Nash-Sutcliffe Model Efficiency Research Articles

Water and Nitrogen Dynamics in Drip Fertigated Tomato for Water of Different Qualities under Polyhouse Conditions


 Water and NO3-N dynamics in the soil during the growing season is an important tool in improving the nitrogen management and environmental protection. HYDRUS-2D has been widely used to predict the water and NO3-N distribution in the soil. The objective of this study was to simulate the water and NO3-N distribution in the soil under drip fertigated tomato irrigated with different water qualities under polyhouse conditions. Field data were collected on spatial and temporal distribution of water and available NO3-N during growing season. The model was calibrated for the hydraulic conductivity and parameters were used for the validation of the model. The model performance in simulating the water and NO3-N was evaluated by using coefficient of determination (R2), root mean square error (RMSE), index of agreement and Nash–Sutcliffe model efficiency (NSE). For both calibration and validation, the higher values of R2 from 0.70 to 0.99 for water distribution and 0.70 to 0.96 for NO3-N distribution showed that observed and predicted values are highly correlated. The value of RMSE ranges from 0.004 to 0.0016 for water and 0.002-0.006 for NO3-N distribution. The index of agreement value varied from 0.86-0.98 for water distribution and 0.89-0.99 for NO3-N distribution. The values of NSE (nearer to 1) i.e. 0.17 to 0.98 for water distribution and -0.09 to 0.94 for NO3-N distribution show that HYDRUS-2D was predicting with good accuracy. From these results, it can be concluded that the model performs well for predicting the water and NO3-N distribution in the tomato crop irrigated with different water qualities under polyhouse conditions.

Application of the JULES-crop model and agrometeorological indicators for forecasting off-season maize yield in Brazil

Zea mays L is a crucial crop for Brazil, ranking second in terms of production and sixth in terms of exports. In Brazil, the second season, or off-season, accounts for 80 % of the overall maize output, which primarily occurs after the soybean main season. A maize yield forecast model for the off-season was developed and implemented throughout Brazilian territory due to its importance to the country's economy and food security. The model was built using multiple linear regressions that connected outputs simulated from a land surface model used in large-scale analysis for agriculture (JULES-crop), to agrometeorological indicators. The application of the developed model occurred every 10 days from the sowing until the maturation. A comparison of the forecasting model was verified with the official off-season maize yields for the years 2003–2016. Agrometeorological indicators during the reproductive phase accounted for 60 % of the interannual variability in maize production. When outputs simulated by JULES-crop were included, the forecasting model achieved Nash-Sutcliffe modeling efficiency (EF) of 0.77 in the maturation and EF = 0.72 in the filling-grain stage, suggesting that this approach can generate useful predictions for final maize yield beginning on the 80th day of the cycle. Outputs of JULES crop enhanced modeling performance during the vegetative stage, reducing the standard deviation error in prediction from 0.59 to 0.49 Mg ha−1.

Predicting sediment yield and locating hotspot areas in the Hamesa watershed of Ethiopia for effective watershed management

Abstract Locating hotspots and assessing sediment accumulation are crucial aspects of water body management. The primary aim of this study was to examine sediment yield in the Hamesa watershed utilizing the Soil and Water Assessment Tool (SWAT) model and to propose best management practices. The basin was divided into 15 sub-basins, with 103 hydrological response units at the outlet of the Hamesa watershed. Simulation was conducted using meteorological and spatial data. Monthly streamflow and sediment data were calibrated for the period from 2000 to 2010 and validated for the period from 2011 to 2015 using the SWAT Uncertainty Calibration Program Sequential Uncertainty Fit (SUFI-2). Model performance was assessed using metrics including the coefficient of determination (R2), Nash–Sutcliffe model efficiency, observation standard deviation ratio, and percentage bias, which demonstrated very good results in both calibration and validation periods. The average annual sediment production in the Hamesa watershed was estimated at 9,800 t/year. Nine out of 15 affected sub-basins were categorized as producing moderate to very high sediment content (4.54–12.82 t/ha/year) and were chosen for sediment reduction scenarios. This study will play a significant role in managing impacted watersheds affected by soil erosion.

A past and present perspective on the European summer vapor pressure deficit

Abstract. The response of evapotranspiration to anthropogenic warming is of critical importance for the water and carbon cycle. Contradictory conclusions about evapotranspiration changes are caused primarily by their brevity in time and sparsity in space, as well as the strong influence of internal variability. Here, we present the first gridded reconstruction of the summer (June, July, and August) vapor pressure deficit (VPD) for the past 4 centuries at the European level. This gridded reconstruction is based on 26 European tree ring oxygen isotope records and is obtained using a random forest approach. According to validation scores obtained with the Nash–Sutcliffe model efficiency, our reconstruction is robust over large parts of Europe since 1600, in particular for the westernmost and northernmost regions, where most tree ring records are located. Based on our reconstruction, we show that from the mid-1700s a trend towards higher summer VPD occurred in central Europe and the Mediterranean region that is related to a simultaneous increase in temperature and decrease in precipitation. This increasing summer VPD trend continues throughout the observational period and in recent times. Moreover, our summer VPD reconstruction helps to visualize the local and regional impacts of the current climate change, as well as to minimize statistical uncertainties of historical VPD variability. This paper provides also new insights into the relationship between summer VPD and large-scale atmospheric circulation, and we show that summer VPD has two preferred modes of variability, namely a NW–SE dipole-like mode and a N–S dipole-like mode. Furthermore, the interdisciplinary use of the data should be emphasized, as summer VPD is a crucial parameter for many climatological feedback processes in the Earth's surface system. The reconstructed summer VPD gridded data over the last 400 years are available at the following link: https://doi.org/10.5281/zenodo.5958836 (Balting et al., 2022).

Determining Effective Factors Regarding Weather and Some Types of Air Pollutants in Seasonal Changes of PM10 Concentration Using Tree-Based Algorithms in Yazd City

Introduction: This study was carried out with the aim of determining weather parameters and air pollutants affecting seasonal changes of particulate matter of less than 10 microns (PM10) in Yazd city using Random Forest (RF) and extreme gradient boosting (Xgboost) models.
 Materials and Methods: The required data was obtained from 2018 to 2022. Levene’s test was applied to investigate the significant difference in the variance of PM10 values in 4 different seasons, and Boruta algorithm was used to select the best predictive variables. RF and Xgboost models were trained using two-thirds of the input data and were tested using the remaining data set. Their performance was evaluated based on R2, Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), and Nash–Sutcliffe Model Efficiency Coefficient (NSE).
 Results: The RF showed a higher performance in predicting PM10 in all the study seasons (R2 > 0.85; RMSE < 22). The contribution of dust concentration and relative humidity in spring PM10 changes was more than other variables. For summer, wind direction and ozone were identified as the most important variables affecting PM10 concentration. In the autumn and winter, air pollutants and dust concentration had the greatest effect on PM10, respectively.
 Conclusion: RF model could explain more than 85% of PM10 seasonal variability in Yazd city. It is recommended to use the model to predict the changes of this air pollutant in other regions with similar climatic and environmental conditions. The results can also be useful for providing suitable solutions to reduce PM10 pollution hazards in Yazd city

A Simulation–Optimization System to Assess Dam Construction with a Focus on Environmental Degradation at Downstream

AbstractThe present study proposes and evaluates an integrated framework to assess dam construction and removal, encompassing the simulation of downstream river habitats and reservoir operation in three distinct statuses: conventional reservoir operation optimization, optimal release considering environmental aspects within the optimization model, and natural flow conditions. Fuzzy physical habitat simulation was employed to assess physical habitats, while an ANFIS-based model was utilized to simulate thermal tension and dissolved oxygen tension at downstream habitats. Particle swarm optimization was applied in the optimization models. To evaluate the effectiveness of the proposed framework, results from the optimization system as well as habitat suitability models in the natural flow and current condition were compared using various measurement indices, including the reliability index, vulnerability index, the Nash–Sutcliffe model efficiency coefficient (NSE), and root mean square error (RMSE). The case study results suggest that the reliability of water supply may be diminished under optimal release for environmental and demand considerations. Additionally, optimal release for the environment may not adequately protect downstream aquatic habitats. Therefore, in cases where the preservation of downstream habitats is a priority, dam removal may be a logical solution. Moreover, it is essential to acknowledge that the main limitation of the proposed method is its high computational complexity.

Assessment of Hydrological Responses to Land Use and Land Cover Changes in Forest-Dominated Watershed Using SWAT Model

Recognizing how human activities affect hydrological systems is vital for the sustainable preservation and effective management of water resources in the watershed. Hence, this paper focuses on the hydrological response to land use and land cover (LULC) change scenarios in the Anyang watershed, South Korea. We obtained LULC data maps for the years 2000, 2013, and 2022 from the local government, revealing significant changes over the years. Agricultural lands experienced a 6.2% increase from 2000 to 2022, and pastureland expanded by 8.67% over two decades. The SWAT model was utilized to assess the impact of LULC on the hydrological components of the study watershed. Model calibration and validation for each LULC change were carried out using the SWAT-CUP program, considering the recorded streamflow information of the region. An excellent agreement was reached between the simulated and measured streamflow in both the calibration and validation stages under various LULC conditions. The Nash–Sutcliffe model efficiency (NSE), the objective function, demonstrated values of 0.9, 0.89, and 0.89 during the calibration for 2000, 2013, and 2022, respectively, in the LULC scenario, while for the validation, we obtained values of 0.82, 0.78, and 0.80 for 2000, 2013, and 2022, respectively. Our findings indicate that the surface runoff rise contributed much to the water yield increase over the two decades compared to the other components in terms of the water yield, while the contribution of evapotranspiration (ET) to the watershed hydrological cycle declined by 1.66% from 2000 to 2022. The southeastern sub-basin part showed a high groundwater recharge distribution due to agricultural land, rice area, and forest area changes.

Quantifying Uncertainty in Runoff Simulation According to Multiple Evaluation Metrics and Varying Calibration Data Length

In this study, the uncertainty in runoff simulations using hydrological models was quantified based on the selection of five evaluation metrics and calibration data length. The calibration data length was considered to vary from 1 to 11 years, and runoff analysis was performed using a soil and water assessment tool (SWAT). SWAT parameter optimization was then performed using R-SWAT. The results show that the uncertainty was lower when using a calibration data length of five to seven years, with seven years achieving the lowest uncertainty. Runoff simulations using a calibration data length of more than seven years yielded higher uncertainty overall but lower uncertainty for extreme runoff simulations compared to parameters with less than five years of calibration data. Different uncertainty evaluation metrics show different levels of uncertainty, which means it is necessary to consider multiple evaluation metrics rather than relying on any one single metric. Among the evaluation metrics, the Nash–Sutcliffe model efficiency coefficient (NSE) and normalized root-mean-squared error (NRMSE) had large uncertainties at short calibration data lengths, whereas the Kling–Gupta efficiency (KGE) and Percent Bias (Pbias) had large uncertainties at long calibration data lengths.

SAMM version 1.0: a numerical model for microbial- mediated soil aggregate formation

Abstract. Maintaining soil organic matter (SOM) is crucial for healthy and productive agricultural soils and requires understanding at the process level, including the role of SOM protection by soil aggregates and the connection between microbial growth and aggregate formation. We developed the Soil Aggregation through Microbial Mediation (SAMM) model, to represent this important connection. The pools of SAMM are fully measurable, and we calibrated and evaluated it against data from a long-term bare fallow experiment in a tropical sandy soil. This experiment received additions of plant litter of different compositions, which resulted in twice the soil carbon stocks in the best treatment compared to the control (about 8 vs. 4 t C ha−1 in 0–15 cm soil depth) after 25 years. As hypothesized, the SAMM model effectively represented the microbial growth response after the addition of litter and the subsequent formation and later destabilization of aggregates. The low correlations between different calibrated model parameters (r<0.5 for all parameters; r>0.4 for only 4 of 22) showed that SAMM is parsimonious. SAMM was able to capture differences between treatments in soil organic carbon (Nash–Sutcliffe modeling efficiency (EF) of 0.68), microbial nitrogen (EF of 0.24), and litter carbon (EF of 0.80). The amount of carbon within the aggregates (EF of 0.60) and in the free silt and clay fraction (EF of 0.24) was also simulated very well to satisfactorily. Our model results suggested that in spite of the sandy soil, up to 50 % of carbon stocks were stabilized through aggregate protection mechanisms; and that microbial and physical aggregate formation coexists. A version of the SAMM model without aggregate protection (SAMMnoAgg) initially failed to stabilize soil organic carbon (EF decreased to −3.68) and the simulation of microbial nitrogen worsened (EF of 0.13). By recalibrating SAMMnoAgg, it was possible to partially correct for the lack of aggregate protection by reducing the rate of mineral-attached carbon decomposition by about 85 % (EF of 0.68, 0.75, and 0.18 for SOC, litter carbon, and microbial nitrogen, respectively). However, the slightly better evaluation statistics of SAMM (e.g., Akaike information criterion of 5351 vs. 5554) suggest that representing aggregate dynamics in SOM models can be beneficial and necessary to understand the mechanism behind SOM dynamics. Our results indicate that current models without aggregate formation partly compensate for the absence of aggregate protection by lowering the turnover rates of other pools. Thus, they remain suitable options where data on aggregate associated carbon are not available.

Predicting Microbiologically Influenced Concrete Corrosion in Self-Cleansing Sewers Using Meta-Learning Techniques

Sewer networks are not only necessary as an infrastructure for human societies, but they can also help humans achieve a stable situation with the surrounding natural environment by controlling and preventing the spread of pollution in the environment. As a result, concrete sewer maintenance and analysis of their damaging elements are critical. In this regard, modeling microbiologically influenced corrosion (MIC) is a challenging phenomenon. Due to the complicated aspects related to the interaction of microorganisms and concrete degradation, this research suggests several machine-learning models as well as traditional multiple linear regression model to predict the MIC in sewer pipelines. The models can be categorized into three sections: (i) stand-alone models (group method of data handling, generalized regression neural network, radial basis function neural network, multilayer perceptron neural network, chi-square automatic interaction detection, and classification and regression tree); (ii) integrative models (adaptive neuro-fuzzy inference system and support vector regression with particle swarm optimization, artificial bee colony, and firefly algorithm); and (iii) ensemble meta-learner stepwise regression (SR) model. After implementing the models, statistical measures, including root mean square error, mean absolute error, mean bias error, Pearson correlation coefficient, and Nash-Sutcliffe model efficiency are considered for evaluating models’ performances. The results indicate that the ensemble meta-learner-SR model is significantly more precise than other models. They also demonstrate that using an integrative model can improve the accuracy of stand-alone models by at least up to 42%. The durability and lifespan of the sewer system are also estimated with the aid of the best predictive model (meta-learner-SR) for two scenario cases of (i) gas phase and (ii) submerged conditions. It is concluded that the sewer systems have a considerably lower life span (24 y less) exposed to submerged sewage than the gas phase with 56 y of durability.

Analysis of Crop Irrigation Water Requirements and Water Scarcity Footprint in the Beijing–Tianjin–Hebei Region Based on the GeoSim–AquaCrop Model

To reduce crop-related water consumption and enhance agricultural water resource efficiency in the Beijing–Tianjin–Hebei region, this study employed the AquaCrop model to simulate crop yield and irrigation water requirements and calculated the water scarcity footprint (WSF). The results were as follows: (1) The AquaCrop model exhibited strong applicability, with R2, RMSE (Root Mean Square Error), EF (Nash–Sutcliffe model efficiency coefficient) and d values of 0.9611, 6.6%, 0.91, and 0.98 (winter wheat), and 0.9571, 5.5%, 0.95, and 0.99 (summer maize) for canopy cover simulation. Similarly, aboveground biomass simulation yielded values of 0.9661, 0.8 t/ha, 0.93, and 0.98 (winter wheat), and 0.9087, 1.3 t/ha, 0.90, and 0.98 (summer maize). Winter wheat soil moisture content simulation showed an R2 of 0.9706, RMSE of 3.7 mm, EF of 0.93, and d of 0.98. (2) The AquaCrop model simulated the winter wheat and summer maize yields and irrigation water requirements for the years 2009, 2014, and 2019, validating the scalability and spatial visualization capabilities of GeoSim in extending AquaCrop simulations. (3) Integrating the water footprint and the water resources system, this study assessed the WSFs of winter wheat and summer maize. From 2009 to 2019, winter wheat production in the region increased by 25.08%, and summer maize production increased by 37.39%. The WSF of winter wheat decreased, whereas the WSF of summer maize increased. It is recommended to reduce crop cultivation areas in regions such as Daming County, Ningjin County, and Dingzhou City while further improving irrigation water efficiency, which would facilitate the sustainable utilization of water resources in the area.

Drought Prediction with Feature Enhanced LSTM Model using Metaheuristic Optimization Algorithms

The impact of drought builds on all three fronts of economy, environment, and society is devastating. Predicting its arrival and duration is highly important to arrange any sort of mitigation plans. The association of detailed relationship between multiple variables makes drought prediction a highly complex task. Especially influence of global warming, polar sea extent variations and their influence on overall ocean temperature have altered the seasonal rainfall behaviors all over the world. In the midst of it, predictions centered on the history of rainfall levels become inaccurate. The proposed system is an optimized deep learning prediction model integrating indigenous knowledge (IK) is proposed to predict the drought. IK expressed in human language is translated using fuzzy function and fed to an improved Long Short Term Memory (LSTM) model. The LSTM model hyperparameters are optimized using a hybrid of Particle Swarm Optimization (PSO) with firefly to produce the meta-heuristics algorithm which will provide the best performance in presence of integration of IK features into modern meteorological features which solves the problem of local minima in LSTM hyperparameter optimization. The performance of the proposed results were tested compared with the meteorological information gathered by the Karnataka Natural Disaster Monitoring Centre (KNDMC) for the district named Chitradurga of the Karnataka state in India. The proposed system which is Indigenous Knowledge merged along the cross model attention network can produce at least 1.4% higher Nash–Sutcliffe model efficiency coefficient (NSE) and 30% lower Mean Absolute Error (MAE) in the prediction of Standard Precipitation Index (SPI) compared to Convolution Neural Networks (CNN) and LSTM based time series prediction models.

LSTM-Based Data Integration to Improve Snow Water Equivalent Prediction and Diagnose Error Sources

Abstract Accurate prediction of snow water equivalent (SWE) can be valuable for water resource managers. Recently, deep learning methods such as long short-term memory (LSTM) have exhibited high accuracy in simulating hydrologic variables and can integrate lagged observations to improve prediction, but their benefits were not clear for SWE simulations. Here we tested an LSTM network with data integration (DI) for SWE in the western United States to integrate 30-day-lagged or 7-day-lagged observations of either SWE or satellite-observed snow cover fraction (SCF) to improve future predictions. SCF proved beneficial only for shallow-snow sites during snowmelt, while lagged SWE integration significantly improved prediction accuracy for both shallow- and deep-snow sites. The median Nash–Sutcliffe model efficiency coefficient (NSE) in temporal testing improved from 0.92 to 0.97 with 30-day-lagged SWE integration, and root-mean-square error (RMSE) and the difference between estimated and observed peak SWE values dmax were reduced by 41% and 57%, respectively. DI effectively mitigated accumulated model and forcing errors that would otherwise be persistent. Moreover, by applying DI to different observations (30-day-lagged, 7-day-lagged), we revealed the spatial distribution of errors with different persistent lengths. For example, integrating 30-day-lagged SWE was ineffective for ephemeral snow sites in the southwestern United States, but significantly reduced monthly-scale biases for regions with stable seasonal snowpack such as high-elevation sites in California. These biases are likely attributable to large interannual variability in snowfall or site-specific snow redistribution patterns that can accumulate to impactful levels over time for nonephemeral sites. These results set up benchmark levels and provide guidance for future model improvement strategies.

Predicted Contribution of Snowmelt to Subsurface Drainage Discharge in Two Subsurface-Drained Fields in Southern Quebec and Ontario

Highlights RZ-SHAW model delivered a satisfactory simulation for winter subsurface drainage discharge. Snowmelt dominated the drainage discharge during the winter and spring snowmelt seasons in southern Quebec. Snowmelt contributed to 29% and 18% of winter and annual drainage discharge, respectively, in southern Ontario. Abstract. Late winter/early spring has been recognized as a critical period for subsurface drainage discharge and associated nutrient losses due to snowmelt and rainfall in croplands under cold, humid climates in North America and Europe. Although the detachment and transport processes of soil particles under snowmelt and rainfall are known to be different, studies quantifying the contribution of snowmelt to winter drainage discharge are limited. This study aims to investigate the contribution of snowmelt to subsurface drainage discharge at two winterized experimental sites in southern Quebec and Ontario, Canada, using observed data and the Root Zone Water Quality Model-Simultaneous Heat and Water model (RZ-SHAW model). The RZ-SHAW model was calibrated and validated against the measured snow depth and year-round subsurface drainage discharge data from 2021 to 2023 at St. Emmanuel, Quebec, and from 2000 to 2003 at Harrow, Ontario. RZ-SHAW demonstrated satisfactory Nash-Sutcliffe model efficiency values (NSE = 0.50) for simulating snow depth, soil temperature, and winter subsurface drainage discharge. The calibrated RZ-SHAW model was used to simulate the contribution of snowmelt to subsurface drainage for the two sites from 1990 to 2022. The snowmelt was predicted to contribute 55% and 44% of the winter and annual subsurface drainage discharge for the cropland in Southern Quebec. In contrast, for the southern Ontario site, the contribution of snowmelt to winter and annual subsurface drainage discharge was 29% and 18%, respectively. Considering that snowmelt in Southern Quebec contributed a significant fraction of winter and annual subsurface drainage discharge, more attention should be devoted to adapting and developing new winter management for nutrient loss in future research in the region. Keywords: Cold region, Cropland drainage, Drainage partition, RZ-SHAW, RZWQM2, Snowmelt, Soil freeze and thaw, Winter subsurface drainage.

Extracting operation rule of cascade reservoirs using a novel framework considering hydrometeorological spatiotemporal information based on artificial intelligence models

The scientific operation of cascade reservoirs is of great significance for completely utilizing water resources and maximizing their comprehensive benefits. The extraction of reservoir operation rules from massive historical operational data still remains a challenge. Therefore, a novel framework for extracting the operation rules of cascade reservoirs was proposed in this study based on artificial intelligence models by considering hydrometeorological spatiotemporal information. First, the observed precipitation was obtained by merging multi-source precipitation products through a convolutional long-short-term memory (ConvLSTM) network. Second, a novel extraction model by considering hydrometeorological spatiotemporal information was established, comprising a process of capturing the spatiotemporal information of interval precipitation using ConvLSTM and a feature aggregation process of aggregating precipitation and reservoir historical operation information using an LSTM network. Finally, three different extraction strategies (recursive, direct, and multi-input multi-output (MIMO) strategies) for cascade reservoir operation rules were developed and compared, and the influence of future multi-step forecast information was analyzed. Considering the example of Yalong River, the results are as follows: (1) the proposed extraction model has excellent fitting performance with a Nash-Sutcliffe model efficiency coefficient (NSE) of more than 0.95 and mean relative error (MRE) of less than 10 %, especially in reservoirs with large regulation capacity; (2) according to the regulation capacity of the reservoirs and reliability of forecast precipitation, appropriately adding multi-step forecast information can improve the performance of the extraction models; (3) each reservoir's NSE when using the three strategies is above 0.9, indicating that they can all be used to extract the operation rules of the cascade reservoir. Among them, the direct strategy is optimal.

Mitigating Nitrate Concentration Using an Optimal Cropping Pattern Developed by Linking a Soil and Water Assessment Tool with Evolutionary Optimization

The unplanned development of agricultural land and urban areas poses threats to water quality, which can lead to the death of the aquatic species in rivers. The present study developed a novel framework by combining a soil and water assessment tool (SWAT) and evolutionary algorithms to optimize the cultivation pattern at the catchment scale in the Tajan River basin, with the aim of mitigating the environmental impacts of surface runoff from farms. This river basin is located in northern Iran, where quick agricultural development is one of the environmental challenges. We utilized a SWAT to simulate the nitrate concentrations for different crops at the river basin scale by applying the Nash–Sutcliffe model efficiency coefficient (NSE) as a measurement index. Then, a novel model was developed to optimize the cultivation pattern by applying different metaheuristic algorithms. Fuzzy technique for order of preference by similarity to ideal solution (TOPSIS) was applied as a decision-making system to select the best optimization algorithm. The results demonstrated that using a SWAT in the optimization model structure is a robust method for the design of an optimal cultivation pattern. The Nash–Sutcliffe model efficiency coefficient (NSE) was 0.74, demonstrating the robust predictive skills of the water quality model. The decision-making system indicated that particle swarm optimization and shuffled complex evolution were the best evolutionary algorithms to optimize the cultivation pattern using the proposed method. The proposed method opens a new window regarding the optimization of cultivation patterns in agriculture and provides an environmental-based optimization to design cultivation patterns on the catchment scale.

Selecting a Suitable Model for Roundabout Entrance Capacity Estimation: A Case Study

Abstract In India, urban vehicular traffic is highly heterogeneous; it appears to be extremely challenging to accurately anticipate roundabout capacity. The main objective of this paper is to develop two models which are predicting the capacity of roundabouts in the Indian scenario. One is Weighted Least Square Regression (WLSR), a linear regression is carried out using weights of each data variable while the other one is Self Adaptive Segregative Genetic Algorithm with Simulated Annealing (SASEGASA), which is a combination of Segregative Genetic Algorithm (SEGA) and self-adaptive selection pressure with the application of simulated annealing. The coefficient of determination (R2) and Nash-Sutcliffe model efficiency coefficient (E) for the models like SASEGASA and WLSR were obtained as (0.947, 0.949) and (0.906, 0.905) respectively. Ranking among the proposed models is carried out using a complex phenomenon known as Modified Ranking Index (MRI), where the highest rank generated from five estimates is chosen as the best model for this study. SASEGASA is ranked as the appropriate model to estimate the roundabout capacity as per MRI. As compared to previously used models, SASEGASA model is quite relevant for field application.

Estimation of the Water Level in the Ili River from Sentinel-2 Optical Data Using Ensemble Machine Learning

Monitoring of the water level and river discharge is an important task, necessary both for assessment of water supply in the current season and for forecasting water consumption and possible prevention of catastrophic events. A network of ground hydrometric stations is used to measure the water level and consumption in rivers. Rivers located in sparsely populated areas in developing countries of Central Asia have a very limited hydrometric network. In addition to the sparse network of stations, in some cases remote probing data (virtual hydrometric stations) are used, which can improve the reliability of water level and discharge estimates, especially for large mountain rivers with large volumes of suspended sediment load and significant channel instability. The aim of this study is to develop a machine learning model for remote monitoring of water levels in the large transboundary (Kazakhstan-People’s Republic of China) Ili River. The optical data from the Sentinel-2 satellite are used as input data. The in situ (ground-based) data collected at the Ili-Dobyn gauging station are used as target values. Application of feature engineering and ensemble machine learning techniques has achieved good accuracy of water level estimation (Nash–Sutcliffe model efficiency coefficient (NSE) >0.8). The coefficient of determination of the model results obtained using cross-validation of random permutations is NSE = 0.89. The method demonstrates good stability under different variations of input data and ranges of water levels (NSE > 0.8). The average absolute error of the method ranges from 0.12 to 0.18 meters against the background of the maximum river water level spread of more than 4 meters. The obtained result is the best current result of water level prediction in the Ili River using the remote probing data and can be recommended for practical use for increasing the reliability of water level estimation and reverse engineering of data in the process of river discharge monitoring.

A methodology to improve the accuracy of Total phosphorous diffuse load estimates from agroforestry watersheds

This study was undertaken to improve the accuracy of Total Phosphorus export coefficient models; an essential need for the water management process. The modeling approach used in this study was supported by the application of four different models to 27 agroforestry watersheds located in the Mediterranean region. The first model definition relies on the correlation between the time of concentration and the Total Phosphorus export coefficient for each watershed. The second model predicts nutrient loads in the watersheds considering export coefficients reported in the literature and watershed land use areas. As an alternative approach, the remaining two models solve a system of equations through the application of a Bayesian optimization algorithm that iteratively modifies the land use export coefficients to minimize the error between the observed and predicted Total Phosphorus export coefficients. The difference between both models relies on the constraints imposed to the variation of the export coefficients. The modeling approach also included watershed clustering and the consideration of a correction factor, computed as absolute differences (observation-prediction). This correction factor encapsulates the widely accepted conclusion that the sediment delivery ratio of a watershed tends to decrease in line with an increase in drainage area, and that a significant amount of the Total Phosphorus transported in rivers is in particulate form and is primarily correlated to suspended solids. The results of the study clearly show that the consideration of Total Phosphorus export coefficients obtained from the literature for watersheds located in the Mediterranean region produce very poor Total Phosphorus diffuse load estimates. When considering the results of all the models, the Percentual bias and Nash-Sutcliffe model efficiency values varied from −80.2 % to 24.7 % (μ = -52.5 ± 20.2) and −8.8 to −1.8 (μ = -4.3 ± 2.8). However, the modeling approach considered in this study generated significant improvements in terms of the accuracy of the Total Phosphorus export coefficient from different land use classes and, consequently, the overall estimation of Total Phosphorus diffuse loads from agroforestry watersheds. When considering the results of all the models, the Percentual bias and Nash-Sutcliffe model efficiency values varied from −13.3 % to −5.0 % (μ = -9.0 ± 2.9) and 0.5 to 0.9 (μ = 0.8 ± 0.2), respectively. Overall, the results of this study suggest that the solution proposed has the potential to significantly improve the quantification of Total P diffuse loads at the watershed level, which is an important conclusion given the present and future challenges currently being faced with regard to the water management process.

Groundwater level forecasting in a data-scarce region through remote sensing data downscaling, hydrological modeling, and machine learning: A case study from Morocco

Study regionThe upstream part of the Essaouira basin, a data-scare region in Morocco, Northwestern Africa. Study focusThe scarcity of hydro-climate data is a significant challenge found in several regions worldwide, where qualitative and quantitative water resource information remains limited. Estimating and predicting groundwater levels (GWL) in such areas is a significant challenge in producing knowledge for effective water resource management. To address this issue, the present study aimed to use the Soil and Water Assessment Tool (SWAT) model in conjunction with downscaled total water storage (TWS) data (9 km) obtained from Gravity Recovery And Climate Experiment (GRACE) and machine learning techniques, specifically random forest (RF) and support vector machine (SVM), to estimate and predict the variation in GWL. New hydrological insights for the regionThis study constitutes a first of its kind in the study area; the SWAT model was set up for 10 years, with a warm-up period from 2000 to 2001, calibration from 2002 to 2007, and validation from 2008 to 2010. The statistical indices (Coefficient of Determination (R²) ≥ 0.73, R² ≥ 0.78, Nash–Sutcliffe model efficiency coefficient (NSE) ≥ 0.67, NSE ≥ 0.80 respectively for calibration and validation) highlight a significant correlation, implying the model's capability to faithfully reproduce the streamflow. The downscaled TWS demonstrates an impressive ability to identify and monitor fluctuations in GWL. Using machine learning algorithms (RF and SVR), the prediction of GWL yielded satisfactory results, NSE = 0.78 and root mean square error (RMSE) = 0.33, NSE = 0.51 and RMSE = 0.49 for the RF and SVR, respectively. Despite some limitations, our approach provided promising results in GWL prediction, with the possibility of expanding to other data-scarce regions.