Multi-objective Calibration Research Articles

The effects of satellite soil moisture data on the parametrization of topsoil and root zone soil moisture in a conceptual hydrological model

Abstract In a previous study, the topsoil and root zone ASCAT satellite soil moisture data were implemented into three multi-objective calibration approaches of the TUW hydrological model in 209 Austrian catchments. This paper examines the model parametrization in those catchments, which in the validation of the dual-layer conceptual semi-distributed model showed improvement in the runoff simulation efficiency compared to the single objective runoff calibration. The runoff simulation efficiency of the three multi-objective approaches was separately considered. Inferences about the specific location and the physiographic properties of the catchments where the inclusion of ASCAT data proved beneficial were made. Improvements were primarily observed in the watersheds with lower slopes (median of the catchment slope less than 15 per cent) and a higher proportion of farming land use (median of the proportion of agricultural land above 20 per cent), as well as in catchments where the runoff is not significantly influenced by snowmelt and glacier runoff. Changes in the mean and variability of the field capacity parameter FC of the soil moisture regime were analysed. The values of FC decreased by 20 per cent on average. Consequently, the catchments’ water balance closure generally improved by the increase in catchment evapotranspiration during the validation period. Improvements in model efficiency could be attributed to better runoff simulation in the spring and autumn month. The findings refine recommendations regarding when hydrological modelling could consider satellite soil moisture data added to runoff signatures in calibration useful.

Precipitation fate and transport in a Mediterranean catchment through models calibrated on plant and stream water isotope data

Abstract. To predict hydrologic responses to inputs and perturbations, it is important to understand how precipitation is stored in catchments, released back to the atmosphere via evapotranspiration (ET), or transported to aquifers and streams. We investigated this partitioning of precipitation using stable isotopes of water (18O) at the Can Vila catchment in the Spanish Pyrenees mountains. The isotope data covered four years of measurements, comprising >550 rainfall and >980 stream water samples, capturing intra-event variations. They were complemented by fortnightly plant water isotope data sampled over eight months. The isotope data were used to quantify how long it takes for water to become evapotranspiration or to be discharged as streamflow using StorAge Selection (SAS) functions. We calibrated the SAS functions using a conventional approach fitting the model solely to stream water isotope data and a multi-objective calibration approach in which the model was simultaneously fitted to tree xylem water isotope data. Our results showed that the conventional model-fitting approach was not able to constrain the model parameters that represented the age of water supplying ET. Consequently, the ET isotope ratios simulated by the conventionally calibrated model failed to adequately simulate the observed xylem isotope ratios. However, the SAS model was capable of adequately simulating both observed stream water and xylem water isotope ratios, if those xylem water isotope observations were used in calibration (i.e., the multi-objective approach). The multi-objective calibration approach led to a more constrained parameter space facilitating parameter value identification. The model was tested on a segment of data reserved for validation showing a Kling–Gupta Efficiency of 0.72 compared to the 0.83 observed during in the calibration period. The water-age dynamics inferred from the model calibrated using the conventional approach differed substantially from those inferred from the multi-objective calibration model. The latter suggested that the median ages of water supplying evapotranspiration is much older (150–300 d) than what was suggested by the former (50–200 d). Regardless, the modeling results support recent findings in ecohydrological field studies that highlighted both subsurface heterogeneity of water storage and fluxes and the use of relatively old water by trees. We contextualized the SAS-derived water ages by also using young-water-fraction and endmember-splitting approaches, which respectively also showed the contribution of young water to streamflow was variable but sensitive to runoff rates and that ET was largely sourced by winter precipitation that must have resided in the subsurface across seasons.

Formulation of Wavelet Based Multi‐Scale Multi‐Objective Performance Evaluation (WMMPE) Metric for Improved Calibration of Hydrological Models

AbstractEstimation of the hydrological model parameters, known as model calibration, is an important step in application of these models for various water resources problems. In the auto‐calibration approach, the match between the simulated and observed variables is assessed using the statistical performance measures, which are used as objective functions. Earlier studies show that the use of traditional single‐objective functions alone cannot capture all the features of the hydrographs, motivating the utilization of multi‐objective calibration. Further, wavelet‐based approaches can capture features that occur at different time scales. In this study, Wavelet Transforms are employed to formulate a Wavelet based Multi‐Scale Multi‐Objective Performance Evaluation (WMMPE) metric that combines multiple standard metrics at multiple time scales. The performance of WMMPE along with other objective functions such as Nash Sutcliffe Efficiency (NSE), Kling Gupta Efficiency (KGE)), Euclidean distance based Multi Objective Calibration Function and wavelet based Multi‐Scale NSE in calibrating two conceptual hydrologic models were tested on 53 watersheds. The performance of the five calibrated models were evaluated for their efficacy in simulating different characteristics of hydrograph such as, low and high flows, water balance components, frequency domain characteristics and average error magnitude. The results show that WMMPE calibrated models are performing well in matching all the flow segments of hydrograph in >22 basins for the two conceptual models. Further, results of the analysis suggested that the presence of outliers have minimum impact on the simulation of WMMPE calibrated models in the validation period.

Enhancing SWAT model predictivity using multi-objective calibration: effects of integrating remotely sensed evapotranspiration and leaf area index

Enhancing SWAT model predictivity using multi-objective calibration: effects of integrating remotely sensed evapotranspiration and leaf area index

A hybrid macro-modelling strategy with multi-objective calibration for accurate simulation of multi-ring masonry arches and bridges

This paper presents an efficient hybrid continuum-discrete macro-modelling strategy with an enhanced multiscale calibration procedure for realistic simulations of brick/block-masonry bridges. The response of these structures is affected by the intrinsic nonlinearity of the masonry material, which in turn depends upon the mechanical properties of units and mortar joints and the bond characteristics. Finite element approaches based upon homogenised representations are widely employed to assess the nonlinear behaviour up to collapse, as they are generally associated with a limited computational demand. However, such models require an accurate calibration of model material parameters to properly allow for masonry bond. According to the proposed approach, the macroscale material parameters are determined by an advanced multi-objective strategy with genetic algorithms from the results of mesoscale “virtual” tests through the minimisation of appropriate functionals of the scale transition error. The developed continuum-discrete finite element macroscale description and the calibration procedure are applied to simulate the nonlinear behaviour up to collapse of multi-ring arch-bridge specimens focusing on the 2D planar response. The results obtained are compared to those achieved using detailed mesoscale models confirming the effectiveness and accuracy of the proposed approach for realistic nonlinear simulations of masonry arch bridges.

Understanding climate-induced changes of snow hydrological processes in the Kaidu River Basin through the CemaNeige-GR6J model

Snow-affected mountainous areas are highly vulnerable to climate changes in arid and semi-arid regions. Previous studies have put more emphasis on runoff, however, research about snow hydrological processes is still insufficient. Based on the optimization of calibration strategies, this study aims to provide a more realistic simulation of snow hydrology by coupling both a lumped rainfall-runoff model (GR6J) and a semi-distributed snow model (CemaNeige). Compared with calibration using only runoff, the multi-objective calibration with both runoff and snow not only significantly improved the KGE (Kling-Gupta efficiency) of snow from 0.34/0.47 to 0.67/0.69 during the calibration/validation period, but also slightly improved the KGE of runoff from 0.84/0.75 to 0.86/0.78, respectively. Additionally, we also found that the calibration using only runoff may underestimate snow cover (41%) and overestimate snowpack (20%), while there is little discrepancy in snowmelt and runoff. Snowmelt runoff is 57.7 mm accounting for 26% of the annual runoff. In the historical period (1980–2011), the runoff varied with a positive trend of 2.15 mm year−1 (p < 0.1), while its tendency became negative in the future. A 15% to 30% decline in runoff was projected for SSP370 (SSP3-RCP7) and SSP585 (SSP5-RCP8.5) in the 2085s (2070–2100) compared with that in the 1985s (1970–2000). Through further study on seasonal variations, we found that the future runoff decline is probably due to mild rise in rain not enough to alleviate the larger AE (actual evaporation), and for the runoff and AE we also found the variations in summer dominated the annual changes. In response to global warming, SCED (snow cover end date) advances while SCOD (snow cover onset date) recedes in all four scenarios as the year increases. Generally, coupling different hydrological models and multi-objective calibration strategies is vital for understanding the hydrological processes in snow-affected catchments.

Towards improving the calibration practice of conceptual hydrological models of extensive green roofs

Conceptual rainfall-runoff models (CRRMs) can be used as a design tool for green roofs due to their simplicity and acceptable accuracy. This study showed how the uncertainty of CRRM parameters could be reduced by changing the calibration practice, which can enhance the interpretation and identifiability of CRRM parameters. A CRRM was developed and tested on a dataset of 14 extensive green roofs located in four Norwegian cities with different climatic conditions. Two calibration schemes were compared: a common scheme using runoff data as a basis for calibration (single-objective), and a scheme combining runoff and soil moisture data for the calibration (multi-objective scheme). The results confirmed the ability of the CRRM to simulate runoff from extensive green roofs across multiple climatic zones and different roof configurations (Kling Gupta Efficiency > 0.75). The multi-objective calibration scheme was found to reduce the uncertainty of the CRRM parameters, especially the storage parameters, enhancing the physical interpretation of parameter values. The study attempted to give guidelines to estimate parameters of the CRRM which can be used by practitioners for new roof configurations under different climatic conditions.

Evaluation of Sentinel-1, SMAP and SMOS surface soil moisture products for distributed eco-hydrological modelling in Mediterranean forest basins

Reliable distributed hydrological modeling, especially in semi-arid areas, must consider the inclusion of surface soil moisture (SSM) spatial information during the calibration process. This variable plays a key role in the evapotranspiration processes that determine the hydrological cycle. The coarse resolution of the SSM estimates by satellite remote sensing has restricted the application of this approach to only large basins, focusing most of the studies in the consideration of simply the temporal dynamics of this variable. The growing efforts in providing higher spatial resolution through disaggregating methodologies or new sensor estimates facilitates the application of this spatial approach to small basins. This paper explores the applicability of the currently available satellite surface soil moisture estimates for distributed eco-hydrological modelling in Mediterranean forest basins. On one hand, this study contributes to fill the existing research gap on the use of remote sensing SSM spatial patterns within the distributed hydrological modelling framework in small basins. On the other hand, it serves as an indirect validation method for the spatial performance of satellite SSM products. To achieve this goal, we implemented the eco-hydrological model TETIS in three case studies named: Hozgarganta (southern Spain), Ceira (western Portugal) and Carraixet (eastern Spain). The SSM estimates selected for comparison were Sentinel-1 SSM provided by the Copernicus Global Land Services (CGLS), SMAP SSM disaggregated using Sentinel-1 (SPL2SMAP_S) provided by the National Aeronautics and Space Administration (NASA), SMOS SSM provided by the Barcelona Expert Center (BEC), and SMOS and SMAP SSM disaggregated using the DISPATCH algorithm provided by Lobelia Earth. The methodology employed involved a multi-objective and multi-variable calibration in terms of remote sensing SSM spatial patterns and in-situ streamflow, using the Spatial Efficiency Metric (SPAEF) and the Nash-Sutcliffe efficiency index (NSE) respectively. Before model calibration a sensitivity analysis of the most influent variables was performed. The temporal and spatial comparison of the reference SSM products revealed inconsistencies amongst products. The disaggregating methodology determined the spatial agreement to a greater degree than the sensor itself (i.e. SMAP, SMOS). In spite of the differences amongst products, the multi-objective calibration approach proposed increased the robustness of the hydrological modelling.

Multi-site multi-objective calibration of SWAT model using a large dataset for improved performance in Ethiopia

Multi-site multi-objective calibration of SWAT model using a large dataset for improved performance in Ethiopia

Climate Normalized Spatial Patterns of Evapotranspiration Enhance the Calibration of a Hydrological Model

Spatial pattern-oriented evaluations of distributed hydrological models have contributed towards an improved realism of hydrological simulations. This advancement has been supported by the broad range of readily available satellite-based datasets of key hydrological variables, such as evapotranspiration (ET). At larger scale, spatial patterns of ET are often driven by underlying climate gradients, and with this study, we argue that gradient dominated patterns may hamper the potential of spatial pattern-oriented evaluation frameworks. We hypothesize that the climate control of spatial patterns of ET overshadows the effect model parameters have on the simulated patterns. To address this, we propose a climate normalization strategy. This is demonstrated for the Senegal River basin as a modeling case study, where the dominant north-south precipitation gradient is the main driver of the observed hydrological variability. We apply the mesoscale Hydrological Model (mHM) to model the hydrological cycle of the Senegal River basin. Two multi-objective calibration experiments investigate the effect of climate normalization. Both calibrations utilize observed discharge (Q) in combination with remote sensing ET data, where one is based on the original ET pattern and the other utilizes the normalized ET pattern. As objective functions we applied the Kling-Gupta-Efficiency (KGE) for Q and the Spatial Efficiency (SPAEF) for ET. We identify parameter sets that balance the tradeoffs between the two independent observations and find that the calibration using the normalized ET pattern does not compromise the spatial pattern performance of the original pattern. However, vice versa, this is not necessarily the case, since the calibration using the original ET pattern showed a poorer performance for the normalized pattern, i.e., a 30% decrease in SPAEF. Both calibrations reached comparable performance of Q, i.e., KGE around 0.7. With this study, we identified a general shortcoming of spatial pattern-oriented model evaluations using ET in basins dominated by a climate gradient, but we argue that this also applies to other variables such as, soil moisture or land surface temperature.

Emulator-based Bayesian optimization for efficient multi-objective calibration of an individual-based model of malaria

Individual-based models have become important tools in the global battle against infectious diseases, yet model complexity can make calibration to biological and epidemiological data challenging. We propose using a Bayesian optimization framework employing Gaussian process or machine learning emulator functions to calibrate a complex malaria transmission simulator. We demonstrate our approach by optimizing over a high-dimensional parameter space with respect to a portfolio of multiple fitting objectives built from datasets capturing the natural history of malaria transmission and disease progression. Our approach quickly outperforms previous calibrations, yielding an improved final goodness of fit. Per-objective parameter importance and sensitivity diagnostics provided by our approach offer epidemiological insights and enhance trust in predictions through greater interpretability.

Propagation of CMORPH rainfall errors to REW streamflow simulation mismatch in the upper Zambezi Basin

Study regionThis research is carried out in the Kabompo Basin, a headwater catchment of the Zambezi Basin which has an area of 67,261 km2. Kabompo River originates in North-Western Province of Zambia between the Zambezi and Congo River Basins. Study focusThis study focuses on error propagation of the National Oceanic and Atmospheric Administration (NOAA) Climate Prediction Center-MORPHing (CMORPH) rainfall product on simulated streamflow. Assessments are based on automated multi-objective calibration of the Representative Elementary Watershed (REW) model (2006–2012). Parameters of the model were optimized using the ɛ-NSGAII algorithm. Assessments on error propagation targeted streamflow modelling for hydrograph shape and volume, specific hydrograph characteristics, and water balance composition. New hydrological insights for the regionBy use of multiple objective functions, this study shows that uncorrected CMORPH results in substantial augmentation of rainfall error to streamflow simulation mismatch whereas bias corrected estimates result in attenuation of error. Analysis for water balance composition has great potential to improve application of satellite precipitation products in water management and decision making in the Zambezi basin. This study advises optimization of model parameters for each respective rainfall input data source so to identify outcomes and effects of respective rainfall data sources on the simulated water balance composition and closure.

Multi-objective calibration and evaluation of SWAT modela case study in Meichuan River basin

Multi-objective calibration and evaluation of SWAT modela case study in Meichuan River basin

REGSim: An open-source framework to estimate recharge and simulate groundwater heads

The computational complexity of distributed groundwater models poses a significant challenge in the conjunctive management of surface and groundwater resources. Here, we propose a multi-model framework to predict groundwater recharge using a simple water balance model. We compare and contrast two lumped parsimonious conceptual models of groundwater recharge, one based on P only and another, additionally including PE as a predictor. Both model variants also include an option to specify annually fixed or seasonally varying recharge factors, as well as a number of pumping scenarios. We incorporate these models within a python based open-source toolbox, REGSim (Recharge Estimation and Groundwater Simulation). The toolbox features a multi-objective evolutionary algorithm (NSGA-II), the Generalized Likelihood Uncertainty Estimation (GLUE) method, and regional sensitivity analysis (RSA). These functionalities allow the users to perform multi-objective calibration of the models, obtain confidence intervals on predicted groundwater heads, and understand the relative importance of different model parameters. REGSim is used to simulate groundwater heads for the urban agglomeration of Hyderabad, India. Using REGSim, we tested alternative conceptualizations of groundwater recharge and pumping processes in the city. We found that the intra-annual dynamics of groundwater levels are better explained by seasonally varying the recharge factors than annually fixed recharge factors. The model achieved an NSE of 0.64 and 0.70 during the validation for the formulations based on P only and using both P and PE, respectively. Furthermore, sensitivity analysis revealed that specific yield is the most influential parameter affecting the simulated groundwater head in the region.

Modeling Groundwater-Fed Irrigation and Its Impact on Streamflow and Groundwater Depth in an Agricultural Area of Huaihe River Basin, China

The amount of water taken from groundwater for agricultural irrigation is often not observed, while hydrological models have been extensively proposed to investigate the irrigation dynamics and impacts in agricultural areas. In this work, we propose an agro-hydrological model that integrates agricultural irrigation with the traditional Xin’anjiang (XAJ) hydrological model. In particular, the proposed model incorporates the FAO guidelines on crop evapotranspiration into hydrological routing of water balance and flow fluxes in unsaturated and saturated zones. The model was used to calibrate the groundwater irrigation amounts in terms of both the observed river discharge and the groundwater depth in the Xuanwu plain area of the Huaihe River Basin in China. The calibration and sensitivity analyses were performed by the shuffled complex evolution (SCE-UA) method. This method can be applied to a single-objective optimization of model parameters, based on either the river discharge or the groundwater depth, or to a multi-objective optimization of model parameters based on both of these objectives. The results show that the multi-objective calibration is more efficient than the single-objective method for capturing dynamics of the river discharge and the groundwater depth. The estimated means of the annual groundwater withdrawal for wheat and maize irrigations were found to be about 140.5 mm and 13.7 mm, respectively. The correlation between the groundwater withdrawal and the change in groundwater depth during crop growing seasons demonstrated that the groundwater withdrawal is the dominant factor for the groundwater depth change in the river basin, particularly in the winter wheat season. Moreover, model simulations show that the combined effects of the reduced precipitation and the increased groundwater withdrawal would lead to a decrease of the average annual runoff and an increase of the average groundwater depth. These estimates can greatly help in understanding the irregular changes in the groundwater withdrawal and offer a quantitative basis for studying future groundwater demands in this area.

Performance Evaluation of Single and Multi-Objective Calibration in Rainfall-Runoff Modelling

"ENGINEER", established in 1973 by The Institution of Engineers, is a Journal for dissemination of Engineering knowledge, published quarterly.Cover DescriptionCover images show a schematic diagram of a Pumped-Storage Hydroelectricity (PSH) facility and one of the promising sites for PSH in Sri Lanka. The latter image was obtained from the publicly available JICA report on “Development Planning on Optimal Power Generation for Peak Demand in Sri Lanka”. The Institution of Engineers, Sri Lanka acknowledges JICA for the image. If you are interested in contributing an Original Technical Paper based on research by the author(s) (intended for Section I) or an Original Article of Professional or Technical interest related to Engineering. (Section II) to this journal please go through the publication checklist which can be downloaded from here.

A new fractal-theory-based criterion for hydrological model calibration

Abstract. Fractality has been found in many areas and has been used to describe the internal features of time series. But is it possible to use fractal theory to improve the performance of hydrological models? This study aims at investigating the potential benefits of applying fractal theory in model calibration. A new criterion named the ratio of fractal dimensions (RD) is defined as the ratio of the fractal dimensions of simulated and observed streamflow series. To combine the advantages of fractal theory with classical criteria based on squared residuals, a multi-objective calibration strategy is designed. The selected classical criterion is the Nash–Sutcliffe efficiency (E). The E–RD strategy is tested in three study cases with different climates and geographies. The results reveal that, in most aspects, introducing RD into model calibration makes the simulation of streamflow components more reasonable. Also, pursuing a better RD during calibration leads to only a small decrease in E. We therefore recommend choosing the parameter set with the best E among the parameter sets with RD values of around 1.

Simulation and validation of indoor temperatures and relative humidity in multi-zone buildings under occupancy conditions using multi-objective calibration

Buildings are expected to be energy efficient but also to provide a comfortable environment to their occupants as well as to be durable. Indoor relative humidity influences the energy consumption, structure and the occupants’ comfort and health. Therefore, it is necessary to control and predict indoor environmental conditions. This work aims to simulate indoor temperatures and relative humidity in a multi-zone building model under realistic conditions, such as occupancy and moisture gains, operation of windows and doors and mechanical ventilation. The combined multi-zone air flow model is developed with TRNSYS and TRNFLOW and it is applied to a phase of the Twin Houses extended experiment of the international energy agency, energy in buildings and communities (IEA EBC) Annex 71. The main novelty of this work is to determine the masses of surface and deep moisture buffer storage by performing a multi-objective calibration of indoor temperatures and relative humidity. Results prove that building energy simulations need to model the moisture buffering of internal materials to accurately predict indoor humidity. The error in simulating relative humidity is reduced by 69% in the house with moisture gains after calibration of the masses of buffering materials. Moreover, the calibrated multi-zone building model shows a great agreement between simulated and measured data with an average root mean square error (RMSE) among thermal zones of 0.51 °C and 0.48 °C in indoor temperatures and 3.58% and 2.21% in relative humidity for the two houses in the validation period.

Multi-objective calibration of the double-ellipsoid heat source model for GMAW process simulation

The scope of application of simulation models in welding is limited by the accuracy of their output results. This paper presents a calibration procedure for a 3-D quasi-stationary model of heat transfer for gas metal arc welding. The double-ellipsoid heat source used in this model has five input parameters whose value cannot be specified accurately. To estimate these values, we employed a multi-objective calibration procedure with two objective functions using the paretosearch optimization algorithm. Objective functions represented the error between simulated and experimentally observed values of penetration depth and weld bead width during gas metal arc welding of P355GH steel plates. All input parameters were assumed to be a power function of line energy. To reduce computational time, we replaced the numerical model with a response surface methodology metamodel based on an optimal set of simulation results from the numerical model. The results of the simulations based on calculated values of input parameters for the heat source model showed excellent matching with the experimental results.

Efficient multi-objective calibration and uncertainty analysis of distributed snow simulations in rugged alpine terrain

In mountainous terrain, reliable snow simulations are crucial for many applications. However, except in highly instrumented research catchments, meteorological data are usually limited, and so the interpolated spatial fields used to force snow models are uncertain. Moreover, certain potentially important processes cannot presently be simulated at catchment scales using entirely physical algorithms. It is therefore often appropriate to introduce empirical parameters into otherwise physically-based snow models. Many opportunities to incorporate snow observations into the parameter estimation process now exist, but they remain to be fully exploited. In this context, a novel approach to the calibration of an energy balance-based snow model that additionally accounts for gravitational redistribution is presented. Several important parameters were estimated using an efficient, gradient-based method with respect to two complementary observation types – Landsat 8-derived snow extent maps, and reconstructed snow water equivalent (SWE) time-series. When assessed on a per-pixel basis, observed patterns were ultimately reproduced with a mean accuracy of 85%. Spatial performance metrics compared favourably with those previously reported, whilst the temporal evolution of SWE at the stations was also satisfactorily captured. Subsequent uncertainty and data worth analyses revealed that: i) the propensity for model predictions to be erroneous was substantially reduced by calibration, ii) pre-calibration uncertainty was largely associated with two parameters which modify the longwave component of the energy balance, but this uncertainty was greatly diminished by calibration, and iii) a lower elevation SWE series was particularly valuable, despite containing comparatively few observations. Overall, our work demonstrates that contemporary snow models, observation technologies, and inverse approaches can be combined to both constrain and quantify the uncertainty associated with simulations of alpine snow dynamics.