon. The Taylor diagram was consistent with the other results, indicating that ST was the best performing model, followed by RSS. The SA demonstrated that RSS was the most robust (mean AUC variation of - 0.022) and ADA was the least robust (mean AUC variation of - 0.038).

A New Spectral Compression Method based on the Minimization of the Color Difference and Root Mean Square Error

This paper describes two new interim connection spaces (ICSs), P-2OC and P-3OC, for spectral image compression and reconstruction. For this type of ICS, the weighting table or compression matrix H is modelled as a variable. The associated reconstruction matrix N is chosen as the Weiner estimation matrix. The objective function f(H) is the combination of the averages of the CIEDE2000 color difference (&#x394;E00) and the root mean square error (RMSE) between the original and reconstructed reflectance values. Hence, the compression matrix H is determined by solving the nonlinear minimization problem based on the Munsell training dataset. The proposed ICSs, P-2OC and P-3OC, were tested and compared, respectively, with ICS-2SI and ICS-3SI developed by Zhang et&#xA0;al.&#xA0;(JOSAA, Vol. 29, pp. 1027&#x2013;1034) in 2012 using the NCS dataset, and 2 spectral images. Performance tests showed that the proposed P-2OC and P-3OC ICSs are better than the ICS-2SI and ICS-3SI ICSs, respectively, in terms of RMSE, goodness of fit coefficient (GFC), and &#x394;E00 under CIE illuminants D65, A, C and F11. Therefore, it is expected that the P-2OC and P-3OC ICSs can find applications in spectral image compression and cross-media reproduction.

Predicting VO2max in Children and Adolescents Aged between 6 and 17 Using Physiological Characteristics and Participation in Sport Activities: A Cross-Sectional Study Comparing Different Regression Models Stratified by Gender.

The aim of this study is to use different regression models to capture the association between cardiorespiratory fitness VO2max (measured in mL/kg/min) and somatometric characteristics and sports activities and making better predictions. multiple linear regression (MLR), quantile regression (QR), ridge regression (RR), support vector regression (SVR) with three different kernels, artificial neural networks (ANNs), and boosted regression trees (RTs) were compared to explain and predict VO2max and to choose the best performance model. The sample consisted of 4908 children (2314 males and 2594 females) aged between 6 and 17. Cardiorespiratory fitness was assessed by the 20 m maximal multistage shuttle run test and maximal oxygen uptake (VO2max) was calculated. Welch t-tests, Mann-Whitney-U tests, X2 tests, and ANOVA tests were performed. The performance measures were root mean square error (RMSE), mean absolute error (MAE), and coefficient of determination (R2). All analyses were stratified by gender. A comparison of the statistical indices for both the predicted and actual data indicated that in boys, the MLR model outperformed all other models in all indices, followed by the linear SVR model. In girls, the MLR model performed better than the other models in R2 but was outperformed by SVR-RBF in terms of RMSE and MAE. The overweight and obesity categories in both sexes (p &lt; 0.001) and maternal prepregnancy obesity in girls had a significant negative effect on VO2max. Age, weekly football training, track and field, basketball, and swimming had different positive effects based on gender. The MLR model showed remarkable performance against all other models and was competitive with the SVR models. In addition, this study's data showed that changes in cardiorespiratory fitness were dependent, to a different extent based on gender, on BMI category, weight, height, age, and participation in some organized sports activities. Predictors that are not considered modifiable, such as gender, can be used to guide targeted interventions and policies.

Application of classical and novel integrated machine learning models to predict sediment discharge during free-flow flushing

In this study, the capabilities of classical and novel integrated machine learning models were investigated to predict sediment discharge (Qs) in free-flow flushing. Developed models include Multivariate Linear Regression (MLR), Artificial Neural Network (ANN), Adaptive Neuro-Fuzzy Inference System (ANFIS), Group Method of Data Handling (GMDH), and four hybrid forms of GMDH and Support Vector Regression (SVR) in combination with Henry Gas Solubility Optimization (HGSO) and Equilibrium Optimizer (EO) algorithms. The number of 160 datasets that were applied to assess these models was extracted from the Janssen (PhD’s Thesis, Engineering, Civil and Environmental Engineering. University of California, 1999) experimental study. Input parameters to predict Qs included the water level in the reservoir (hw), bed level in the flushing channel (hb), outflow (Qout), inflow (Qin), and elapsed time of flushing (T). The performance of all models was evaluated by four statistical indices of root mean square error (RMSE), mean absolute error (MAE), correlation coefficient (R2), and Mean absolute relative error (MARE). Evaluation of results demonstrated that the HGSO and EO algorithms could enhance the accuracy of the GMDH model (up to 26% and 22% in terms of RMSE), respectively. According to statistical criteria, the SVR-EO and SVR-HGSO provided the highest accuracy in both training (R2 = 0.98) and validation phases (R2 = 0.96). Moreover, among the developed models, the GMDH-HGSO algorithm provided excellent fitness to the observed data (R2 = 0.96, RMSE = 22.37, MAE = 15.65, and MARE = 0.26). The results indicated the high efficiency of the HGSO and EO algorithms in improving the accuracy of the GMDH and SVR models. However, among the developed models, the GMDH-HGSO is the most accurate model and is recommended for sediment transport modelling.

Spatiotemporal deep learning rainfall-runoff forecasting combined with remote sensing precipitation products in large scale basins

Rainfall-runoff modeling is a complex nonlinear spatiotemporal prediction problem. However, few studies have considered the spatial characteristics of rainfall-runoff relationship in runoff forecasts based on machine learning. With the emergence of high-resolution Satellite-based Precipitation Products (SPPs) and the continuous improvement of rainfall estimation accuracy, the shortcoming of sparse spatial information for in-situ rainfall monitoring has been made up. Therefore, this study developed a large scale spatiotemporal deep learning rainfall-runoff (SDLRR) forecasting model for hydrological stations in the upper Yangtze River, and evaluated the positive impact of utilizing spatial information of three SPPs on reducing errors of runoff forecasts. The adopted remote sensing precipitation products are bias-corrected Climate Hazards Group InfraRed Precipitation with Station data (CHIRPS), Integrated Multi-satellite Retrievals for Global Precipitation Measurement data (IMERG) and Tropical Rainfall Measuring Mission Multi-satellite Precipitation Analysis data (TMPA). For runoff forecasting at the Luoduxi (LDX) hydrological station, compared to regular Long Short Term Memory Network (LSTM) model, the proposed SDLRR model that utilizing IMERG data as precipitation input (IMERG_SDLRR) improved 15% in terms of Coefficient of Determination (R2) and improved 25% in terms of Root Mean Squared Error (RMSE). Compared to the best performance model among models using area-averaged precipitation as input, IMERG_SDLRR improved 5% in terms of R2 and 11% in terms of RMSE. Good performance was also acquired in the other hydrological stations. For extreme flood forecasts, IMERG_SDLRR decreased Mean Relative Error (MRE) by 0.29 and increased Qualified Rate (QR) by 53% compared to LSTM, and decreased MRE by 0.08 and increased QR by 6% compared to the best performance model using area-averaged precipitation as input. The utilization of IMERG or TMPA spatial information improved the accuracy of runoff forecasting. The accuracy evaluation of SPPs based on the results of spatiotemporal rainfall-runoff forecasts method was also demonstrated. The research is of great significance for developing runoff forecasting methods and optimizing water resources management.