Parameter Uncertainty Analysis Research Articles

A Monte Carlo-based Workflow for Geochemical Anomaly Identification Under Uncertainty and Global Sensitivity Analysis of Model Parameters

A Monte Carlo-based Workflow for Geochemical Anomaly Identification Under Uncertainty and Global Sensitivity Analysis of Model Parameters

Thermal performance enhancement on a box-type solar cooker using a triangular fin over a conventional cooking pot

The thermal performance of existing box-type solar cookers is limited and to improve the performance many research works reported on modification of solar cooker sub-systems i.e., cylindrical fin over the cooking pot, extra air duct, integrating with a photovoltaic panel, incorporation of heat storage materials and multiple reflectors. This paper presents a novel study of a triangular-finned cooking pot to increase the convective heat transfer rate and solar irradiance over the cooking pot. The thermal network diagram is presented and thermal losses are measured using MATLAB. The thermal visualization is presented with the help of COMSOL Multiphysics 5.5 and uncertainty analysis of different performance parameters is completed with Engineering Equation Solver (EES). The simulation results found thermal losses occur from the top, bottom and side are 4.26, 0.77 and 0.64 W/m2 respectively. The experimentally observed result indicates the cooking time is reduced from 94–95 min to 85–86 min during the boiling test of 1.5 kg water for the triangular finned cooking pot. The first figure of merit is 0.139 and the second figure of merit is 0.563 respectively for the same loading conditions. The cooking power is increased from 68.71 W to 77.06 W. The thermal efficiency is increased from 36.16 % to 40.55 %. The uncertainty analysis emphasizes the value of the performance parameter lies within an acceptable range. From thermal visualization, It is concluded that wall edges are less affected by heat flow and the correct placement of the cooking pot is necessary for uniform heating.

Comprehensive analysis of model parameter uncertainty influence on evaluation of HVAC operation to mitigate indoor virus: A case study for an office building in a cold and dry climate

Simulation-based studies of HVAC operation to mitigate indoor virus have been conducted to understand tradeoffs between indoor air quality (IAQ) and energy consumption. However, the influence of model parameter uncertainty in these studies has not been systematically quantified, which is critical when providing guidance to building operators. To address this gap, we identify 20 model parameters for a typical medium office building system in a cold and dry climate that can influence IAQ and energy consumption for indoor virus scenarios. Next, the distributions of the parameter values are estimated from literature and a set of simulation samples for three representative days is generated that simultaneously sample all of the parameters from their distributions. Two HVAC virus mitigation strategies are studied: increased filtration using MERV 13 filters and increased ventilation, via supply of 100% outdoor air into buildings. The model parameter uncertainty leads to significant variability in the results, particularly for the IAQ because of the highly uncertain virus generation rate. Use of 100% outdoor air can be beneficial for some uncertain scenarios on the hot day, but shows less IAQ improvement and/or significant energy increases on the other days. The virus removal efficiency and pressure drop of the HVAC filter, fan efficiency, and internal heat gain are the most important parameters to determining the tradeoffs of the two strategies. The results demonstrate how this model parameter uncertainty analysis methodology can provide practical guidance to building operators.

Environmental Footprints of the Catalytic Pyrolysis of Pine Needles through Integration of a Nickel-Decorated Chemically Modified Biochar Catalyst

The current study presents the life cycle analysis (LCA) of a biochar-catalyzed pyrolysis-based biorefinery system. Noncondensable gases (NCGs), produced as a coproduct, have been recycled back to the system, while biochar produced was employed for catalyst preparation, to visualize a biorefinery model. Results showed the considerable environmental impact of the process, in particular of the catalyst and energy sources utilized during the process. Matching up with the LCA metrics, both single parameter and Monte Carlo uncertainty analyses also revealed the high sensitivity of the obtained impacts for the impregnating chemical (i.e., ZnCl2, H3PO4, and NaOH) employed during catalyst preparation. Among different processes, the Ni/BC-H3PO4 catalyzed process with NCG recycling has emerged as the best possible case with the lowest environmental emissions (GWP ≈ 0.109 kg CO2 equivalent) and low cumulative energy demand (∼2.47 MJ) together with high process efficacy and productivity. Furthermore, sustaining process energy needs with varieties of different sources advocates for renewable sources (more specifically hydropower) over nonrenewable sources of energy. The study highlights the hotspots of the current technology, envisaging ways of reducing emissions through clean/renewable energy sources as well as utilizing less impact-causing intermediate materials, and thus, acting like a building stone in creating a base toward the vision of achieving net zero emissions.

Forest fire susceptibility mapping with sensitivity and uncertainty analysis using machine learning and deep learning algorithms

In the hilly region of the Western Himalayas, forest fires play a crucial role in forest destruction and biodiversity loss. Therefore, addressing the problem of forest fires is an immediate priority to mitigate the impact of forest fires. Therefore, the objective of this research is to assess and map the susceptibility zones of forest fire in the hilly state of Uttarakhand using geospatial techniques and machine learning algorithms (MLA) and deep Learning based sensitivity and uncertainty analysis of the forest fire ignition parameters. Thirteen forest fire ignition parameters were used to map forest fire susceptibility. Further, we used five distinct MLAs, including artificial neural networks (ANN), random forest (RF), logistic regression (LR), support vector machine (SVM) and an ensemble machine learning model. The generated models were validated using the receiver operating characteristic (ROC). The result shows that the ''very high'' and ''high'' zones for forest fires cover almost 55% of the total area of Uttarakhand. Validation of the models used indicated that all models produced most significant results. The ensemble machine learning had the highest accuracy (AUC = 0.977), followed by RF (0.973) and ANN (0.972). In addition, a Deep Neural Networks (DNN) based sensitivity approach is proposed to perform sensitivity and uncertainty analysis of the ignition parameters. For this, we have selected the best predictive model (Ensemble Machine Learning) to evaluate the sensitivity and uncertainty analysis. The sensitivity analysis shows that evapotranspiration has the highly influencing parameter to forest fires, followed by annual rainfall, wind speed, distance to agriculture, distance to tourist spots, distance to built-up and distance to road. However, aspect, temperature and distance to religious centers have less influence on susceptibility to forest fires in this study area. This study is unique because it provides new insights into the construction of an ensemble machine learning model and the integration of DNN-based sensitivity and uncertainty for robust and scientific modelling of forest fire susceptibility. The contribution of parameters demonstrates a novel approach in understanding the impact of a single parameter sample on forest fire susceptibility, which will help in developing prevention strategies. A similar approach may be used in other parts of the world with similar natural and anthropogenic conditions for mapping forest fire susceptibility.

Parametric MELCOR 2.2 sensitivity and uncertainty study with a focus on aerosols based on Phébus test FPT1

The Phébus test FPT-1 was carried out to investigate the fission product release, transportation, and distribution in the reactor coolant system and the simulated containment during a severe accident. The test was divided into three phases: the core degradation, aerosol transport, and containment phenomena. In this work, a parametric sensitivity and uncertainty analysis of the two first phases, i.e., core degradation and aerosol transport, is conducted with the focus on aerosol parameters.The study was done in a frame of WP4 of the European MUSA project whose main objective is to develop a methodology for severe accident uncertainty analysis.Due to the wide scope of the MUSA project, this study focuses not only on the UQ analysis but also on the parametric dependency and data interpretation. The main subject was investigating the uncertainty of the MELCOR 2.2 aerosol models and the relation between the chosen parameters as well as evaluating the usability of essential analytical tools available for such analyses.The calculations were carried out using the integral severe accident code MELCOR2.2 and the generic input deck of the FPT-1 experiment. The uncertainty quantification tool DAKOTA was used as a statistical tool. It was coupled with MELCOR within the SNAP environment.Fifteen uncertain parameters related with the aerosol release and transport were investigated. Two figures of merit were chosen to be the amount of deposition of Cs in the reactor coolant circuit, and the mass of air-borne aerosol in the containment. Preliminary analysis showed a significant correlation between the aerosol dynamic and agglomeration shape factors, density and minimum aerosol diameter with both FOMs. In addition to the standard uncertainty analysis, five MELCOR/DAKOTA UQ sensitivity calculations with modified parameters were performed and analyzed. Additionally, the impact of the parameters was investigated by excluding the most correlated ones from the analysis. The results confirm the strong correlation of shape factors and the importance of their ranges for overall UQ results.The authors show that by coupling the chosen analytical tools together, it is possible to carry out uncertainty quantification of severe accident analysis.Based on the study performed, the authors underline the need for further investigation of parameter distributions, the effect of the chosen number of samples on the results, and a global time-dependent UQ analysis.

Decision Support System for Sustainable Exploitation of the Eocene Aquifer in the West Bank, Palestine

Groundwater is a crucial resource for water supply and irrigation in many parts of the world, especially in the Middle East. The Eocene aquifer, located in the northern part of the West Bank, Palestine, is threatened by unsustainable groundwater abstractions and on-ground pollution. Analysis and management of this aquifer are challenging because of limited data availability. This research contributes to the long-term sustainability of the aquifer by model-based design of future abstraction strategies considered within an uncertainty analysis framework. The methodology employed started with development of a single-layer steady-state MODFLOW groundwater model of the area, followed by uncertainty analysis of model parameters using Monte Carlo simulations. The same model was afterwards coupled with a Successive Linear Programming (SLP) optimization algorithm, implemented in the Groundwater Management tool (GWM) of the United States Geological Survey (USGS). The purpose of optimization was deriving five optimal abstraction strategies, each aiming to maximize groundwater abstraction, subject to different constraints regarding groundwater depletion. Given the uncertainty of model parameters, the sensitivity and reliability of these optimal strategies were then tested. Sensitivity was checked for two optimal strategies by performing re-optimization with different values of uncertain model parameters (one at a time). Reliability of the five strategies was tested by analyzing the extent of constraints’ violation for each strategy when varying the uncertain parameters using Monte Carlo simulations. Finally, the model was used for determining capture zones of wells for the five optimal abstraction strategies, land-use in these capture zones, and the associated estimates of on-ground nitrogen loading. The developed strategies were then deployed in a web-based decision support application (named Groundwater Decision Support System—GDSS), together with other relevant information. Users can analyze results of different optimal strategies in terms of groundwater level variations and total water balance results, and test consequences of uncertain parameters. Capture zones of wells for different abstraction strategies, together with land-use and on-ground nitrogen loading in these capture zones, are also presented. Results show that critical uncertain parameters are recharge, hydraulic conductivity, and conductance at key boundary condition locations. Optimal abstraction strategies results indicate that an increase in total abstractions could be between 5% and 20% from the current level (estimated at about 56 × 106 m3/year, which is about 74% of estimated annual recharge). The uncertain parameters, however, are impacting the sensitivity and the reliability of the optimal strategies to variable degrees. Recharge and hydraulic conductivity are the most critical uncertain parameters regarding sensitivity of the optimal strategies, while reliability is also impacted by the level of abstraction proposed in a given strategy (number, locations, and abstraction rates of new wells). The main novelty and contribution of this research is in combining modelling, uncertainty analysis, and optimization techniques in a comprehensive decision support system for the area of the Eocene aquifer, characterized with limited data availability.

Shape adjustment for uncertain mesh reflectors using machine learning

Mesh reflectors are uncertain structures due to manufacturing and assembly errors. Shape adjustment must be made to meet the surface precision requirement of space missions by changing the lengths of adjustable cables. It is desirable to find the optimum adjustment amount based on the current state of the actual prototype. However, it is hard to build an accuracy simulation model reflecting mechanical behavior of the actual prototype because only part information about the actual prototype can be measured. Therefore, it is necessary to deal with the parameter uncertainty of the actual prototype in order to provide a reliable structure state to guide the shape adjustment. This field has not been well studied because it involves an uncertainty analysis of parameters and the nonlinearity of high coupling of node displacements. To improve the efficiency of shape adjustment of uncertain mesh reflectors, this paper combines finite element model, actual prototype, and machine learning algorithm to develop a prediction model for precision analysis of the mesh reflector under multi-uncertainties using a machine learning algorithm (extreme gradient boosting), with the reflector surface node position as the input and the cable length adjustment amount required to displace it to a high precision surface as the output. First, based on the nominal design model, the sample data is collected by the Latin hypercube design method; the nonlinearity of the structure was taken into account by iterative solution during the data sampling, and the uncertain models containing different random errors are fused from the perspective of data sampling. The Bayesian optimization algorithm is used to obtain the optimal hyperparameters that enables strong generalization of the prediction model. Next, learning from the idea of K-nearest neighbor algorithm, the prediction model is updated based on the actual prototype data, which further improves the prediction ability of the current prototype to be adjusted. The incremental learning algorithm is used to reduce the time-consuming of the prediction model updating. Finally, numerical examples and experiments show that the proposed method has higher adjustment efficiency and precision with less workload.

Uncertainty and Sensitivity Analysis of Hydrogen Source Term under Severe Accident of Marine Reactor

In order to explore the hydrogen source term characteristics under severe accidents of marine pressurized water reactors (PWR) and effectively assess the hydrogen risk, the best estimation program SCDAP/RELAP5/MOD3.2 is used to establish the marine reactor severe accident analysis model. Based on the Latin Hypercube sampling (LHS) method and the Wilks sampling theory, a set of methods for the uncertainty analysis of severe accidents is developed. This method can be applied to the uncertainty and sensitivity analysis of different target parameters. The phenomenon identification and ranking table (PIRT) under the severe accident induced by the break are established, and 14 uncertain parameters are selected as input variables. The established PIRT fills the gap in the uncertainty and sensitivity analysis of severe accidents of marine reactors and provides a reference for subsequent research. The quantitative uncertainty analysis of the calculation results is carried out, and the uncertainty range of hydrogen production is defined. The Spearman correlation coefficient is used to evaluate the sensitivity of input parameters, and the sensitivity of each parameter to hydrogen production is obtained. The results show that under the severe accident caused by the medium equivalent diameter break, the uncertainty range of hydrogen production in the zirconium–water reaction in the pressure vessel is 20.14 kg~22.19 kg with 95% confidence, and the fuel cladding thickness has a significant positive correlation on the hydrogen production.

Parameter range estimation and uncertainty analysis for high strain rate material response via model fusion

The dynamic stress–strain response of a material can be described by a number of different models of varying fidelity. However, an individual model’s ability to replicate the dynamic stress–strain response of a material can be hindered by experimental variability. Reification, an approach to fusing models and experimental data with inherent scatter, is presented. It is then used to determine the optimum parameters of the Johnson–Cook (JC) and Zerilli–Armstrong (ZA) models using a fusion of Split-Hopkinson Pressure Bar (SHPB) data and the JC and ZA models fit to the SHPB data using a traditional approach. The output of the fused model is a dataset that represents a “best-guess” sampling of the possible stress–strain response of a high strength steel. In the present work, the dynamic response of a newly developed steel, AF9628, is evaluated. Under the reification framework, the experimental variability and limitations of the mathematical model expressions are addressed by the optimized sampling of data and combined fitting process. The JC and ZA models are then re-fit to partitions of the fused dataset, which bound the responses of the traditionally fit JC and ZA models. The behavior of the re-fit models and the traditionally fit models are compared via a simulated Taylor anvil test.

Correlations on heat transfer rate and friction factor of a rectangular toothed v-cut twisted tape exhibiting the combined effects of primary and secondary vortex flows

It is generally recognized that the heat transfer rate can be improved using active, passive and compound methods. The use of twisted tape inserts in a pipe flow is one of the passive methods. The influence of swirl flow, primary and secondary vortex flow generated due to twist in tape, v-cut on the tape, and the teeth on the v-cut respectively on Nu (heat transfer rate), f (friction factor), and CPE (criterion for performance evaluation) of an inserted toothed v-cut twisted tape is investigated in this study. Experiments were carried out on a double pipe heat exchanger using water as the working medium for a Reynolds number range of 6000–13,000, with the uncertainty analysis of the relevant parameters taken into account. The correlations for Nu and f in terms of factors such as Re (Reynolds number), Pr (Prandtl number), b/c (ratio of depth of v-cut to width of cut), and e/c (ratio of depth of teeth on v-cut to depth of v-cut) are also developed using regression analysis. The results of correlations lie within ±10% for Nu whereas within ±15% for f. The investigations show that the inclusion of teeth on the v-cut develops a secondary vortex flow which further enhances the Nu (though not up to the comparable magnitudes of the primary vortex) and is capable of minimizing the percent decrement of heat transfer rate with increasing depth of cut at higher Re. The friction factor increased up to 7.7 times for e/c ratio equals 0.14 (depth of the tooth to depth of the v-cut) in comparison to the plain tube, while the increase in heat transfer rate for the same was 84% for Re = 6000.

Specifying the hydrogeological parameters range for a groundwater flow model in El-Qaa Plain aquifer, Sinai, Egypt

Hydrogeological parameters are very uncertain to specify in the groundwater flow models. Hence, it is important to specify a range for these parameters to be further utilized by the groundwater model users. In El-Qaa Plain, previous researchers have determined the hydrogeological parameters such as transmissivity, hydraulic conductivity, and mean annual recharge rate using geophysical methods, pumping tests, hydrological models, and groundwater flow models. However, they are also limited to specific locations and do not spatially distribute. In this paper, an effort has been accomplished to specify a specific range for the aquifer hydrogeological parameters using inverse groundwater modeling taking into account the spatial heterogeneity. These ranges could be taken into consideration during modeling groundwater flow. Hence, a finite-difference groundwater flow model has been applied using the ModFLOW model. Regularized pilot points method (RPPM) has been chosen as an inverse modeling technique to specify the accurate values of the aquifer parameters. Model-independent parameter estimation and uncertainty analysis (PEST) has been utilized to estimate the parameters. The results show that the geometric mean of the hydraulic conductivity in the study area ranges from proximity 1 to 2 m/day. Additionally, the geometric mean of the transmissivity ranges from nearly 200 to 500 m2/day. A relationship between the mean annual recharge rate and hydrogeological parameters has been determined. Besides, maps have been generated showing the spatial heterogeneity of the hydraulic conductivity. These maps could be further utilized by groundwater flow modelers to predict the potential of groundwater flow and pollution in the future.

How to keep it adequate: A protocol for ensuring validity in agent-based simulation

There has so far been no shared understanding of validity in agent-based simulation. We here conceptualise validation as systematically substantiating the premises on which conclusions from simulation analysis for a particular modelling context are built. Given such a systematic perspective, validity of agent-based models cannot be ensured if validation is merely understood as an isolated step in the modelling process. Rather, valid conclusions from simulation analysis require context-adequate method choices at all steps of the simulation analysis including model construction, model and parameter inference, uncertainty analysis and simulation. We present a twelve-step protocol to highlight the (often hidden) premises for methodological choices and their link to the modelling context. It is designed to aid modelers in understanding their context and in choosing and documenting context-adequate and mutually consistent methods throughout the modelling process. Its purpose is to assist reviewers and the community as a whole in assessing and discussing context-adequacy.

Recycling waste vehicle tyres into crumb rubber and the transition to renewable energy sources: A comprehensive life cycle assessment

This study conducts a comprehensive life cycle assessment (LCA) on converting waste vehicle tyres into recycled crumb rubber (CR) granules as an alternative polymer for enhancing asphalt properties. The LCA study has been performed on acquired industrial primary data by incorporating CR at different proportions of binder in one ton (1-ton) of asphalt mix following the wet method. The uncertainty analysis of design variables identified a relatively strong positive relation of emissions with the equipment energy consumption (r = 0.98). Monte Carlo simulations evaluate the potential renewable sources (solar, hydro, and wind) in sequence over fossil fuels for the possible transition in the Australian grid by 2030 and 2050, as per the Paris Agreement. 71.91% reduction of CO2 emissions is achievable by recycling vehicle tyres into crumb rubber compared to landfill and incineration. Recycling by-products of CR production, such as steel and textile, significantly mitigates negative impacts. A decrease of 2.23% emissions was associated to the use of crumb rubber as a binder modifier in the asphalt mixture via the midpoint assessment. In endpoint LCA, a higher association of resource (US$) saving costs was observed than for other protective zones, i.e., human health and ecosystem damage. Recycling 466,000 tonnes of disposable waste tyres contributes to 16.1 million US$ worth of resource savings. An equitable industry-based LCA and uncertainty analysis of design parameters can assist in prioritizing suitable options to improve efficiency and future emission strategies on a global scale.

RESCUE: A geomorphology‐based, hydrologic‐hydraulic model for large‐scale inundation mapping

AbstractFlood mapping is a key stage for detecting vulnerable areas, assessing floods impacts, identifying damages, and mitigation plans. Despite of the available guidelines for flood mapping following EU Floods Directive, coherent and comprehensive official flood hazard maps at the national scale are still missing, at least in the EU. In fact, flood mapping generally requires high economical and computational costs for setup/calibration of hydrological/hydraulic models, leading to partial and fragmented vulnerable areas. Here, we propose RESCUE, a laRgE SCale inUndation modEl; the scope is providing a simple and inexpensive tool for flood mapping at the regional scale, overcoming the current fragmentation. RESCUE takes advantage from coupling geomorphological analysis and simplified hydrologic‐hydraulic modeling, providing simple and reliable large scales inundation estimates. It is as parsimonious as purely DTM‐based methods and physically based as any hydrodynamic model, yet computationally more efficient. The model allows for parametric uncertainty analysis for a probabilistic assessment of inundation areas. We demonstrate RESCUE potentialities and limitations using two illustrative examples: Paglia‐Chiani River system, and Central Apennines District (Central Italy). Notwithstanding the inaccuracies of large scale DTMs and the missed representation of flood‐structure interactions, results are promising, suggesting that RESCUE might be an effective tool for preliminary hazard assessment.

Cost-effectiveness of media reporting guidelines for the prevention of suicide.

Media guidelines for the responsible reporting of suicide are a recognized universal suicide prevention intervention. While implemented in numerous countries, including Australia, little is known about whether they are cost-effective. We aimed to determine the cost-effectiveness of Mindframe, the national initiative implementing media guidelines in Australia. We conducted a modelled economic evaluation (5-year time-horizon) incorporating two types of economic analysis: (i) return-on-investment (ROI) comparing estimated cost savings from the intervention to the total intervention cost, and (ii) cost-effectiveness analysis comparing the net intervention costs to health outcomes: suicide deaths prevented and quality-adjusted life-years (QALYs). We also included uncertainty analyses to propagate parameter uncertainty and sensitivity analyses to test the robustness of the model outputs to changes in input parameters and assumptions. The estimated ROI ratio for the main analysis was 94:1 (95% uncertainty interval [UI]: 37 to 170). The intervention was associated with cost savings of A$596M (95% UI: A$228M to A$1,081M), 139 (95% UI: 55 to 252) suicides prevented and 107 (95% UI: 42 to 192) QALYs gained. The intervention was dominant, or cost-saving, compared with no intervention with results being robust to sensitivity analysis but varying based on the conservativeness of the parameters entered. Mindframe was found to be cost-saving, and therefore, worthy of investment and inclusion as part of national suicide prevention strategies.

Uncertainty analysis of acoustic and non-acoustic parameters derived from four-microphone impedance tube measurements

The uncertainty related to measurements is fundamental for qualifying the method applied to the estimation of the quantities derived from an experimental test. In this paper an uncertainty analysis is given for the estimation of the intrinsic properties of fibrous and porous materials, i.e. the characteristic impedance and the characteristic wavenumber, from measurement data obtained by a four-microphone impedance tube. In addition, an uncertainty analysis is given for some other derived material properties, amongst which the flow resistivity. The uncertainty analysis is based on the principles contained in the Guide to the Expression of Uncertainty in Measurements (GUM) and on its supplement 1, concerning the propagation of distributions using a Monte Carlo (MC) method. The GUM and the MC frameworks are compared with each other in terms of accuracy and in terms of flexibility of use. The uncertainty arising from the process of sampling is also evaluated by testing several samples cut from the same material batch.

An interactive graphical interface tool for parameter calibration, sensitivity analysis, uncertainty analysis, and visualization for the Soil and Water Assessment Tool

The Soil and Water Assessment Tool (SWAT) is one of the most widely used and well-tested eco-hydrological models. However, parameter calibration, sensitivity analysis and uncertainty analysis remain among the most challenging tasks. Existing SWAT parameter calibration, sensitivity analysis, and uncertainty analysis tools are either commercial products or free tools with limited options. This study demonstrates an interactive graphical user interface tool in the R environment for SWAT parameter calibration, sensitivity and uncertainty analyses, and visualization, called R-SWAT. Different R functions/packages for parameter calibration, sensitivity analysis, and uncertainty analysis have been incorporated into R-SWAT. Third-party packages can be integrated into R-SWAT with minimum effort. The application of R-SWAT for a test case study demonstrates its functionalities. In general, R-SWAT (1) is a potential platform for developing and testing new sensitivity or optimization packages, and (2) promotes the understanding of hydrological processes with open-source SWAT and R.

Estimation and uncertainty analysis of groundwater quality parameters in a coastal aquifer under seawater intrusion: a comparative study of deep learning and classic machine learning methods.

Excessive withdrawal of groundwater for agricultural irrigation can cause seawater intrusion into coastal aquifers. Such a case will in turn results in deterioration of irrigation water quality. Determination of irrigation water quality with traditional methods is a time-consuming and costly process. However, machine learning algorithms can be useful tools for modeling and estimating groundwater quality used for irrigation water purposes. In this study, TDS, PS, SAR, and Cl parameters of groundwater were estimated with models based on EC and pH variables. For this purpose, prediction performances of two different deep learning methods (convolutional neural network (CNN) and deep neural network (DNN)) and two different classical machine learning (Random Forest (RF) and extreme gradient boosting (XGBoost)) methods were compared. In addition, predictive uncertainty of the models was determined by quantile regression (QR) analysis. Performance criteria and results of uncertainty analysis revealed that CNN (in testing phase, NSE = 0.95 for TDS, NSE = 0.96 for PS, NSE = 0.67 for SAR and NSE = 0.93 for CI) and DNN (in testing phase, NSE = 0.91 for TDS, NSE = 0.91 for PS, NSE = 0.57 for SAR and NSE = 0.94 for Cl) models had quite a close performance in estimation of TDS, PS, SAR, and Cl parameters and higher than the other two classical machine learning methods. As a result, the CNN model can be considered the best performing model in estimating all quality parameters due to the highest NSE and lowest RMSE values. In addition, the Taylor diagram showed that the values estimated using the CNN model had the highest correlation with the measured data. It was determined that the model with the lowest uncertainty based on the PICP statistics was DNN, followed by the CNN model. However, the CNN model has predicted outliers more accurately. Present findings proved that deep learning models could offer efficient tools for predicting irrigation water quality parameters.

Assessment of multiple precipitation interpolation methods and uncertainty analysis of hydrological models in Chaohe River basin, China

Precipitation interpolation is widely used to generate continuous rainfall surfaces for hydrological simulations. However, increasing the precision of values at the unknown points generated by different spatial interpolation methods is challenging. This study used the Chaohe River Basin, which is an important source of Beijing’s drinking water, as a research area to comprehensively evaluate several precipitation interpolation methods (Thiessen polygon, inverse distance weighting, ordinary kriging and ANUSPLIN) for inputs in hydrological simulations. This research showed that the precipitation time-series surface generated using the ANUSPLIN interpolation method had higher accuracy and reliability. Using this precipitation input to drive the hydrological models, we explored the parameter uncertainties of four typical hydrological models (GR4J, IHACRES, Sacramento and MIKE SHE) based on the multi-objective generalized likelihood uncertainty estimation (GLUE) method. The GLUE method was used to study the parameter sensitivity and uncertainty of the model. Results showed that the ANUSPLIN precipitation interpolation surface combined with the Sacramento model performed best. The multi-objective GLUE method had obvious advantages in parameter uncertainty analysis in hydrological models. Simultaneously exploring the convex line and point density distributions of the behavioural parameters with multi-objective functions determined their distribution and sensitivity more effectively.