RMSE Range Research Articles

Long Short-Term Memory vs Gated Recurrent Unit: A Literature Review on the Performance of Deep Learning Methods in Temperature Time Series Forecasting

Temperature forecasting is a crucial aspect of meteorology and climate change studies, but challenges arise due to the complexity of time series data involving seasonal patterns and long-term trends. Traditional methods often fall short in handling this variability, necessitating more advanced solutions to enhance prediction accuracy. LSTM and GRU models have emerged as promising alternatives for modeling temperature data. This study is a literature review comparing the effectiveness of LSTM and GRU based on previous research in temperature forecasting. The goal of this review is to evaluate the performance of both models using various evaluation metrics such as MSE, RMSE, and MAE to identify gaps in previous research and suggest improvements for future studies. The method involves a comprehensive analysis of previous studies using LSTM and GRU for temperature forecasting. Assessment is based on RMSE values and other metrics to compare the accuracy and consistency of both models across different conditions and temperature datasets. The analysis results show that LSTM has an RMSE range of 0.37 to 2.28. While LSTM demonstrates good performance in handling long-term dependencies, GRU provides more stable and accurate performance with an RMSE range of 0.03 to 2.00. This review underscores the importance of selecting the appropriate model based on data characteristics to improve the reliability of temperature forecasting.

Assessments of various precipitation product performances and disaster monitoring utilities over the Tibetan Plateau

The Tibetan Plateau, often referred to as Asia’s water tower, is a focal point for studying spatiotemporal changes in water resources amidst global warming. Precipitation is a crucial water resource for the Tibetan Plateau. Precipitation information holds significant importance in supporting research on the Tibetan Plateau. In this study, we estimate the performance and applicability of Climate Prediction Center Merged Analysis of Precipitation (CMAP), Integrated Multi-Satellite Retrievals for Global Precipitation Measurement (IMERG), Global Land Data Assimilation System (GLDAS), and Global Precipitation Climatology Project (GPCP) precipitation products for estimating precipitation and different disaster scenarios (including extreme precipitation, drought, and snow) across the Tibetan Plateau. Extreme precipitation and drought indexes are employed to describe extreme precipitation and drought conditions. We evaluated the performance of various precipitation products using daily precipitation time series from 2000 to 2014. Statistical metrics were used to estimate and compare the performances of different precipitation products. The results indicate that (1) Both CMAP and IMERG showed higher fitting degrees with gauge precipitation observations in daily precipitation. Probability of detection, False Alarm Ratio, and Critical Success Index values of CMAP and IMERG were approximately 0.42 to 0.72, 0.38 to 0.56, and 0.30 to 0.42, respectively. Different precipitation products presented higher daily average precipitation amount and frequency in southeastern Tibetan Plateau. (2) CMAP and GPCP precipitation products showed relatively great and poor performance, respectively, in predicting daily and monthly precipitation on the plateau. False alarms might have a notable impact on the accuracy of precipitation products. (3) Extreme precipitation amount could be better predicted by precipitation products. Extreme precipitation day could be badly predicted by precipitation products. Different precipitation products showed that the bias of drought estimation increased as the time scale increased. (4) GLDAS series products might have relatively better performance in simulating (main range of RMSE: 2.0–4.5) snowfall than rainfall and sleet in plateau. G-Noah demonstrated slightly better performance in simulating snowfall (main range of RMSE: 1.0–2.1) than rainfall (main range of RMSE: 2.0–3.8) and sleet (main range of RMSE: 1.5–3.8). This study’s findings contribute to understanding the performance variations among different precipitation products and identifying potential factors contributing to biases within these products. Additionally, the study sheds light on disaster characteristics and warning systems specific to the Tibetan Plateau.

A high-resolution coupled circulation-wave model for regional dynamic downscaling of water levels and wind waves in the western North Atlantic ocean

Reanalysis datasets provide temporally/spatially continuous data for sea level and wind wave climate studies but often lack the resolution to resolve the complex coastal physical processes. To address this issue, a high-resolution coupled circulation-wave model (ADCIRC + SWAN) is applied to dynamically downscale storm tides (tide + surge + wave setup) and waves in the western North Atlantic Ocean. The model is forced at its sea surface boundary with hourly surface pressure and wind fields, and its open boundary with water levels and direction-frequency wave spectrum from the ERA5 reanalysis dataset. The sensitivity of the model is evaluated for different combinations of internal tide energy conversion, quadratic friction coefficients, and wave physics packages. For the best model setup, the results show an RMSE range of 0.114 m–0.295 m for storm tide, 0.176 m–0.298 m for non-tidal residual, 0.143 m–0.39 m for significant wave height, and 0.42 s–1.04 s for mean wave period. Incorporating the direction-frequency wave spectrum as an open boundary condition improves the model's accuracy, reducing the RMSE for significant wave heights by up to 7% and for the mean wave period by up to 30% over a one-month simulation period.

Predicting daily recovery during long-term endurance training using machine learning analysis

PurposeThe aim of this study was to determine if machine learning models could predict the perceived morning recovery status (AM PRS) and daily change in heart rate variability (HRV change) of endurance athletes based on training, dietary intake, sleep, HRV, and subjective well-being measures.MethodsSelf-selected nutrition intake, exercise training, sleep habits, HRV, and subjective well-being of 43 endurance athletes ranging from professional to recreationally trained were monitored daily for 12 weeks (3572 days of tracking). Global and individualized models were constructed using machine learning techniques, with the single best algorithm chosen for each model. The model performance was compared with a baseline intercept-only model.ResultsPrediction error (root mean square error [RMSE]) was lower than baseline for the group models (11.8 vs. 14.1 and 0.22 vs. 0.29 for AM PRS and HRV change, respectively). At the individual level, prediction accuracy outperformed the baseline model but varied greatly across participants (RMSE range 5.5–23.6 and 0.05–0.44 for AM PRS and HRV change, respectively).ConclusionAt the group level, daily recovery measures can be predicted based on commonly measured variables, with a small subset of variables providing most of the predictive power. However, at the individual level, the key variables may vary, and additional data may be needed to improve the prediction accuracy.

Evaluating Performance of Multiple Machine Learning Models for Drought Monitoring: A Case Study of Typical Grassland in Inner Mongolia

Driven by continuously evolving precipitation shifts and temperature increases, the frequency and intensity of droughts have increased. There is an obvious need to accurately monitor drought. With the popularity of machine learning, many studies have attempted to use machine learning combined with multiple indicators to construct comprehensive drought monitoring models. This study tests four machine learning model frameworks, including random forest (RF), convolutional neural network (CNN), support vector regression (SVR), and BP neural network (BP), which were used to construct four comprehensive drought monitoring models. The accuracy and drought monitoring ability of the four models when simulating a well-documented Inner Mongolian grassland site were compared. The results show that the random forest model is the best among the four models. The R2 range of the test set is 0.44–0.79, the RMSE range is 0.44–0.72, and the fitting accuracy relationship could be described as RF > CNN > SVR ≈ BP. Correlation analysis between the fitting results of the four models and SPEI found that the correlation coefficient of RF from June to September was higher than that of the other three models, though we noted the correlation coefficient of CNN in May was slightly higher than that of RF (CNN = 0.79; RF = 0.78). Our results demonstrate that comprehensive drought monitoring indices developed from RF models are accurate, have high drought monitoring ability, and can achieve the same monitoring effect as SPEI. This study can provide new technical support for comprehensive regional drought monitoring.

Evaluation of the Potential of Using Machine Learning and the Savitzky–Golay Filter to Estimate the Daily Soil Temperature in Gully Regions of the Chinese Loess Plateau

Soil temperature directly affects the germination of seeds and the growth of crops. In order to accurately predict soil temperature, this study used RF and MLP to simulate shallow soil temperature, and then the shallow soil temperature with the best simulation effect will be used to predict the deep soil temperature. The models were forced by combinations of environmental factors, including daily air temperature (Tair), water vapor pressure (Pw), net radiation (Rn), and soil moisture (VWC), which were observed in the Hejiashan watershed on the Loess Plateau in China. The results showed that the accuracy of the model for predicting deep soil temperature proposed in this paper is higher than that of directly using environmental factors to predict deep soil temperature. In testing data, the range of MAE was 1.158–1.610 °C, the range of RMSE was 1.449–2.088 °C, the range of R2 was 0.665–0.928, and the range of KGE was 0.708–0.885 at different depths. The study not only provides a critical reference for predicting soil temperature but also helps people to better carry out agricultural production activities.

Enhancing short-term streamflow forecasting of extreme events: A wavelet-artificial neural network hybrid approach

Accurate short-term streamflow forecasting models are crucial for effective water resource management, enabling timely responses to extreme flood or drought events and mitigating potential socioeconomic damage. This study proposes robust hybrid Wavelet Artificial Neural Network (WANN) models for real-world hydrological applications. Two WANN variants, WANNone and WANNmulti, are proposed for short-term streamflow forecasting of extreme (high and low) flows at eight gauging stations within Brazil's Paraíba do Sul River basin. WANNone directly feeds both the original streamflow data and the decomposed components obtained through an À Trous wavelet transform into the ANN architecture. Conversely, WANNmulti utilizes separate ANNs for the original data, with the final streamflow estimate reconstructed via the inverse wavelet transform of the individual ANN outputs. The performance of these WANN models is then compared against conventional ANN models. In both approaches, Bayesian optimization is employed to fine-tune the hyperparameters within the ANN architecture. The WANN models achieved superior performance for 7-day streamflow forecasts compared to conventional ANN models. WANN models yielded high R2 values (>0.9) and low MAPE (4.8%–14.7%) within the expected RMSE range, demonstrating statistically significant improvements over ANN models (71% and 75% reduction in RMSE and MAPE, respectively, and 69% increase in R2). Further analysis revealed that WANNmulti models generally exhibited superior performance for low extreme flow predictions, while WANNone models achieved the highest accuracy for high extreme flows at most stations. WANN models' strong performance suggests their value for real-time flood warnings, enabling improved decision-making in areas like flood/drought mitigation and urban water planning.

Towards Affordable Precision Irrigation: An Experimental Comparison of Weather-Based and Soil Water Potential-Based Irrigation Using Low-Cost IoT-Tensiometers on Drip Irrigated Lettuce

Predictive weather-based models are widely used to schedule irrigation through the estimation of crop evapotranspiration. However, perceiving real-time crop water requirements remains a challenge. This research aims at field validating and exploiting a low-cost IoT soil moisture tensiometer prototype to consequently compare weather-based irrigation to soil water moisture-based irrigation in terms of yield and crop water productivity. The prototype is based on the ESP32 microcontroller and BMP180 barometric sensor. When compared to a mechanical tensiometer, the IoT prototype proved its accuracy, registering an average R2 equal to 0.8 and an RMSE range of 4.25–7.1 kPa. In a second step, the irrigation of a Romaine lettuce field (Lactuca sativa L.) cultivated under a drip system was managed according to two different scenarios: (1) using the data feed from the IoT tensiometers, irrigation was performed to keep the soil water potential between −15 and −25 kPa; (2) using the data provided by the in-situ weather station to estimate the crop water requirements. When comparing the yield, no significant difference was registered between the two scenarios. However, the water productivity was significantly higher, registering a 36.44% increment in scenario 1. The experiment highlights the water-saving potential achievable through real-time monitoring of soil moisture conditions. Since it is a low-cost device (82.20 USD), the introduced prototype facilitates deploying and managing a fleet of sensors for soil water potential live mapping.

COMPARATIVE ANALYSIS BETWEEN GPS-BASED FIELD DIGITAL TERRAIN MODEL WITH TANDEM- X 90M, ASTER AND SRTM DIGITAL ELEVATION MODELS

This study compared the vertical accuracy of existing satellite’s Digital Elevation Model (STRM, ASTER) and the latest satellite remote sensing height data set TanDEM-90m. ASTER and STRM have vertical accuracies of ± 20m and ± 16m, respectively. TanDEMX-90m was processed before use, and the RMSE range of 19-20m was confirmed for TANDEM X 90m. The result of the study shows that ASTER DEM performed better than the rest of the global digital elevation models. The SRTM error of 4m between the first and second locations may be due to systematic error due to slightly different versions of SRTM used for processing. The TANDEM X 90m had the same resolution but performed poorer in positional fitness. The study found that digital elevation models have different resolutions and accuracy levels. Tandem-X90m has a resolution of up to 3 meters, while Aster and SRTM have 30 meters. GPS-based field digital terrain models provide higher accuracy, while Tandem-X90m has a 10 cm accuracy. The models cover a limited area, making them suitable for high-resolution elevation data. The data collection methods used by the models, such as SAR sensors, also affect their accuracy and resolution. Based on the above comparative analysis, it can be recommended that the GPS-based field digital terrain model is suitable for high-accuracy and high-resolution elevation data in a limited area. Tandem-X90m is suitable for high-resolution elevation data in a limited area but is expensive. Aster and SRTM are suitable for global-scale studies but have lower accuracy and resolution.

Estimating Blood Glucose Levels Using Machine Learning Models with Non-Invasive Wearable Device Data.

In 2019 alone, Diabetes Mellitus impacted 463 million individuals worldwide. Blood glucose levels (BGL) are often monitored via invasive techniques as part of routine protocols. Recently, AI-based approaches have shown the ability to predict BGL using data acquired by non-invasive Wearable Devices (WDs), therefore improving diabetes monitoring and treatment. It is crucial to study the relationships between non-invasive WD features and markers of glycemic health. Therefore, this study aimed to investigate accuracy of linear and non-linear models in estimating BGL. A dataset containing digital metrics as well as diabetic status collected using traditional means was used. Data consisted of 13 participants data collected from WDs, these participants were divided in two groups young, and Adult Our experimental design included Data Collection, Feature Engineering, ML model selection/development, and reporting evaluation of metrics. The study showed that linear and non-linear models both have high accuracy in estimating BGL using WD data (RMSE range: 0.181 to 0.271, MAE range: 0.093 to 0.142). We provide further evidence of the feasibility of using commercially available WDs for the purpose of BGL estimation amongst diabetics when using Machine learning approaches.

Downscaling wind speed based on coupled environmental factors and machine learning

AbstractWind speed changes impact society and have important implications for climate change studies. Thus, high‐resolution and high‐quality wind speed datasets are necessary for environmental monitoring and ecosystem research. However, there is no complete set of high spatial and temporal resolution wind speed datasets for China. Additionally, it is extremely challenging to produce wind speed data at high spatial and temporal resolution for large‐scale regions with diverse climate types and complex topographies, such as China. In this study, we used multisource remote sensing images, obtained data on various environmental factors through the Google Earth Engine and Evapotranspiration (ET) Watch Cloud platforms, and combined machine learning algorithms to downscale the ERA5 reanalysis wind speed data, and finally obtained the daily wind speed datasets with 1 km spatial resolution for China in 2015. To verify the accuracy of the model and data products, we selected several metrics to evaluate in conjunction with the actual site observed data. The results show that the multifactor combination model of artificial neural network combining land surface temperature, sunshine durations and roughness factors outperforms a single‐factor combination model, and the results were in good agreement with the original data (R2 of 0.95 and RMSE of 0.40 m·s−1). The final wind speed data products were also in good agreement with the observed meteorological data (R2 range of 0.86–0.95 and RMSE range of 0.33–0.44 m·s−1); moreover, the accuracy and precision were greatly improved over the original data. This study provided a dataset that has potential applications in future climate change and ecosystem studies.

Method of Validating Satellite Surface Reflectance Product Using Empirical Line Method

Atmospherically corrected surface reflectance (SR) products are used for reliable monitoring of land surfaces and are the standard products of Landsat sensors. Due to increased demand for SR products, a need exists to verify that the L2C2 (Level-2 Collection-2) SR products are precise and accurate. The Level-2 Collection 2 (L2C2) SR Product has processed satellite imagery data that correct for atmospheric effects such as absorption and scattering, providing a more accurate representation of Earth’s surface. The validation of SR products using ground truth measurement is essential. This study aims to develop and evaluate a validation methodology for satellite SR products. Thus, the Empirical Line Method (ELM) is used here for atmospheric validation of remotely sensed data. Validation is performed using the SR derived from ELM tied to ground truth measurement. Absolute surface reflectance models of Algodones Dunes and the Salton Sea located in North America Sonoran Desert are developed to extend the temporally limited ground truth measurements. This model can give ground truth reflectance in any time frame independent of time constraints. The result of the absolute surface reflectance model of Algodones Dunes indicates that the model predicts the response of Algodones Dunes with an average accuracy of 0.0041 and precision of 0.0063 and gives ground measurements across all multispectral between 350 and 2500 nm. For the Salton Sea, the model predicts the response of the Salton Sea with mean absolute error (MAE) of 0.0035 and gives ground measurements across all multispectral between 350 and 2500 nm. The ELM generates atmospheric coefficients (gain and bias), which are applied to an image to obtain SR. Validation results indicated that for L9-OLI-2, L8-OLI, and L5-TM-SR products, the RMSE range is 0.0019 to 0.0106, 0.0019 to 0.0148 and 0.0026 to 0.0045 reflectance unit, respectively, and accuracy is within 0.0038, 0.0022, and 0.0055 reflectance unit across all spectral bands of L9, L8, and L5, respectively. On average, the validation result showed a strong linear relation between the L2C2 SR products and ELM SR within 0.5 to 1 reflectance units. These results demonstrate the high accuracy and reliability of the L2C2 SR product, providing valuable information for a wide range of remote sensing applications, including land cover and land use mapping, vegetation monitoring, and climate change studies.

Development of Psoriasis Assessment Tools Among Patients in the CorEvitas Psoriasis Registry.

Dermatologists would benefit from an easy to use psoriasis severity assessment tool in the clinic. To develop psoriasis assessment tools to predict PASI and Dermatology Life Quality Index (DLQI) using simple measures typically collected in clinical practice. Data included 33605 dermatology visits among plaque psoriasis patients enrolled in the CorEvitas Psoriasis Registry (4/15/15-7/11/20). Performance (adjusted coefficient of determination [R2 adj], root mean square error [RMSE]) in predicting PASI and DLQI was assessed for 16 different linear regression models (specified a priori based on combinations of BSA, Investigator's Global Assessment [IGA], itch, skin pain, patient global assessment, age, sex, BMI, comorbidity index, prior biologic use), and 2 stepwise selection models and 1 elastic net model based on 56 available variables. For each prediction model, concordance (sensitivity, specificity) of predicted PASI75, PASI90 and DLQI 0/1 with observed values was evaluated. Mean (SD) age, BSA, and PASI were 51 (14) years, 6 (11), and 4 (6), respectively; 46% were women, and 87% were biologic experienced. A model predicting PASI using BSA plus IGA performed best among a priori specified models (R2 adj = .72, RMSE = 2.93) and only marginally worse than models including additional variables (R2 adj range .64-.74, RMSE range 2.82-3.36). Models including IGA had the best concordance between predicted and observed PASI75 (sensitivity range 83-85%, specificity range 88-91%) and PASI90 (sensitivity range 76-82%, specificity range 94-98%). DLQI prediction was limited. An assessment tool for psoriasis including BSA and IGA may be an ideal option to predict PASI in a clinic setting.

Estimation of Evapotranspiration in the Desert–Oasis Transition Zone Using the Water Balance Method and Groundwater Level Fluctuation Method—Taking the Haloxylon ammodendron Forest at the Edge of the Gurbantunggut Desert as an Example

Shallow groundwater is an important water source for Haloxylon ammodendron (H. ammodendron). The accurate estimation of evapotranspiration (ETg) from groundwater is of great significance for the water cycle and the maintenance of ecological stability. Using a combination of the water balance method and the groundwater level fluctuation method (WTF), the water balance components (precipitation, soil moisture, groundwater depth, and Bowen ratio meteorological data) in the desert–oasis transition zone were continuously monitored from 2015 to 2018 and the ETg was estimated The results showed that the closed degree of Bowen specific energy after data screening was higher, and the annual actual evapotranspiration (ETa) value could be reliably calculated at 260.87 mm. As the main contributor to water consumption in the growing season, latent heat accounted for 70.16~91.86% of the energy balance. Precipitation had no significant impact on water consumption for H. ammodendron vegetation growth, and the precipitation in the main growing season accounted for 59.44% of the ETa. The groundwater depth in the study area decreased yearly and had a significant impact on the growth of H. ammodendron vegetation. Although the groundwater depth in the study area was greater than 9 m, the ETg, as an important part of the water balance, was found to participate in the evapotranspiration process brought about by H. ammodendron due to the strong root system and supporting capillary water in the soil. The actual evapotranspiration ETa for H. ammodendron in the main growing season was 244.32 mm, and the contribution rate for ETg was as high as 74.78% or approximately 182.35 mm. After the ETg was verified using the water balance method and WTF, R was greater than 0.96, the RMSE range was 1.5931~4.5706, the bias range was −0.15~0.11, and the IOA value was greater than 0.95. The accuracy of the estimation model was high, and the results were relatively accurate. The model can be applied in the desert–oasis transition zone to obtain accurate ETg estimations and provide theoretical guidance and a scientific basis for local water resource management and ecological protection.

Simulation and assessment of water supply network for specified districts at Najaf Governorate

AbstractThis study aims to simulate and assess the hydraulic characteristics and residual chlorine in the water supply network of a selected area in Al-Najaf City using WaterGEMS software. Field and laboratory work were conducted to measure the pressure heads and velocities, and water was sampled from different sites in the network and then tested to estimate chlorine residual. Records and field measurements were utilized to validate WaterGEMS software. Good agreement was obtained between the observed and predicted values of pressure with RMSE range between 0.09–0.17 and 0.08–0.09 for chlorine residual. The results of the analysis of water distribution systems (WDS) during maximum demand hours showed that the pumps unit capability cannot cover the high water demand during that time and resulted in a loss of pressure values, which were ranged between 0.2 and 2.1 bar. Moreover, the simulated results of the residual chlorine levels were within the permissible limits of 0.4–0.7 ppm, in different locations in the network. Providing good quality and adequate water supply is an important component for human life development. Modeling WDS is an efficient method of gaining a true understanding of the functioning of the network and determining the factors and conditions affecting the performance of the network.

Predicting riverine fish production using empirical models and the metabolic theory of ecology

AbstractFish production integrates many different measures of community performance, such as abundance, biomass, growth, and reproduction, into one valuable quantitative metric but requires resource intensive data for empirical estimation. While published empirical models and the metabolic theory of ecology (MTE) represent alternative methods to estimate fish production, few studies have focused on productivity models for stream fish assemblages. The goal of our study was to determine whether existing empirical models and elements of the metabolic theory of ecology can reliably estimate stream fish productivity. We used production estimates from the literature (n = 107) to parameterize models based on the metabolic theory of ecology and new estimates of stream fish production from North America (n = 78) to compare and validate all models. Using major axis regression, we determined that while all models had strongly correlated production estimates relative to the observed values (r2 range: [0.496, 0.815]), not all the models produced accurate estimates. The MTE model with the temperature component had a poorer predictive performance (RMSE = 0.502) relative to models based solely on allometric scaling (RMSE range: [0.299, 0.380]). We conclude that standard production models can generate relative estimates of production using general fish sample data, however, the accuracy and precision of the estimates can vary among the models. Our study highlights the need for productivity estimates for stream fish assemblages from different geographic regions, to test empirical models with novel datasets, and for further investigation of temperature effects on fish productivity.

Estimation of ground reaction force waveforms during fixed pace running outside the laboratory.

In laboratory experiments, biomechanical data collections with wearable technologies and machine learning have been promising. Despite the development of lightweight portable sensors and algorithms for the identification of gait events and estimation of kinetic waveforms, machine learning models have yet to be used to full potential. We propose the use of a Long Short Term Memory network to map inertial data to ground reaction force data gathered in a semi-uncontrolled environment. Fifteen healthy runners were recruited for this study, with varied running experience: novice to highly trained runners (<15 min 5 km race), and ages ranging from 18 to 64 years old. Force sensing insoles were used to measure normal foot-shoe forces, providing the standard for identification of gait events and measurement of kinetic waveforms. Three inertial measurement units (IMUs) were mounted to each participant, two bilaterally on the dorsal aspect of the foot and one clipped to the back of each participant's waistband, approximating their sacrum. Data input into the Long Short Term Memory network were from the three IMUs and output were estimated kinetic waveforms, compared against the standard of the force sensing insoles. The range of RMSE for each stance phase was from 0.189-0.288 BW, which is similar to multiple previous studies. Estimation of foot contact had an r 2 = 0.795. Estimation of kinetic variables varied, with peak force presenting the best output with an r 2 = 0.614. In conclusion, we have shown that at controlled paces over level ground a Long Short Term Memory network can estimate 4 s temporal windows of ground reaction force data across a range of running speeds.

Refinement of global gridded ray-traced Zenith tropospheric delay over Nigeria based on local GNSS network observations

• Application of deep structured supervised neural network modelling technique. • Data derived from the local GNSS observation within the study area. • A refinement model is developed to predict ZTD parameters at user-defined points. • Improvement in the accuracy of the local refine VMF1-ZTD estimates. Recent studies on the application of Zenith Tropospheric Delay (ZTD) in meteorology and geodesy have shown improved accuracy with the use of locally estimated ZTD compared to Global ZTD. This paper focuses on the refinement of global gridded ray-traced Vienna Mapping Functions 1 (VMF1) ZTD over Nigeria, using ZTD estimates derived from local GNSS Continuously Operating Reference Stations (CORS) observations. Three (3) years of GNSS observations collected at 13 CORS located within the study area were post-processed using GNSS Analysis and Processing Software (GAPS) to derive the local GNSS-ZTD estimates. The global VMF1-ZTD were interpolated at the locations of the local GNSS-CORS. An optimal deep structured supervised Neural Network (NN) was derived by training feedforward networks using the Levenberg-Marquardt backpropagation algorithm, with spatiotemporal variables and VMF1-ZTD as inputs and the GNSS-ZTD as the target. The outputs from the NN model were assessed using internal and external data. Results from internal data evaluations indicated a robust and accurate model output with optimum network performances indicating MSE of ∼7.22 × 10 −6 m and ∼4.56 × 10 −4 m for hydrostatic and wet ZTD networks respectively. The external data evaluations performed using the International GNSS Services (IGS) Zenith Path Delay (ZPD), indicated an RMSE range of ∼0.041 m to 0.095 m for global-VMF1 and ∼0.030 m to 0.037 m for the Local Refine (LR)-VMF1. The results suggest that the application of ZTD estimates derived from local GNSS observations in the refinement of global gridded VMF1-ZTD yields an improved accuracy over Nigeria. Consequently, the LR-VMF1 ZTD estimates are recommended for improved accuracy, required for geodetic and meteorological applications in and around Nigeria :

Performance of artificial intelligence models in estimating blood glucose level among diabetic patients using non-invasive wearable device data

IntroductionDiabetes Mellitus (DM) is characterized by impaired ability to metabolize glucose for use in cells for energy, resulting in high blood sugar (hyperglycemia). DM impacted 463 million individuals worldwide in 2019, with over four million fatalities documented. Blood glucose levels (BGL) are usually measured, as standard protocols, through invasive procedures. Recently, Artificial Intelligence (AI) based techniques have demonstrated the potential to estimate BGL using data collected by non-invasive Wearable Devices (WDs), thereby, facilitating monitoring and management of diabetics. One of the key aspects of WDs with machine learning (ML) algorithms is to find specific data signatures, called Digital biomarkers, that can be used in classification or gaging the extent of the underlying condition. The use of such biomarkers to monitor glycemic events represents a major shift in technology for self-monitoring and developing digital biomarkers using non-invasive WDs. To do this, it is necessary to investigate the correlations between characteristics acquired from non-invasive WDs and indicators of glycemic health; furthermore, much work is needed to validate accuracy. Research Design & MethodsThe study aimed to investigate performance of AI models in estimating BGL among diabetic patients using non-invasive wearable devices data An open-source dataset was used which provided BGL readings, diabetic status (Diabetic or non-diabetic), heart rate, Blood oxygen level (SPO2), Diastolic Blood pressure, Systolic Blood Pressure, Body temperature, Sweating, and Shivering for 13 participants by age group taken from WDs. Our experimental design included Data Collection, Feature Engineering, ML model selection/development, and reporting evaluation of metrics. ResultsWe were able to estimate with high accuracy (RMSE range: 0.099 to 0.197) the relationship between glycemic metrics and features that can be derived from non-invasive WDs when utilizing AI models. ConclusionWe provide further evidence of the feasibility of using commercially available WDs for the purpose of BGL estimation amongst diabetics.

A framework for predicting the carbonation depth of concrete incorporating fly ash based on a least squares support vector machine and metaheuristic algorithms

The carbonation of concrete structures is becoming a more serious problem as the global CO2 level increases, and poses a great challenge to concrete durability research. In this study, a hybrid prediction framework on the basis of the least squares support vector machine (LSSVM) and metaheuristic algorithms is proposed to accurately predict the carbonation depth of concrete incorporating fly ash. The fly ash replacement level, cement content, water-to-binder ratio, CO2 content, relative humidity of environmental and exposure time are treated as input variables of the framework. A database containing 500 sets of samples is created for model building and comparison. The results show that five hybrid models have superior accuracy in the prediction of concrete carbonation depth, where the R2, RMSE, MAE and VAF range from 0.9683 to 0.9819, 1.7309–2.2936 mm, 1.2255–1.5478 mm, and 96.8317 to 98.1995, respectively. In addition, the proposed prediction framework has better performance (i.e., smaller RMSE) compared to reported empirical prediction models. The Sobol' global sensitivity index analysis results indicate that the water-to-binder ratio, exposure time and cement content have higher sensitivity to predict the carbonation depth. The effect of dataset division ratio on the models (an aspect that has not been investigated in previous research) is further explored in this study, and the optimal dataset division ratio for each model is determined.