Willmott Index Research Articles

The Superiority of Data-Driven Techniques for Estimation of Daily Pan Evaporation

In the present study, estimating pan evaporation (Epan) was evaluated based on different input parameters: maximum and minimum temperatures, relative humidity, wind speed, and bright sunshine hours. The techniques used for estimating Epan were the artificial neural network (ANN), wavelet-based ANN (WANN), radial function-based support vector machine (SVM-RF), linear function-based SVM (SVM-LF), and multi-linear regression (MLR) models. The proposed models were trained and tested in three different scenarios (Scenario 1, Scenario 2, and Scenario 3) utilizing different percentages of data points. Scenario 1 includes 60%: 40%, Scenario 2 includes 70%: 30%, and Scenario 3 includes 80%: 20% accounting for the training and testing dataset, respectively. The various statistical tools such as Pearson’s correlation coefficient (PCC), root mean square error (RMSE), Nash–Sutcliffe efficiency (NSE), and Willmott Index (WI) were used to evaluate the performance of the models. The graphical representation, such as a line diagram, scatter plot, and the Taylor diagram, were also used to evaluate the proposed model’s performance. The model results showed that the SVM-RF model’s performance is superior to other proposed models in all three scenarios. The most accurate values of PCC, RMSE, NSE, and WI were found to be 0.607, 1.349, 0.183, and 0.749, respectively, for the SVM-RF model during Scenario 1 (60%: 40% training: testing) among all scenarios. This showed that with an increase in the sample set for training, the testing data would show a less accurate modeled result. Thus, the evolved models produce comparatively better outcomes and foster decision-making for water managers and planners.

Comparison and modification of measurement and simulation techniques for estimating Tmrt in summer and winter in a severely cold region

Mean Radiant Temperature (Tmrt) is a crucial indicator to assess urban radiant and thermal environments. In this study, Tmrt derived from the globe thermometer method and three simulation models (SOLWEIG, RayMan, and ENVI-met) were analyzed and modified based on the standard six-directional method (Tmrt(six)). Performances of the techniques were compared for two contrasting seasons, summer and winter, and at sites with different degrees of openness. The accuracies of each method were assessed via patterns in time series, coefficient of determination, root mean square error, and Willmott index of agreement. Results showed that globe thermometer method was more suitable for estimating Tmrt in summer than winter, and it showed better performance in more open spaces. The simulation inaccuracies were mainly reflected in winter. Comparatively, SOLWEIG exhibited the best performance in correlation and discrepancy; RayMan presented an advantage in describing the trend consistency of Tmrt than ENVI-met; ENVI-met simulated Tmrt closer to Tmrt(six) than RayMan. Principle-based causes of the techniques’ inaccuracies were analyzed. Correlation and cluster analyses were applied for the globe thermometer method, and global radiation showed the strongest correlation and nearest distance with globe temperature, followed by air temperature. Sensitivity and components analyses were utilized for the simulation models involving isotropic and anisotropic issues considering direct and diffuse radiation, solar position, and surrounding environment. Based on the characteristics of different techniques, multiple nonlinear (introducing equivalent globe temperature) and unary linear regressions were utilized to modify globe thermometer and simulation results, respectively. Both modifications significantly improved Tmrt determination accuracies.

The overall framework design of automatic logistics system using a hybrid ANN-PSO model

Automated logistic systems are commonly applied in the enterprise logistic processes while allowing to improve the reliability and efficiency of logistics procedures with computer simulation. The logistics system focuses on realizing the space and time efficiency of material and understanding the links of logistics to achieve the optimum economic effect. Therefore, in this study, 150 data were collected from 20 hospitals in Malaysia to adopt the automated logistic system under simulation of the computer to decrease the costs of hospitals and increase the serviceability (output). Since obtaining this information for the automated logistic systems is not easy and has a high cost in a traditional way, this study was performed by the integration of artificial neural network (ANN) and particle swarm optimization (PSO) as the ANN-PSO. The input data that were obtained from the processing of automated logistic system in hospitals verified by using a hybrid of ANN-PSO. Later, the results were measured with five regression indicators of determination coefficient (R2), root mean square (RMSE), Pearson correlation coefficient (r), Nash–Sutcliffe model efficiency coefficient (NSE) and Willmott’s index (WI). Considering the results of RMSE, RSQR and r in both testing and training phases, the good performance of ANN-PSO in determining the effectiveness of applying automated logistic system using computer simulation method in hospital management, in terms of raising its serviceability and reducing the costs was proved. As a result, it was shown that ANN-PSO can successfully determine the effectiveness of using an automated logistic system under computer simulation method in hospital management.

Prediction of meteorological drought by using hybrid support vector regression optimized with HHO versus PSO algorithms.

Drought is considered one of the costliest natural disasters that result in water scarcity and crop damage almost every year. Drought monitoring and forecasting are essential for the efficient management of water resources and sustainability in agriculture. However, the design of a consistent drought prediction model based on the dynamic relationship of the drought index with its antecedent values remains a challenging task. In the present research, the SVR (support vector regression) model was hybridized with two different optimization algorithms namely; Particle Swarm Optimization (PSO) and Harris Hawks Optimization (HHO) for reliable prediction of effective drought index (EDI) 1 month ahead, at different locations of Uttarakhand State of India. The inputs of the models were selected through partial autocorrelation function (PACF) analysis. The output produced by the SVR-HHO and SVR-PSO models was compared with the EDI estimated from observed data using five statistical indicators, i.e., RMSE (Root Mean Square Error), MAE (Mean Absolute Error), COC (Coefficient of Correlation), NSE (Nash-Sutcliffe Efficiency), WI (Willmott Index), and graphical inspection of radar-chart, time-variation plot, box-whisker plot, and Taylor diagram. Appraisal of results indicates that the SVR-HHO model (RMSE = 0.535-0.965, MAE = 0.363-0.622, NSE = 0.558-0.860, COC = 0.760-0.930, and WI = 0.862-0.959) outperformed the SVR-PSO model (RMSE = 0.546-0.967, MAE = 0.372-0.625, NSE = 0.556-0.855, COC = 0.758-0.929, and WI = 0.861-0.956) in predicting EDI. Visual inspection of model performances also showed a better performance of SVR-HHO compared to SVR-PSO in replicating the median, inter-quartile range, spread, and pattern of the EDI estimated from observed rainfall. The results indicate that the hybrid SVR-HHO approach can be utilized for reliable EDI predictions in the study area.

Assessment of spectral indexes for estimating soil water content in the Brazilian Pantanal

The Pantanal is the largest floodplain of the globe occupying 138,183 km2 in Brazil. The fundamental ecological factor of interaction between the Pantanal ecosystems is the flooding regime. Connected to properties of the soil-plant-atmosphere system, knowledge on the soil water content becomes increasingly necessary. The high temporal and spatial variability of water content in the soil caused by the vast heterogeneity of soil texture, vegetation, topography and climate makes it a difficult physical variable to be measured. However, its spatial and temporal variability can be determined by recent modern techniques of remote sensing based on data obtained by microwave or infrared sensors. Thus, the aim of this study was to evaluate the accuracy of vegetation and soil water indexes through satellite images from Landsat 5 in the northern Brazilian Pantanal. The study was conducted in a pasture, experimental site in the Northern Pantanal in Mato Grosso state. Soil moisture was measured using a TDR probe installed at 10 cm depth in the period from 2009 to 2011. For comparison, spectral indexes and the surface temperature provided by Geological Survey (USGS) were used, these indexes are derived from bands ratios of satellite reflectance products Landsat 5 TM. The data evaluation was performed using some indicators: accuracy - Willmott index, Root Mean Square Error and the Mean Absolute Error. This study demonstrated that the application of remote sensing in the management of water resources is very promising. The indexes correlated with soil moisture measurement. Among the soil water indexes the NBR-2 showed related to soil moisture measurement. For both types of soils EVI had the highest determination coefficient, lowest errors and highest Willmott’s index of agreement.

Fine-resolution precipitation mapping over Syria using local regression and spatial interpolation

Annual precipitation at 1 km2 spatial resolution have been produced over Syria for a referenced period of 1975–2010. The observations from 410 rain-gauges were interpolated over a regular grid by applying multivariate regression models (PSMRM) and local equations for sub-regions of the study area. This statistical method aims to model the influences of the essential geographical and topographical climatic factors, such as longitude, latitude, elevation, slopes, and aspects on the precipitation field in multiple local regions. The PSMRM is composed of two steps, (i) a potential surface of precipitation is calculated through multi-linear local regressions based on geographical and topographical information, then (ii) a kriging and IDW interpolation is applied to adjust the potential surface so as to better fit the station residuals (i.e. the difference between the observed values and the predicted values which are obtained from PSMRM). Ultimately, the models' accuracy was evaluated by 43 stations. The PSRMR-IDW-3 is found to be superior to all other models; the value of RMSE was 92.5 mm and the Nash-Sutcliffe efficiency NSE was 0.9187, while the Willmott index of agreement was 0.9808. In contrast, the PSMRM-OK-EXP was only superior to other models with the least mean absolute error (MEA) and the mean absolute percentage error (MAPE); the difference was 64.07 mm, i.e. 11.44%. However, all the proposed models were shown to be highly efficient compared to global models and can be considered an appropriate alternative to studying precipitation variability spatially over Syria.

Comparison of meteorological data, related to reference evapotranspiration, from conventional and automatic stations in the Sertão and Agreste regions of Pernambuco, Brazil

The objective was to perform a comparative study of the meteorological elements data that most cause changes in the reference Evapotranspiration (ETo, mm) and its own value, of automatic weather stations AWS and conventional weather stations CWS of the Sertão and Agreste regions of Pernambuco State. The ETo was calculated on a daily scale using the standard method proposed by the Food and Agriculture Organization (FAO), Penman-Monteith (FAO-56). The ETo information obtained from AWS data can be used to update the weather database of stations, since there is a good relationship between the ETo data obtained from CWS and AWS, statistically determined by the Willmott's concordance index (d > 0.7). The observed variations in the weather elements: air temperature, relative humidity, wind speed, and global solar radiation have not caused significant changes in the ETo calculation.

Investigating the reliability of machine learning algorithms as a sustainable tool for total suspended solid prediction

This research studies the implementation of artificial neural networks (ANN) in predicting the concentration of total suspended solids (TSS) for the Fei Tsui reservoir in Taiwan. The prediction model developed in this study is designed to be used for monitoring the water quality in the Fei Tsui reservoir. High concentrations of total suspended solids (TSS) have been a crucial problem in the Fei Tsui reservoir for decades. As the Fei Tsui reservoir is a primary water source for Taipei City, this issue impacts the drinking water supply for the city due to etherification problems in the reservoir. 10-year average monthly records and 13-year average annual records have been collected for 26 parameters and correlated with the TSS concentrations to determine the parameters that have a strong relationship with the TSS concentrations. The parameters that were shown to have a strong correlation with the TSS concentration are the trophic state index (TSI), nitrate (NO3) concentration, total phosphorous (TP) concentration, iron concentration (IRON), and turbidity. Linear regression was used to develop the model that estimates the TSS concentration in the Fei Tsui Reservoir. The results show that model 3, a three-layer ANN model that uses three-input parameters namely NO3 concentration, TP concentration, and turbidity, with five neurons, to predict the output parameter which is TSS concentration, produces the highest coefficient of determination (R2) and Willmott Index (WI), which are 0.9589 and 0.9933 respectively, and the lowest root mean square error, which is 0.4753. Based on these performance criteria, model 3 is concluded as the best model to predict TSS concentrations in this study.

Comprehensive evaluation of machine learning models for suspended sediment load inflow prediction in a reservoir

Suspended sediment load (SSL) flowing into a reservoir contributes to the overall safety of dam. Owing to the complexity and stochastic nature of sedimentation, accurate prediction of reservoir SSL inflow is still challenging. Moreover, research and application of machine learning (ML) techniques for reservoir sedimentation are still deficient. A comprehensive evaluation of six ML models for a reservoir SSL inflow prediction was performed in this study. ML techniques including artificial neural network (ANN), adaptive neuro-fuzzy inference system (ANFIS), radial basis function neural network (RBFNN), support vector machine (SVM), genetic programming (GP), and deep learning (DL) were applied to develop predictive models of daily SSL inflow at Sangju Weir, South Korea. Significant input vectors for each model were selected with streamflow, water temperature, water stage, reservoir outflow for different time lags. Model performances were evaluated using various statistical indices including the coefficient of determination (R2), mean absolute error (MAE), percentage of bias (PBIAS), Willmott index (WI), Nash–Sutcliffe efficiency (NSE), root mean square error (RMSE), and Pearson correlation coefficient (PCC). The best input combinations were found to be unique for each ML model, but all six models performed reasonably well for SSL inflow predictions. ANN model outperformed other models with R2 = 0.821, MAE = 4.244 tons/day, PBIAS = 0.055, WI = 0.891, NSE = 0.991, RMSE = 11.692 tons/day, PCC = 0.826. The models were ranked based on their SSL prediction capabilities as ANN > ANFIS > DL > RBFNN > SVM > GP from best to worst. The findings are expected to be useful for future dam safety and risk assessment, and for achieving sustainability of reservoir operation through comprehensive sediment management.

DINÂMICA E ESTIMATIVAS DA TEMPERATURA E UMIDADE RELATIVA DO AR EM VIVEIROS PROTEGIDOS COM DIFERENTES SOMBREAMENTOS

Objetivou-se descrever a dinâmica diária da temperatura (Tar) e umidade relativa do ar (UR) em ambientes protegidos com diferentes telas poliefinas de sombreamento, na região de transição Cerrado-Amazônia do Mato Grosso; ademais, foram avaliados os usos de regressões de estimativa de Tar e UR nos ambientes sombreados com base nas mesmas variáveis medidas em pleno sol. As avaliações micrometeorológicas foram realizadas em viveiros florestais modulares suspensos, alinhados no sentido Leste-Oeste, entre junho de 2017 e abril de 2019, sob telas pretas com níveis crescentes de sombreamento (35, 50, 65 e 80%) e coloridas/espectrais (termorefletora, vermelha, azul e verde, todas com 50% de sombreamento). Os dados do monitoramento micrometeorológico foram agrupados em função das estações hídricas regionais (seca, seca-chuvosa, chuvosa e chuvosa-seca), com separação da base de dados por decêndios. Foram empregados 70 e 30% dos dados para geração e validação das regressões, em cada agrupamento de dados, respectivamente. Na avaliação do desempenho estatístico das regressões foram empregados os indicadores estatísticos: coeficiente de determinação (R2), erro absoluto médio (MBE), raiz quadrada do erro quadrático médio (RMSE) e índice de Willmott (d). Houve dinâmica similar de Tar e UR entre a condição de pleno sol e as telas poliefinas pretas e coloridas ao longo do dia e do ano; porém, com aumento expressivo nos valores médios da Tar e UR nesses ambientes protegidos. Os valores de R² foram satisfatórios, demonstrando que mais de 60% da variável dependente (Tar nas telas de sombreamento) está relacionada à variável independente (Tar na condição de pleno sol). O d variou de 0,96 a 0,99, indicando que as regressões de estimativas da Tar e UR ajustadas apresentam desempenho satisfatório para todas as estações hídricas regional nos ambientes sombreados. Palavras-chave: ambientes protegidos; micrometeorologia; indicadores estatísticos; transição Cerrado-Amazônia. Dynamics and estimates of air temperature and relative humidity in nurseries protected with different shading ABSTRACT: The objective was to describe the daily dynamics of temperature (Tar) and relative humidity (RH) in protected environments with different polyolefin shading screens in transition region of Cerrado-Amazonia of Mato Grosso; in addition, the uses regressions of estimation of Tar and RH in shaded environments based on the same variables measured in full sun were evaluated. Micrometeorological assessments were performed in suspended modules forest nurseries, aligned to the East-West direction, between June 2017 and April 2019, under black screens with increasing levels of shading (35, 50, 65 and 80%) and colored / spectral (thermo-reflector, red, blue and green, all with 50% shading). The data of the micrometeorological monitoring were grouped according to the regional water stations (dry, dry-rainy, rainy and rainy-dry), with separation of the database for ten years. 70 and 30% of the data were used to generation and validation the regressions, in each data group, respectively. In the Statistical performance evaluation of the regressions were used the statistical indicators: coefficient of determination (R2), the mean error (MBE), root mean square error (RMSE) and Willmott's index of adjustment (d). There was similar dynamic of Tar and UR between the condition of full sun and the black and colored polyolefin screens throughout the day and year; however, with a significant increase in the mean values of Tar and UR in these protected environments. The R² values were satisfactory, showing that more than 60% of the dependent variable (Tar in the shading screens) is related to the independent variable (Tar in full sun). The d ranged from 0.96 to 0.99, indicating that the adjusted regressions of Tar and UR present satisfactory performance for all regional water stations in shaded environments. Palavras-chave: protected environments; micrometeorology; statistical indicators; Cerrado-Amazônia transition.

Estimating the daily nutritive value of bermudagrass for grazing livestock

ABSTRACT Objective Our objective was to develop models to estimate the CP, NDF, lignin, and ash concentrations of bermudagrass [Cynodondactylon (L.) Pers.] forage daily throughout the grazing season and use them to estimate TDN daily. We hypothesized that the CP, NDF, lignin, ash, and TDN concentrations might change daily. Materials and Methods Coastal bermudagrass forage CP, NDF, lignin, and ash concentrations spanning several years were obtained from Overton, Texas. The forage samples used comprised hand-plucked plant parts intended to estimate animal-selected diets. We developed an empirical model for each nutritive variable with the day of the year (DOY) as a predictor. By incorporating these models into a pre-existing summative equation, we estimated the daily values of TDN. These nutritive value models were evaluated using the Willmott index, modeling efficiency, root mean squared error, and R2 as the measures of fit. Results and Discussion Bermudagrass CP concentration had a negative, curvilinear association with DOY, whereas lignin and NDF concentrations were positively, curvilinearly related to DOY. The ash concentration had a negative, linear association with DOY. The likely causal factors for these relationships were plant maturity and weather conditions, especially temperature. The Willmott index, modeling efficiency, and R2 values were greater than 0.87, 0.64, and 0.65, respectively. The root mean squared error and percent error values were less than 3.1 and 11%, respectively. These values indicated that the nutritive value models were reliable. Implications and Applications We showed that the CP, NDF, lignin, ash, and TDN values of bermudagrass forage changed daily. By incorporating these trends into a nutritive value prediction system, accurate estimates of daily TDN could be made. These nutritive value models could be a valuable asset in forage-animal modeling for more accurately studying the effects of various management strategies and environmental factors on forage nutritive value and animal performance.

Assessment of Remote Sensing and Re-Analysis Estimates of Regional Precipitation over Mato Grosso, Brazil

The spatial and temporal distribution of precipitation is of great importance for the rain-fed agricultural production and the socioeconomics of Mato Grosso (MT), Brazil. MT has a sparse network of ground rain gauges that limits the effective use of precipitation information for sustainable agricultural production and water resources in the region. Several gridded precipitation products from remote sensing and reanalysis of land surface models are currently available that can enhance the use of such information. However, these products are available at different spatial and temporal resolutions which add some challenges to stakeholders (users) to identify their appropriateness for specific applications (e.g., irrigation requirements, length of growing season, and drought monitoring). Thus, it is necessary to provide an assessment of the reliability of these precipitation estimates. The objective of this work was to compare regional precipitation estimates over MT as provided by the Global Land Data Assimilation (GLDAS), Modern-Era Retrospective Analysis for Research and Applications (MERRA), Tropical Rainfall Measurement Mission (TRMM), Global Precipitation Measurement (GPM), and the Global Precipitation Climatology Project (GPCP) with ground-based measurements. The comparison was conducted for the 2000–2018 period at eleven ground-based weather stations that covered different climate zones in MT using daily, monthly, and annual temporal resolutions. The comparison used the Pearson correlation index–r, Willmott index–d, root mean square error—RMSE, and the Wilks methods. The results showed GPM and GLDAS estimates did not differ significantly with the measured daily, monthly, and annual precipitation. TRMM estimates slightly overestimated daily precipitation by about 4.7% but did not show significant difference on the monthly and annual scales when compared with local measurements. The GPCP underestimated annual precipitation by about 7.1%. MERRA underestimated daily, monthly, and annual precipitation by about 22.9% on average. In general, all products satisfactorily estimated monthly precipitation, and most of them satisfactorily estimated annual precipitation; however, they showed low accuracy when estimating daily precipitation. The TRMM, GPM, GPCP, and GLDAS estimates had the highest performance, from high to low, while MERRA showed the lowest performance. The findings of this study can be used to support the decision-making process in the region in application related to water resources management, sustainability of agriculture production, and drought management.

Performance of a process-based model for predicting robusta coffee yield at the regional scale in Vietnam

Reliable and timely prediction of robusta coffee (Coffea canephora Pierre ex A. Froehner) yield is pivotal to the profitability of the coffee industry worldwide. In this study we assess the performance of a simple process-based model for simulating and predicting robusta coffee yield at the regional scale in Vietnam. The model includes the key processes of coffee growth and development and simulates its response to variation in climate and potential water requirements throughout the growing season. The model was built and evaluated for the major Vietnamese robusta coffee-producing provinces Dak Lak, Dak Nong, Gia Lai, Kon Tum, and Lam Dong, using official provincial coffee yield data and climate station data for the 2001–2014 period, and field data collected during a 10-year (2008–2017) survey. Overall, good agreements were found between the observed and predicted coffee yields. Root mean square error (RMSE) and mean absolute percentage error (MAPE) values ranged from 0.24 to 0.33 t ha−1, and 9% to 14%, respectively. Willmott's index of agreement (WI) was greater than or equal to 0.710 in model evaluation steps for three out of five provinces. The relatively low values of WI were found for provinces with relatively low inter-annual yield variability (i.e. Dak Lak and Dak Nong). Moreover, the model was successfully tested using remote sensing satellite and model-based gridded climate data: MAPE values were ≤ 12% and RMSE were ≤ 0.29 t ha−1. Such evaluation is important for long-term coffee productivity studies in these regions where long-term climate stations data are not readily available. The simple process-based model presented in this study could serve as a basis for developing an integrated seasonal climate-robusta coffee yield forecasting system, which would offer substantial benefits to coffee growers and industry through better supply chain management and preparedness for extreme climate events, and increased profitability.

Prediction of tropospheric ozone concentration using artificial neural networks at traffic and background urban locations in Novi Sad, Serbia.

In this paper, we described generation and performances of feedforward neural network models that could be used for a day ahead predictions of the daily maximum 1-h ozone concentration (1hO3) and 8-h average ozone concentration (8hO3) at one traffic and one background station in the urban area of Novi Sad, Serbia. The six meteorological variables for the day preceding the forecast and forecast day, ozone concentrations in the day preceding the forecast, the number of the day of the year, and the number of the weekday for which ozone prediction was performed were utilized as inputs. The three-layer perceptron neural network models with the best performance were chosen by testing with different numbers of neurons in the hidden layer and different activation functions. The mean bias error, mean absolute error, root mean squared error, correlation coefficient, and index of agreement or Willmott's Index for the validation data for 1hO3 forecasting were 0.005 μg m-3, 12.149 μg m-3, 15.926 μg m-3, 0.988, and 0.950, respectively, for the traffic station (Dnevnik), and - 0.565 μg m-3, 10.101 μg m-3, 12.962 μg m-3, 0.911, and 0.953, respectively, for the background station (Liman). For 8hO3 forecasting, statistical indicators were - 1.126 μg m-3, 10.614 μg m-3, 12.962 μg m-3, 0.910, and 0.948 respectively for the station Dnevnik and - 0.001 μg m-3, 8.574 μg m-3, 10.741 μg m-3, 0.936, and 0.966, respectively, for the station Liman. According to the Kolmogorov-Smirnov test, there is no significant difference between measured and predicted data. Models showed a good performance in forecasting days with the high values over a certain threshold.

Proposition of New Ensemble Data-Intelligence Models for Surface Water Quality Prediction

An accurate prediction of water quality (WQ) related parameters is considered as pivotal decisive tool in sustainable water resources management. In this study, five different ensemble machine learning (ML) models including Quantile regression forest (QRF), Random Forest (RF), radial support vector machine (SVM), Stochastic Gradient Boosting (GBM) and Gradient Boosting Machines (GBM_H2O) were developed to predict the monthly biochemical oxygen demand (BOD) values of the Euphrates River, Iraq. For this aim, monthly average data of water temperature (T), Turbidity, pH, Electrical Conductivity (EC), Alkalinity (Alk), Calcium (Ca), chemical oxygen demand (COD), Sulfate (SO <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">4</sub> ), total dissolved solids (TDS), total suspended solids (TSS), and BOD measured for ten years period were used in this study. The performances of these standalone models were compared with integrative models developed by coupling the applied ML models with two different feature extraction algorithms i.e., Genetic Algorithm (GA) and Principal Components Analysis (PCA). The reliability of the applied models was evaluated based on the statistical performance criteria of determination coefficient (R <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> ), root mean square error (RMSE), mean absolute error (MAE), Nash-Sutcliffe model efficiency coefficient (NSE), Willmott index (d), and percent bias (PBIAS). Results showed that among the developed models, QRF model attained the superior performance. The performance of the evaluated models presented in this study proved that the developed integrative PCA-QRF model presented much better performance compared with the standalone ones and with those integrated with GA. The statistical criteria of R <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> , RMSE, MAE, NSE, d, and PBIAS of PCA-QRF were 0.94, 0.12, 0.05, 0.93, 0.98, and 0.3, respectively.

Hybrid deep learning model for ozone concentration prediction: comprehensive evaluation and comparison with various machine and deep learning algorithms

To accurately predict tropospheric ozone concentration(O3), it is needed to investigate the variety of artificial intelligence techniques’ performance, such as machine learning, deep learning and hybrid models. This research aims to effectively predict the hourly ozone trend via fewer input variables. This ozone prediction attempt is performed on diversity data of air pollutants (NO2, NOx, CO, SO2) and meteorological parameters (wind-speed and humidity). The historical datasets are collected from 3 sites in Malaysia. The study’s methodology progressed in two paths: standalone and hybrid models where hourly-averaged datasets are applied based on 5-time horizon analysis scenario, with different inputs’ combinations. For evaluation, all models are tested throughout 5-performance indicator and illustrated on Modified Taylor diagram. Sensitivity analysis of input variables is quantified. Additionally, uncertainty analysis is conducted to assess their confidence level associated with Willmott Index. Based on R 2, results indicated that XGBoost has higher accuracy compared to MLP and SVR; meanwhile, LSTM and CNN outweighs XGBoost. In terms of robustness and accuracy, the proposed hybrid model possesses superlative performance compared to all above-mentioned techniques. The proposed model achieved exceptional results as the highest R 2, the highest 95% confidence degree, and narrower confidence interval width, are 93.48%, 98.16%, and 0.0014195, respectively.

Evaluation of the MOD11A2 product for canopy temperature monitoring in the Brazilian Atlantic Forest

Forest canopies have an important influence on the global climate balance. Through the analysis of the temperature of the canopy, it is possible to infer about the physiological aspects of the plants, helping to understand the behavior of the vegetation and, consequently, in the environmental monitoring and management of green areas. This study aims to validate the MOD11A2 V006 product from canopy surface temperature data obtained by an infrared radiation sensor. For the validation of the MOD11A2 product, a comparative analysis was performed between the land surface temperature (LST) data, obtained by the MODIS sensor, and the canopy temperature data, obtained by the SI-111 infrared radiation sensor coupled to the Itatiaia National Park (PNI) micrometeorological tower. Meteorological variables and land surface temperature collected from January to December 2018 in the PNI were also analyzed. The results reveal that the MOD11A2 product overestimates the canopy temperature in the daytime (MB ranging from 1.56 to 3.57 °C) and underestimates in the night time (MB ranging from - 0.18 to - 4.22 °C). During daytime, the months corresponding to the dry season presented a very high correlation (r = 0.74 and 0.86) and the highest values for the Willmott index (d = 0.70 and 0.64). At nighttime, the MOD11A2 product did not present a good performance for the LST estimation, especially in the rainy season. Therefore, we observed that the MOD11A2 product has limitations to estimate the land surface temperature and that possible changes in the algorithm of this product can be performed for high atmospheric humidity conditions.

A decomposition and multi-objective evolutionary optimization model for suspended sediment load prediction in rivers

Suspended sediment load (SSL) estimation is essential for both short- and long-term water resources management. Suspended sediments are taken into account as an important factor of the service life of hydraulic structures such as dams. The aim of this research is to estimat SSL by coupling intrinsic time-scale decomposition (ITD) and two kinds of DDM, namely evolutionary polynomial regression (EPR) and model tree (MT) DDMs, at the Sarighamish and Varand Stations in Iran. Measured data based on their lag times are decomposed into several proper rotation components (PRCs) and a residual, which are then considered as inputs for the proposed model. Results indicate that the prediction accuracy of ITD-EPR is the best for both the Sarighamish (R 2 = 0.92 and WI = 0.96) and Varand (R 2 = 0.92 and WI = 0.93) Stations (WI is the Willmott index of agreement), while a standalone MT model performs poorly for these stations compared with other approaches (EPR, ITD-EPR and ITD-MT) although peak SSL values are approximately equal to those by ITD-EPR. Results of the proposed models are also compared with those of the sediment rating curve (SRC) method. The ITD-EPR predictions are remarkably superior to those by the SRC method with respect to several conventional performance evaluation metrics.

Artificial Intelligence for Modeling Real Estate Price Using Call Detail Records and Hybrid Machine Learning Approach.

Advancement of accurate models for predicting real estate price is of utmost importance for urban development and several critical economic functions. Due to the significant uncertainties and dynamic variables, modeling real estate has been studied as complex systems. In this study, a novel machine learning method is proposed to tackle real estate modeling complexity. Call detail records (CDR) provides excellent opportunities for in-depth investigation of the mobility characterization. This study explores the CDR potential for predicting the real estate price with the aid of artificial intelligence (AI). Several essential mobility entropy factors, including dweller entropy, dweller gyration, workers’ entropy, worker gyration, dwellers’ work distance, and workers’ home distance, are used as input variables. The prediction model is developed using the machine learning method of multi-layered perceptron (MLP) trained with the evolutionary algorithm of particle swarm optimization (PSO). Model performance is evaluated using mean square error (MSE), sustainability index (SI), and Willmott’s index (WI). The proposed model showed promising results revealing that the workers’ entropy and the dwellers’ work distances directly influence the real estate price. However, the dweller gyration, dweller entropy, workers’ gyration, and the workers’ home had a minimum effect on the price. Furthermore, it is shown that the flow of activities and entropy of mobility are often associated with the regions with lower real estate prices.

Spatial methods to analyze the relationship between Spanish soil properties and cadmium content

In this study, concentrations of cadmium using 3778 samples encompassing the total size of Spain (about 505 km2) were investgated. Two novel spatial methods namely Moran eigenvector spatially varying coefficient (MESVC) and spatially filtered unconditional quantile regression (SF-UQR) were employed with the aim of avoiding the problem of local collinearity which is prevalent in regression models. Additionally, the spatially varying coefficients methods were applied to assess the influence of soil properties together with soil texture on the spatial variations of cadmium. It was indicated that the overall level of cadmium is low compared to the concentrations found around the world. In particular, the values of Cd varied between 0.01 and 2.00 mgkg−1, with the median of 0.23 mgkg−1. The residual standard error and adjusted R2 produced by MESVC were 0.16 and 0.69, respectively which are better than 0.21 and 0.39 yielded by the SF-UQR model. Both of these models outperformed compared to the geographically weighted regression (GWR) and the performance of MESVC was also better than the traditional method of kriging. For instance, in terms of willmott index (d) and root mean squared relative error (RMSRE), the MESVC had superior performance with values equal to 0.612 and 0.275 compared to 0.399 and 0.379 obtained for the ordinary kriging. The MESVC and GWR demonstrated that CaCO3, sand, silt and clay had a negligible influence on spatial variations of cadmium whereas, EC had the largest contribution followed by SOM and pH.