Nash-Sutcliffe Efficiency Index Research Articles

Wheat Yield Forecasting for the Tisza River Catchment Using Landsat 8 NDVI and SAVI Time Series and Reported Crop Statistics

Due to the increasing global demand of food grain, early and reliable information on crop production is important in decision making in agricultural production. Remote sensing (RS)-based forecast models developed from vegetation indices have the potential to give quantitative and timely information on crops for larger regions or even at farm scale. Different vegetation indices are being used for this purpose, however, their efficiency in estimating crop yield certainly needs to be tested. In this study, wheat yield was derived by linear regressing reported yield values against a time series of six different peak-seasons (2013–2018) using the Landsat 8-derived Normalized Difference Vegetation Index (NDVI) and Soil Adjusted Vegetation Index (SAVI). NDVI- and SAVI-based forecasting models were validated based on 2018–2019 datasets and compared to evaluate the most appropriate index that performs better in forecasting wheat production in the Tisza river basin. Nash-Sutcliffe efficiency index was positive with E1 = 0.716 for the model from NDVI and for SAVI E1 = 0.909, which means that the forecasting method developed and performed good forecast efficiency. The best time for wheat yield prediction with Landsat 8-SAVI and NDVI was found to be the beginning of full biomass period from the 138th to 167th day of the year (18 May to 16 June; BBCH scale: 41–71) with high regression coefficients between the vegetation indices and the wheat yield. The RMSE of the NDVI-based prediction model was 0.357 t/ha (NRMSE: 7.33%). The RMSE of the SAVI-based prediction model was 0.191 t/ha (NRMSE 3.86%). The validation of the results revealed that the SAVI-based model provided more accurate forecasts compared to NDVI. Overall, probable yield amount is possible to predict far before harvest (six weeks earlier) based on Landsat 8 NDVI and SAVI and generating simple thresholds for yield forecasting, and a potential loss of wheat yield can be mapped.

Irrigation water infiltration modeling using machine learning

The present study proposes five standard artificial intelligence models including Artificial Neural Network (ANN), Adaptive Neuro-Fuzzy Inference System (ANFIS), Group Method of Data Handling (GMDH), Multivariate Linear Regression (MLR), and Support Vector Regression (SVR) as well as their integrative models combined with the nature-inspired Firefly Algorithm (FA) to forecast the infiltrated water in the furrow irrigation system. The FA functions as an optimizer tool in constructing the integrative models. Data for assessing these models was gathered from published literatures and field experiments conducted on researching farm at SBUK University, Kerman, Iran. The input parameters used in modeling were furrow length (L), the inflow rate (Q), advance time at the end of the furrow (TL), cross-sectional area of inflow (Ao) and infiltration opportunity time (To). The root mean square (RMSE), the mean absolute error (MAE), the correlation coefficient (R2), the Nash–Sutcliffe efficiency index (NSE), and the index of agreement (IA) were used to evaluate the predicting performance of the suggested models. The best values of the statistical criteria and in training and testing phase of the simulation imply that the MLPNN-FA and SVR-FA provided an excellent fit to the observed data. Comparison of the results revealed that the FA is able to boost the accuracy of the models so that the improvement in RMSE values was 5%, 1%, 4%, 47% in the ANFIS, GMDH, MLPNN, and SVR, respectively. Having the calculated value for the comprehensive synthesis index (SI) indicated that using the combination of the SVR and FA noticeably enhance the performance of the standard SVR model by 97%. Even though the FA algorithm had a trivial influence on the GMDH model; however, it can be introduced as a powerful tool to enhance the complicated modeling processes.

Development and Validation of a New Calibration Model for Diviner 2000® Probe Based on Soil Physical Attributes

This study aimed to develop a new model, valid for soil with and without expandable characters, to estimate volumetric soil water content (θ) from readings of scaled frequency (SF) acquired with the Diviner 2000® sensor. The analysis was carried out on six soils collected in western Sicily, sieved at 5 mm, and repacked to obtain the maximum and minimum bulk density (ρb). During an air-drying process SF values, the corresponding gravimetric soil water content (U) and ρb were monitored. In shrinking/swelling clay soils, due to the contraction process, the variation of dielectric permittivity was affected by the combination of the mutual proportions between the water volumes and the air present in the soil. Thus, to account for the changes of ρb with U, the proposed model assumed θ as the dependent variable being SF and ρb the independent variables; then the model’s parameters were estimated based on the sand and clay fractions. The model validation was finally carried out based on data acquired in undisturbed monoliths sampled in the same areas. The estimated θ, θestim, was generally close to the corresponding measured, θmeas, with Root Mean Square Errors (RMSE) generally lower than 0.049 cm3 cm−3, quite low Mean Bias Errors (MBE), ranging between −0.028 and 0.045 cm3 cm−3, and always positive Nash-Sutcliffe Efficiency index (NSE), confirming the good performance of the model.

CFSR- NCEP Performance for weather data forecasting in the Pernambuco Semiarid, Brazil

The present study aims to evaluate meteorological data -with real time actualization- from the Climate Forecast System Reanalysis (CFRS) of the National Centers for Environmental Prediction (NCEP), comparing them with data from local stations in two mesoregions: Sertão de Pernambuco (SP) and Sertão do São Francisco (SFF), semi-arid region of Pernambuco, Brazil. Statistical performance indicators were used for the period since 1979 to 2014 and the variables: precipitation (P), average, minimum and maximum temperature (Tm, Tn, Tx respectively), relative humidity (HR), wind speed (Vv), solar radiation (RS) and potential evapotranspiration (ETo). Tn, Tm and Tx showed the best results for the determination coefficient (R2), Willmott concordance index (d), Nash-Sutcliffe efficiency index (NSE) and percentage bias (PBIAS). ETo, P and HR obtained acceptable values for R2, d and NSE. CFSR data shows good performance with d values between 0.63 and 0.94.

Combination of data-driven models and interpolation technique to develop of PM10 map for Hanoi, Vietnam

The degradation of air quality is the most concerned issue of our society due to its harmful impacts on human health, especially in cities with rapid urbanization and population growth like Hanoi, the capital of Vietnam. This study aims at developing a new approach that combines data-driven models and interpolation technique to develop the PM10 concentration maps from meteorological factors for the central area of Hanoi. Data-driven models that relate the PM10 concentration with the meteorological factors at the air quality monitoring stations in the study area were developed using the Multiple Linear Regression (MLR) and Artificial Neural Network (ANN) algorithms. Models’ performance comparison showed that ANN models yielded better goodness-of-fit indices than MLR models at all stations in the study area with average coefficient of correlation (r) and Nash–Sutcliffe Efficiency Index (NSE) of 0.51 and 0.34 for the former, and 0.7 and 0.49 for the latter. These indices indicates that the ANN-based data-driven models outperformed the MLR-based models. Thus, the ANN-based models and the Inverse Distance Weighting (IDW) interpolation technique were then combined for mapping the monthly PM10 concentration with a spatial resolution of 1 km from global meteorological data. With this combination, the PM10 concentration maps account for both local PM10 concentration and impacts of spatio-temporal variations of meteorological factors on the PM10 concentration. This study provides a promising method to predict the PM concentration with a high spatio-temporal resolution from meteorological data.

A Self-adaptive differential evolutionary extreme learning machine (SaDE-ELM): a novel approach to blast-induced ground vibration prediction

Blast-induced ground vibration is still an adverse impact of blasting in civil and mining engineering projects that need much consideration and attention. This study proposes the use of Self-Adaptive Differential Evolutionary Extreme Learning Machine (SaDE-ELM) for the prediction of ground vibration due to blasting using 210 blasting data points from an open pit mine in Ghana. To ascertain the predictive performance of the proposed SaDE-ELM approach, several artificial intelligence and empirical approaches were developed for comparative purposes. The performances of various developed models were assessed using model performance indicators of mean squared error (MSE), Nash–Sutcliffe Efficiency Index (NSEI) and correlation coefficient (R). Furthermore, the Bayesian Information Criterion (BIC) was applied to select the best performing approach. The obtained prediction results based on the performance indicators showed that the SaDE-ELM outperformed all the competing models as it had the lowest MSE value of 0.01942, respectively. The SaDE-ELM also achieved the highest R and NSEI values of 0.8711 and 0.7537, respectively. The other artificial intelligent approaches had MSE, R and NSEI in the ranges of (0.02166–0.03006), (0.8012–0.8537) and (0.6188–0.7254), respectively. The empirical approaches performed poorly relative to the artificial intelligence approaches by having had MSE, R and NSEI in the ranges of (0.03419–0.06587), (0.7466–0.7833) and (0.1649–0.5665), respectively. The prediction superiority of SaDE-ELM was confirmed when it is achieved the lowest BIC value of − 293.40. Therefore, the proposed SaDE-ELM has demonstrated great potential to be used for on-site prediction, control and management of blast-induced ground vibration to prevent unwanted effects on the environment.

Exploring the application of soft computing techniques for spatial evaluation of groundwater quality variables

The precise estimation of spatial variation of groundwater quality variables is necessary for making appropriate decisions regarding aquifer resources management and environmental systems. The aim of this research is to evaluate five different data-intelligent methods so as to get the best one in spatial estimation of electrical conductivity (EC) and total hardness (TH) of groundwater resources. The study uses water quality data including Na, Cl, Mg, Ca, SO4, HCO3, TH, and EC, observed at 367 wells, in the Chhattisgarh state, central India. Four different scenarios are decided by implementing Gamma and linear regression methods. Data-intelligent methods applied to selected scenarios to estimate EC and TH are dynamic evolving neural-fuzzy inference system (DENFIS), group method of data handling (GMDH), multivariate adaptive regression spline (MARS), M5 Tree model (M5 Tree), and gene expression programming (GEP). Evaluation of the methods with respect to correlation coefficient (R), root mean square error (RMSE), mean absolute error (MAE), Nash-Sutcliffe efficiency (NSE), Legates and McCabe index (LMI), and Willmott index of agreement (WI) indicates that the GEP for the input combination including TH, Na, Cl, HCO3 variables and the DENFIS by employing Ca, Mg, EC variables provide the best accuracy in estimating EC and TH, respectively. The statistical metrics of the GEP (DENFIS) for the best scenario are R = 0.970 (0.958), NSE = 0.941 (0.917), RMSE = 105.317 μs/cm (34.692 mg/L), MAE = 57.075 μs/cm (12.603 mg/L), WI = 0.984 (0.979) and LMI = 0.824 (0.868). The spatial distribution maps indicate that the EC and TH values of groundwater resources in the studied area are in their permissible limits for drinking consumption and the interpolated maps of the data-intelligent techniques have suitable precision in comparison with the observed maps.

Estimation of Suspended Sediment Concentration by M5 Model Tree Based on Hydrological and Moderate Resolution Imaging Spectroradiometer (MODIS) Data

The Estimation of suspended sediment concentration (SSC) is an important factor in river engineering, which is used as an indicator of land-use change, water quality studies, and all projects related to constructions in rivers. In this research, the M5 model tree and the Moderate Resolution Imaging Spectroradiometer (MODIS) data were utilized to estimate the SSC at Ahvaz station on the Karun River. In this study, 135 cloud-free images of the MODIS sensor on the Terra satellite were taken for days corresponding to field SSC data, during the years 2000 to 2015. Input parameters of the model tree in this study were flow discharge, derived from hydrological data, and red (R), near-infrared (NIR) bands, and NIR/R ratio extracted from MODIS imagery. The results of statistical analysis illustrate that the M5 model outperforms the sediment rating curve (SRC) method, which is the most common method of estimating suspended sediment load. The Nash-Sutcliffe efficiency index for the M5 model tree of 0.58 was achieved, which was much better than that of the SRC method (0.26). At high fluxes, the efficiency of the SRC method significantly reduced, while the model tree provides acceptable results. The global sensitivity analysis on the M5 model pointed out that 93% of output variance was established by the main effects of input parameters, and less than 7% belong to the interaction effects. 73% and 12% of output variance specified by the main effects of flow discharge and NIR/R ratio, respectively.

Assessing the Performance of Different Time of Concentration Equations in Urban Ungauged Watersheds: Case Study of Cartagena de Indias, Colombia

In ungauged watersheds, the estimation of the time of concentration (Tc) is always a challenging task due to the intrinsic uncertainty involved when making assumptions. Given that Tc is one of the main inputs in a hydrological analysis for the design of hydraulic structures for stormwater management, ten equations (including one proposed in several local studies) and two Tc methodologies (overland flow time plus channel flow time) were used to compute the Tc in fifteen urban ungauged watersheds, located in Cartagena de Indias (Colombia), with different area sizes and slopes to statistically assess their performance against the value obtained via the Natural Resources Conservation Service (NRCS) velocity method (assumed to be the true value). According to the Nash–Sutcliffe efficiency index, none of the equations proved to be reliable in all watersheds as only four equations predicted the Tc value in 53% of the cases. In addition, based on the percent bias, all equations tended to significantly over- or underestimate the Tc, which affects the quantification of the runoff volume necessary for, among others, the implementation of best management practices for watershed management (e.g., conventional and/or sustainable drainage system design), flood-prone area delineation and flood risk analyses, urban planning, and stream restoration.

Spatial year-ahead forecast of air temperature and precipitation in large mountain areas

The method for spatial reconstruction of monthly mean air temperature and monthly precipitation is proposed for large mountain areas. Meteorological factor dynamics is described in percent of long-term monthly average values measured “in situ”. It is almost independent of height, exposure and climate of characterized sites and remains the same throughout the mountainous territory. By the example of the Altai-Sayan mountain country covering more than 2,500,000 km2, we averaged (spatially generalized) the data of 11 weather stations and obtained the country's long-term dynamics of monthly mean air temperature and monthly precipitation for 1951–2016. Using a simple moving average of 6–7 recent years of stations observations, we provide a reliable year-ahead prediction of a factor value for any spot of the country without additional data. Modification of RSR and Nash–Sutcliffe efficiency indices is proposed to correct their shortcomings in assessing the quality of various predictive methods and models. The developed method of four-season weather forecast has the quality close to the theoretically best for statistical forecasts.

Improving regional groundwater storage estimates from GRACE and global hydrological models over Tasmania, Australia

Accuracy of groundwater storage (GWS) estimates from the Gravity Recovery and Climate Experiment (GRACE) mission usually has certain relations with hydrological models. This study develops a statistical selection approach to optimally estimate GWS from GRACE using two hydrological models: the Global Land Data Assimilation System (GLDAS) and the WaterGAP Global Hydrology Model (WGHM), over Tasmania, Australia. This approach involves three variables: the long-term trend, Pearson correlation coefficient (PR), and root mean square error (RMSE). The results show that in-situ observations are highly correlated with GRACE-GLDAS (PR from 0.64 to 0.85) and GRACE-WGHM (PR from 0.69 to 0.88) in eastern and northern regions of Tasmania, respectively. The interannual trends of GRACE-GLDAS estimates are generally ~1.8 times larger than those from GRACE-WGHM solutions. With regard to the standard method, the statistical selection approach can effectively improve the PR and Nash-Sutcliffe efficiency index (NSE) by 3.80 and 1.38%, respectively, over the northern region, while it decreases the RMSE by 1.07%. Similar improvements can also be detected in the eastern region. In terms of spatial distribution, the statistical approach benefits from advantages of the different models, especially to preserve the characteristics of Central Highland. Overall, according to the models, Tasmania experienced a pronounced GWS decline during the Millennium Drought (2003–2010), at a depletion rate of –2.57 mm/year, mainly due to decreasing precipitation. The increasing precipitation infiltration after 2010 lead to the GWS recovery by 3.94 mm/year. The limitation of the method is that it depends on the availability of in-situ groundwater level data.

Fusion of MODIS and Landsat-Like Images for Daily High Spatial Resolution NDVI

One of the obstacles in monitoring agricultural crops is the difficulty in understanding and mapping rapid changes of these crops. With the purpose of addressing this issue, this study aimed to model and fuse the Moderate Resolution Imaging Spectroradiometer (MODIS) Normalized Difference Vegetation Index (NDVI) using Landsat-like images to achieve daily high spatial resolution NDVI. The study was performed for the period of 2017 on a commercial farm of irrigated maize-soybean rotation in the western region of the state of Bahia, Brazil. To achieve the objective, the following procedures were performed: (i) Landsat-like images were upscaled to match the Landsat-8 spatial resolution (30 m); (ii) the reflectance of Landsat-like images was intercalibrated using the Landsat-8 as a reference; (iii) Landsat-like reflectance images were upscaled to match the MODIS sensor spatial resolution (250 m); (iv) regression models were trained daily to model MODIS NDVI using the upscaled Landsat-like reflectance images (250 m) of the closest day as the input; and (v) the intercalibrated version of the Landsat-like images (30 m) used in the previous step was used as the input for the trained model, resulting in a downscaled MODIS NDVI (30 m). To determine the best fitting model, we used the following statistical metrics: coefficient of determination (r2), root mean square error (RMSE), Nash–Sutcliffe efficiency index (NSE), mean bias error (MBE), and mean absolute error (MAE). Among the assessed regression models, the Cubist algorithm was sensitive to changes in agriculture and performed best in modeling of the Landsat-like MODIS NDVI. The results obtained in the present research are promising and can enable the monitoring of dynamic phenomena with images available free of charge, changing the way in which decisions are made using satellite images.

An approach to runoff modelling in small karst watersheds using the SWAT model

The soil and water assessment tool (SWAT) is one of the most widely used hydrological models. However, it has not been extensively applied in karst watersheds due to their geomorphological heterogeneity. Karst watersheds contribute up to 50% of water supply in Europe, while in South China alone, over 100 million people subsist on it. Hence, improving model performance for karst areas is an important area of research. Here, we propose a methodology based on field investigation and tracer test aimed at improving the conceptual understanding of the hydrological processes to guide the modeller. The Miaogou watershed was used as a case study to run the SWAT model using daily streamflow data from 2013 to 2016. Model calibration and validation results were found to be positive. Average R2 and Nash–Sutcliffe efficiency index (NSE) solutions for the validation period are 0.70 and 0.65 respectively. CN2 was the most sensitive parameter due to the concave shape of the basin as well as the presence of a sinkhole that conveys recharge faster to the main stream outlet via a spring. Our findings show that a field-based approach to modelling karst areas helps define watershed surface and groundwater boundaries, which is often neglected. It also serves as a secondary validation that can be applied anywhere rather than relying on computer code alone. In addition, tracer tests highlight the subterranean hydrological processes which define hydrological gains and losses necessary for model parameterization. Overall, this study establishes a positive outlook towards extensive application in other karstified regions of the world.

Impact of rainfall properties on the performance of hydrological models for green roofs simulation.

Green roofs (GR) are effective tools for the mitigation of the negative hydrological impact linked to uncontrolled urbanization. Models for runoff response of vegetated covers support planning decisions about the use of this technology in ever-expanding areas, but there is still large uncertainty in this research area. The goal of the present study was to define the accuracy of three selected models for the simulation of the hydrological behavior of a GR, with a particular focus on the precipitation pattern characteristics. The Nash cascade model, Storm Water Management Model (SWMM), and HYDRUS-1D models were selected. Each model was calibrated over 24 rainfall-runoff events collected at the experimental site located in Southern Italy. Rainfall events were characterized using several criteria and were divided into convective, stratiform, and tropical precipitation types according to the shape of the rainfall profile identified by the rainfall binary shape code. The models returned overall satisfactory performances with average Nash-Sutcliffe Efficiency index of 0.65 for the SWMM and HYDRUS and 0.73 for the Nash model. In general, the models were better performing for stratiform and tropical events. SWMM and HYDRUS predicted with higher accuracy the convective events while the Nash model appeared more suitable for stratiform events.

Evaluation of the performance of Euro-CORDEX Regional Climate Models for assessing hydrological climate change impacts in Great Britain: A comparison of different spatial resolutions and quantile mapping bias correction methods

Regional Climate Models (RCMs) are an essential tool for analysing regional climate change impacts, such as hydrological change, as they provide simulations with more small-scale details and expected smaller errors than global climate models. There has been much effort to increase the spatial resolution and simulation skill of RCMs (i.e. through bias correction), yet the extent to which this improves the projection of hydrological change is unclear. Here, we evaluate the skill of five reanalysis-driven Euro-CORDEX RCMs in simulating precipitation and temperature, and as drivers of a hydrological model to simulate river flow on four UK catchments covering different physical, climatic and hydrological characteristics. We use a comprehensive range of evaluation indices for aspects of the distribution such as means and extremes, as well as for the structure of time series. We test whether high-resolution RCMs provide added value, through analysis of two RCM resolutions, 0.44° (50 km) and 0.11° (12.5 km), which are also bias-corrected employing the parametric quantile-mapping (QM) method, using the normal distribution for temperature, and the Gamma (GQM) and Double Gamma (DGQM) distributions for precipitation. The performance of these is considered for a range of meteorological variables and for the skill in simulating hydrological impacts at the catchment scale.In a small catchment with complex topography, the 0.11° RCMs clearly outperform their 0.44° version for precipitation and temperature, but when used in combination with the hydrological model, fail to capture the observed river flow distribution. In the other (larger) catchments, only one high-resolution RCM consistently outperforms its low-resolution version, implying that in general there is no added value from using the high-resolution RCMs in those catchments. Both resolutions produce river flow simulations that cover the observed flow duration curve, but the ensemble spread is large and therefore the simulations are difficult to use in practice. GQM decreases most of the simulation biases, except for extreme precipitation and high flows, which are further decreased by DGQM, which also reduces the multi-model simulation spread. Bias correction does not improve the representation of daily temporal variability measured by the Nash-Sutcliffe Efficiency Index, but it does for monthly variability, in particular when applying DGQM, which reduces most of the simulation biases. Overall, an increase in RCM resolution does not imply a better simulation of hydrology and bias-correction represents an alternative to ease decision-making.

Intensity-duration-frequency of maximum rainfall in Mato Grosso State

Intensive rainfall is an important meteorological variable that is of technical interest in hydraulic projects. This study therefore generated Intensity-Duration-Frequency equations (IDF) for 14 weather stations in Mato Grosso State, based on pluviograph analysis. Annual maximum rainfall data regarding 10-to-1440-minute long rainfall events were collected from digitized daily pluviographs. Data adherence to the generalized extreme value distribution (GEV) was checked through the Kolmogorov-Smirnov test at a 20% significance level. Next, the maximum probable rainfall for return periods such as 2, 5, 10, 20, 30, 50 and 100 years was calculated and the IDF equations were adjusted. The performance of the IDF equations was evaluated based on mean absolute error (MAE), root mean square error (RMSE), bias, Willmott's concordance index and Nash-Sutcliffe efficiency index (ENS). Adjusting the IDF equations was only possible for rainfall durations ranging from 10 to 360 min at each station due to the low frequency of longer rainfalls. High variation was present in parameters of the IDF equation and in maximum rainfall intensity between stations. The satisfactory performance of the models, as attested to by statistical indices, allows using IDF equations adjusted for rainfall durations from 10 to 360 min, and return periods from 2 to 100 years, in the regions of the Mato Grosso weather stations.

Assessment of Remotely Sensed Near-Surface Soil Moisture for Distributed Eco-Hydrological Model Implementation

The aim of this study was to implement an eco-hydrological distributed model using only remotely sensed information (soil moisture and leaf area index) during the calibration phase. Four soil moisture-based metrics were assessed, and the best alternative was chosen, which was a metric based on the similarity between the principal components that explained at least 95% of the soil moisture variation and the Nash-Sutcliffe Efficiency (NSE) index between simulated and observed surface soil moisture. The selected alternative was compared with a streamflow-based calibration approach. The results showed that the streamflow-based calibration approach, even presenting satisfactory results in the calibration period (NSE = 0.91), performed poorly in the validation period (NSE = 0.47) and Leaf Area Index (LAI) and soil moisture were neither sensitive to the spatio-temporal pattern nor to the spatial correlation in both calibration and validation periods. Hence, the selected soil moisture-based approach showed an acceptable performance in terms of discharges, presenting a negligible decrease in the validation period (ΔNSE = 0.1) and greater sensitivity to the spatio-temporal variables’ spatial representation.

Hydrological Modeling of Green Roofs Runoff by Nash Cascade Model

Background: Green roofs (GRs) technology has gained increasing interest in recent years since it offers multiple benefits to urban environments, citizens and buildings. Eco-covers can capture some water nutrient pollutants, filter air pollutants and moderate the urban heat island effect. Beyond these benefits, abundant literature stresses the role played by the GRs from the hydrological perspective. They allow to face the increasing stress on the traditional urban drainage systems by reducing the annual stormwater runoff. In light of this, the hydrological behavior prediction of a vegetated cover is essential for urban planners, policy makers and engineers in order to quantify runoff mitigation potential so as to optimize their application. Objective: The aim of the present research is to meet this need by testing the accuracy of Nash cascade model in predicting the stormwater production of GR systems. Materials and Methods: The proposed model has been calibrated against hourly data of thirteen rainfall-runoff events observed at two experimental sites, both located within the campus of the University of Salerno, southern Italy. Event scale model calibration, aimed at the identification of the storage coefficient parameter, has been based on the optimization of a number of error statistics. Results and Conclusion: Despite its simplicity, the investigated model appears adequately to be able to predict the runoff production from the experimental green roofs with a good degree of accuracy, as described by the Nash-Sutcliffe Efficiency index, which ranges between 0.54 and 0.94.

Experimental and numerical investigation of drainage mechanisms at sports fields under simulated rainfall

Infiltration and drainage are complex hydrological considerations to determine unsaturated flow processes in case of irrigation, root growth and pollutant transport in porous media consisting of multi-layered soils. Considering sports fields, this affects the quality of sports organization in terms of player’s health, the participation of spectators and the cancellation of games. This study investigated the drainage behavior of sports fields both experimentally and numerically under simulated rainfall. To achieve this goal, a newly-developed experimental setup which consists of a rainfall simulator (RS) that is able to simulate the natural rainfall, and a drainage tank (DT) that is able to pack stratified drainage layers same as in the sports fields were used. Experimental rainfall hyetographs were designed to simulate various rainfall conditions for the relevant region in Istanbul, Turkey. Numerous experiments were conducted to investigate hydrological description of unsaturated flow by using packed drainage layers in the laboratory. Time-dependent water contents were also monitored using soil moisture sensors (10-HS, Decagon Devices) at different depths in the drainage layers. Soil water retention curve (SWRC) of each drainage layer obtained from calibration tests and empirical parameters (α,n) were optimized with HYDRUS-3D model by using water contents and suction pressure results. Observed drain outflow hydrographs were compared with simulated drain outflow hydrographs by using statistical indices of Nash-Sutcliffe Efficiency (NSE) index, Kling and Gupta Efficiency (KGE) index and determination coefficient (R2). Experimental results and HYDRUS-3D simulations showed good compatibility with the values of NSE, KGE and R2 varied between 0.859–0.958, 0.594–0.972 and 0.868–0.975, respectively. In the present study, experimental and numerical investigation of the drainage mechanisms for sports fields was evaluated by considering unsaturated flow characteristics through the different drainage layers.

Long-term forecasting system using wavelet – nonlinear autoregressive neural network conjunction model

Purpose An accurate long-term multi-step forecast provides crucial basic information for planning and reinforcing managerial decision-support. However, nonstationarity and nonlinearity, normally consisted of several types of managerial data can seriously ruin the forecasting computation. This paper aims to propose an effective long-term multi-step forecasting conjunction model, namely, wavelet–nonlinear autoregressive neural network (WNAR) conjunction model. The WNAR combines discrete wavelet transform (DWT) and nonlinear autoregressive neural network (NAR) to cope with such nonstationarity and nonlinearity within the managerial data; as a consequence, provides insight information that enhances accuracy and reliability of long-term multi-step perspective, leading to effective management decision-making. Design/methodology/approach Based on WNAR conjunction model, wavelet decomposition is executed for efficiently extracting hidden significant, temporal features contained in each of six benchmark nonstationary data sets from different managerial domains. Then, each extracted feature set at a particular resolution level is fed into NAR for the further forecast. Finally, NAR forecasting results are reconstructed. Forecasting performance measures throughout 1 to 30-time lags rely on mean absolute percentage error (MAPE), root mean square error (RMSE), Nash-Sutcliffe efficiency index or the coefficient of efficiency (Ef) and Diebold–Mariano (DM) test. An effect of data characteristic in terms of autocorrelation on forecasting performances of each data set are observed. Findings Long-term multi-step forecasting results show the best accuracy and high-reliability performance of the proposed WNAR conjunction model over some other efficient forecasting models including a single NAR model. This is confirmed by DM test, especially for the short-forecasting horizon. In addition, rather steady, effective long-term multi-step forecasting performances are yielded with slight effect from time lag changes especially for the data sets having particular high autocorrelation, relative against 95 per cent degree of confidence normal distribution bounds. Research limitations/implications The WNAR, which combines DWT with NAR can be accounted as a bridge for the gap between machine learning, engineering signal processing and management decision-support systems. Thus, WNAR is referred to as a forecasting tool that provides insight long-term information for managerial practices. However, in practice, suitable exogenous input forecast factors are required on the managerial domain-by-domain basis to correctly foresee and effectively prepare necessary reasonable management activities. Originality/value Few works have been implemented to handle the nonstationarity, consisted of nonlinear managerial data to attain high-accurate long-term multi-step forecast. Combining DWT and NAR capabilities would comprehensively and specifically deal with the nonstationarity and nonlinearity difficulties at once. In addition, it is found that the proposed WNAR yields rather steady, effective long-term multi-step forecasting performance throughout specific long time lags regarding the data, having certainly high autocorrelation levels across such long time lags.