The effect of different soil databases on parameter and prediction uncertainty quantification for hydrological modelling

Integrating global land-cover and soil datasets to update saturated hydraulic conductivity parameterization in hydrologic modeling

Model evaluation is a crucial process in model development and quantified model evaluation metrics play a pivotal role in model calibration and validation. However, in current water quality and ecosystem models, conventional model evaluation metrics are derived from hydrological models and often overlook the non-normal distribution of water quality and ecological state variables. In this study, we proposed a series of metrics that consider non-normal distributions by modifying components of the Kling-Gupta efficiency. Subsequently, we tested the existing metrics and newly proposed metrics using four different synthetic datasets and the actual simulation case of total phosphorus, chlorophyll a, and dissolved oxygen concentrations. The results demonstrate that the newly proposed metric, Fu-Zhang efficiency, which replaces the ratio of standard deviations with the ratio of interquartile ranges to measure dispersion similarity, is more suitable for evaluating simulation results with lots of outliers in the measurements. Our study reveals the hidden perils in conventional integrated model evaluation metrics of water quality and ecosystem models and calls for the development of comprehensive and feasible quantitative evaluation frameworks to drive the next paradigm shift.

<p>To achieve and maintain the good ecological status of surface waters, each EU Member State is expected to invest economic resources in a programme of measures. To do this, a preliminary analysis to quantify the anthropogenic pressures and their impacts on surface waters is necessary to prioritise the measures to be undertaken.</p><p>The objectives of this work were to: (i) quantify the nutrient loads from point and diffuse pollution to the Rio Mannu stream identifying the critical time in terms of water quality, and (ii) simulate some mitigation measures for reducing the nutrient loads being delivered to the wetland. Two “measures” were tested to mitigate nutrient pollution in high flow and low flow conditions: (1) the use of treated wastewater from urban wastewater treatment plants (WWTPs) for the irrigation of cultivated olive trees and a (2) reduction in fertiliser usage rates (20%).</p><p>Results showed that under high flow conditions, NO<sub>3</sub>-N and TP loads accounted for 89% and 99% of the total load, respectively. The low flow contribution to the total load was very low, with NO<sub>3</sub>-N and TP accounting for 2.8% and 0.7%, respectively. However, the natural hydrological regime in the study area is intermittent, and low flow represents a critical condition for the water quality due to the high concentrations of TP and NO<sub>3</sub>-N from WWTP discharge. Results indicated that agriculture is the main source of NO<sub>3</sub>-N in the surface waters. The point sources made a minor contribution, in terms of nutrient load to the surface waters, but they constitute a relevant hydrological pressure for the Rio Mannu, producing a shift from an intermittent hydrological regime (natural conditions) towards a perennial regime. The measures are effective to reduce nutrient loads in surface waters at the outlet for all hydrological conditions. Under high flow conditions, the reduction was 9% and 12% for NO<sub>3</sub>-N and TP loads, respectively. The reduction increased under normal and low flow conditions (75% and 83% for NO<sub>3</sub>-N and TP loads, respectively).</p><p>Based on this study, it is evident that a “programme of measures” for improving the current status of surface waters must be oriented towards reducing nutrient loads, both under high and low flow conditions. The results show that a 20% reduction in fertiliser usage on the main crops in this area, and the reuse of wastewater from selected WWTPs, would result in a significant reduction in the nutrient loads delivered to the wetland, although this reduction is not large enough, and supplementary measures would be required.</p>

Estimation of parameter and predictive uncertainty of hydrologic models has traditionally relied on several simplifying assumptions. Residual errors are often assumed to be independent and to be adequately described by a Gaussian probability distribution with a mean of zero and a constant variance. Here we investigate to what extent estimates of parameter and predictive uncertainty are affected when these assumptions are relaxed. A formal generalized likelihood function is presented, which extends the applicability of previously used likelihood functions to situations where residual errors are correlated, heteroscedastic, and non‐Gaussian with varying degrees of kurtosis and skewness. The approach focuses on a correct statistical description of the data and the total model residuals, without separating out various error sources. Application to Bayesian uncertainty analysis of a conceptual rainfall‐runoff model simultaneously identifies the hydrologic model parameters and the appropriate statistical distribution of the residual errors. When applied to daily rainfall‐runoff data from a humid basin we find that (1) residual errors are much better described by a heteroscedastic, first‐order, auto‐correlated error model with a Laplacian distribution function characterized by heavier tails than a Gaussian distribution; and (2) compared to a standard least‐squares approach, proper representation of the statistical distribution of residual errors yields tighter predictive uncertainty bands and different parameter uncertainty estimates that are less sensitive to the particular time period used for inference. Application to daily rainfall‐runoff data from a semiarid basin with more significant residual errors and systematic underprediction of peak flows shows that (1) multiplicative bias factors can be used to compensate for some of the largest errors and (2) a skewed error distribution yields improved estimates of predictive uncertainty in this semiarid basin with near‐zero flows. We conclude that the presented methodology provides improved estimates of parameter and total prediction uncertainty and should be useful for handling complex residual errors in other hydrologic regression models as well.

Seasonal hydrological forecasts are of critical importance for the effective management of water resources, particularly in complex and vulnerable basins such as the Mediterranean area which have both a deficitary water regimen and a high anthropogenic pressure. Nevertheless, inaccuracies in meteorological inputs can propagate through hydrological models, amplifying uncertainties in flow predictions. It is imperative to rectify these forecasts, whether meteorological or hydrological, to enhance prediction reliability and provide robust data for informed decision-making. This study proposes an advanced framework for seasonal hydrological forecasting that integrates raw and corrected weather forecasts with distributed hydrological modeling and sophisticated post-processing techniques to enhance flow prediction accuracy in the Júcar River in Spain as a representative case study of the Mediterranean area.The catchment hydrological model was implemented using the model TETIS v9.1 wich was calibrated and validated using observed records from 1981 to 2019 at seven control points. To do this, a split sample test was conducted using the period 2009–2019 for calibration, and the rest of the time series for validation (1981–2008). The process involved refining parameter maps to ensure good or acceptable performance at all control points. Meteorological data were sourced from the W5E5 dataset, downscaled to a 0.09° resolution using ERA5-Land. This improved the spatial and temporal resolution of the hydrological model. Once we have established an acceptable hydrological model,, the seasonal forecast hindcasts were evaluated using meteorological inputs from global forecasting systems, including ECMWF-SEAS5, CMCC_SPSv35, DWD_GCFS21, and MeteoFrance System8. To address uncertainties in the meteorological forecasts and their propagation to hydrological outputs, two complementary correction strategies were implemented. First, artificial intelligence(fuzzy logic) was applied to correct meteorological inputs before integration into the hydrological model, assuming errors originate solely from meteorological data and treating the hydrological model as a “perfect” simulator. Second, a hydrological error model was developed to identify and adjust discrepancies between simulated and observed flows, addressing systematic biases and errors in the hydrological simulation.The results demonstrated that forecasts based on corrected meteorological inputs exhibited significant accuracy improvements compared to those using unprocessed inputs. The hydrological error model further enhanced prediction reliability by mitigating systematic biases. These findings underscore the effectiveness of combining meteorological forecasts with AI-driven corrections to address uncertainties, thereby improving the robustness of seasonal hydrological predictions. This study highlights the potential for integrating advanced correction techniques into seasonal hydrological forecasting frameworks, offering a replicable methodology for other basins with similar complexities. The proposed framework enhances both the reliability and applicability of forecasts, ensuring their relevance for effective decision-making in complex hydrological systems as the Mediterranean area. The improved accuracy of these forecasts provides a sound scientific support for adaptive water resource management, particularly in the face of increasing climatic variability and environmental changes.Acknowledgments: This study was funded by the Colombian Ministry of Science, Technology, and Innovation (MINCIENCIAS) through the Call for Doctorates Abroad 885-2; by the Valencian Regional Government through the WATER4CAST 2.0 (CIPROM/2023/5) research project; and Spanish Ministry of Science and Innovation through the research project TETISPREDICT (PID2022-141631OB-I00). 

Assessing parameter, precipitation, and predictive uncertainty in a distributed hydrological model using sequential data assimilation with the particle filter

Dividends in flow prediction improvement using high-resolution soil database

We previously found that local O2 extraction efficacy in isolated pump-perfused intestine was enhanced when systemic reflex vasoconstriction was stimulated by hypovolemia (Samsel, R.W., and P.T. Schumacker. 1994. J. Appl. Physiol. 77: 2291-2298). The microvascular mechanism underlying this beneficial effect could involve a redistribution of flow between mucosa and serosa, or an adjustment in the heterogeneity of perfusion within those regions. We measured regional blood flows and distributions of flow and capillary erythrocyte transit times in two segments of small intestine in anesthetized dogs (n = 10). Each vascularly isolated segment of intestine was pump-perfused under high flow (O2 supply-independent VO2) and low flow (O2 supply-dependent) conditions. During the first gut segment, the animal was kept normovolemic using i.v. fluids to minimize reflex vasoconstriction. During the second, the animal was hemorrhaged to augment vasoconstriction (n = 7), or kept normovolemic to control for the effects of time (n = 3). Blood flow distributions were measured using 15 microm radiolabeled microspheres. Tissue blood volume was measured using 99mTc-labeled red blood cells. Capillary volume was determined as the product of tissue blood volume and the histologically derived fraction of vascular volume in the capillaries. Transit times were calculated as the ratio of capillary volume to flow. Each gut segment was fixed and sectioned into 350 approximately 100 mg tissue pieces for analysis. Data revealed significant spatial heterogeneity of blood flow and capillary transit times in both mucosa and muscularis, with relative dispersions (SD/Mean) ranging from 23 to 97%. Hypovolemia caused an increase in flow heterogeneity in muscularis at both high and low flow states, and in mucosa under high flow conditions. However, hypovolemia also elicited changes in capillary volume, such that transit time heterogeneity remained unchanged. Augmentation of vasoconstrictor tone caused a redistribution of flow toward mucosa (P < 0.003) under high and low flow conditions. This redistribution correlated with the improvements in O2 extraction ratio (P = 0.022). Thus, the improvement in gut O2 extraction efficacy seen with increased vasoconstriction may be explained mostly by an intramural redistribution of flow between mucosa and muscularis. Capillary transit time heterogeneity remained unchanged, suggesting that this variable is tightly regulated.

This report presents the results of the design testing of large (36-inch diameter) butterfly valves under high flow conditions. The two butterfly valves were pneumatically operated air-open, air-shut valves (termed valves 1 and 2). These butterfly valves were redesigned to improve their ability to function under high flow conditions. Concern was raised regarding the ability of the butterfly valves to function as required with high flow-induced torque imposed on the valve discs during high steam flow conditions. High flow testing was required to address the flow-induced torque concerns. The valve testing was done using a heavily instrumented piping system. This test program was called the Large Scale Steam Valve Test (LSSVT). The LSSVT program demonstrated that the redesigned valves operated satisfactorily under high flow conditions.

Abstract. This paper addresses the following points: how can whole soil data from normally available soil mapping databases (both conventional and those integrated by digital soil mapping procedures) be usefully employed in hydrology? Answering this question requires a detailed knowledge of the quality and quantity of information embedded in and behind a soil map. To this end a description of the process of drafting soil maps was prepared (which is included in Appendix A of this paper). Then a detailed screening of content and availability of soil maps and database was performed, with the objective of an analytical evaluation of the potential and the limitations of soil data obtained through soil surveys and soil mapping. Then we reclassified the soil features according to their direct, indirect or low hydrologic relevance. During this phase, we also included information regarding whether this data was obtained by qualitative, semi-quantitative or quantitative methods. The analysis was performed according to two main points of concern: (i) the hydrological interpretation of the soil data and (ii) the quality of the estimate or measurement of the soil feature. The interaction between pedology and hydrology processes representation was developed through the following Italian case studies with different hydropedological inputs: (i) comparative land evaluation models, by means of an exhaustive itinerary from simple to complex modelling applications depending on soil data availability, (ii) mapping of soil hydrological behaviour for irrigation management at the district scale, where the main hydropedological input was the application of calibrated pedo-transfer functions and the Hydrological Function Unit concept, and (iii) flood event simulation in an ungauged basin, with the functional aggregation of different soil units for a simplified soil pattern. In conclusion, we show that special care is required in handling data from soil databases if full potential is to be achieved. Further, all the case studies agree on the appropriate degree of complexity of the soil hydrological model to be applied. We also emphasise that effective interaction between pedology and hydrology to address landscape hydrology requires (i) greater awareness of the hydrological community about the type of soil information behind a soil map or a soil database, (ii) the development of a better quantitative framework by the pedological community for evaluating hydrological features, and (iii) quantitative information on soil spatial variability.

Continuous hydrological records at the desired site with spatial and temporal coverage are essential to water resources management and flood prevention. Hydrological gauging stations and observations are limited at transboundary Himalayan River basins in developing countries. This study is carried out to estimate discharge at ungauged sites from donor catchments based on calibrated discharge. In the Tamakoshi River basin, Spatial Process in HYdrology (SPHY), Hydrologiska Byråns Vattenbalansavdelning (HBV)-light, and Hydrologic Engineering Centre Hydrologic Modelling System (HEC HMS) models were used for hydrological simulation. The ungauged discharge estimated at Benighat is based on daily, monthly, and annual bases from upstream gauged stations. Statistical indices were obtained as NSE &amp;gt; 0.62, R2 &amp;gt; 0.77, and PBias &amp;lt;26% in the simulation of these models. The low flow prediction is more reliable than the high flow prediction. The deviation in predicted flow mostly appeared in high flow periods. All the hydrological models simulated ungauged streamflow were similar in the flow pattern, and the estimated discharge at the ungauged receiver site was greater than at the donor site. Comparative simulations are better alternatives to estimate ungauged discharge than individual simulations. This research will support future reference to generate streamflow data at ungauged sites and can help fill the data recording gap at similar mountain river basins.

Green roofs can be valuable components in sustainable urban drainage systems, and hydrological models may provide useful information about the runoff from green roofs for planning purposes. Various models have been proposed in the literature, but so far no papers have compared the performance of multiple models across multiple full-size green roofs. This paper compared 4 models: the conceptual models Urbis and SWMM and the physically-based models Hydrus-1D and Mike SHE, across two field sites (Lyon, France and Umeå, Sweden) and two calibration periods for each site. The uncertainty and accuracy of model predictions were dependent on the selected calibration site and period. Overall model predictions from the simple conceptual model Urbis were least accurate and most uncertain; predictions from SWMM and Mike SHE were jointly the best in terms of raw percentage observations covered by their flow prediction intervals, but the uncertainty in the predictions in SWMM was smaller. However, predictions from Hydrus were more accurate in terms of how well the observations conformed to probabilistic flow predictions. Mike SHE performed best in terms of total runoff volume. In Urbis, SWMM and Hydrus uncertainty in model predictions was almost completely driven by random uncertainty, while parametric uncertainty played a significant role in Mike SHE. Parameter identifiability and most likely parameter values determined with the DREAM Bayesian algorithm were found to be inconsistent across calibration periods in all models, raising questions about the generalizability of model applications. Calibration periods where rainfall retention was highly variable between events were more informative for parameter values in all models.

Land surface models (LSMs) and hydrologic models are parameterized models. The number of involved parameters is often large. Sensitivity analysis (SA) is a key step to understand the complex relationships between parameters and between state variables and parameters. SA is also critical to understand system dynamics and to examine the parameter identifiability. In this paper, parameter SA for a fully coupled, physically based, distributed land surface hydrologic model, namely, the Flux–Penn State Integrated Hydrologic Model (Flux–PIHM), is performed. Multiparameter and single-parameter tests are performed to examine the three dimensions of identifiability: distinguishability, observability, and simplicity. Results show that Flux–PIHM model predictions of discharge, water table depth, soil moisture, land surface temperature, and surface heat fluxes are very sensitive to the selection of parameter values. Parameter uncertainties produce large uncertainties in hydrologic and land surface variable predictions. The van Genuchten parameters α and β and the Zilitinkevich parameter Czil are the most identifiable among the 20 tested parameters. Results indicate that the land surface and the subsurface are closely coupled. Hydrologic parameters have significant influence on land surface simulations. At the same time, land surface parameters have considerable impacts on hydrologic simulations; the evapotranspiration prediction prior to a strong precipitation event is critical for initializing accurate prediction of discharge peaks. Results also show that parameter identifiability depends on seasons and canopy wetness. Parameter identifiability at high and low flow conditions can be extremely different. Complex system dynamics have been revealed during the SA.

Sensitivity of flood dynamics to different soil information sources in urbanized areas

Rising sea levels and anthropogenic activities are intensifying pressure on coastal zones. Process-based coastal morphodynamic models are increasingly used to forecast natural and anthropogenic beach morphology changes at various spatio-temporal scales. Such predictions are crucial for the sustainable management of coasts. However, process-based morphodynamic models contain numerous free model parameters, introducing uncertainty in predictions. Systematically exploring the parameter space has remained a challenge due to the high computational demands of these morphodynamic models. Here, for the first time we quantify parameter uncertainty of a state-of-the-art morphodynamic (2DH) coastal area model (Delft3D) by systematically varying key model parameters, utilizing the Dutch national supercomputer: SurfSara. We simulate the initial (14-month) response of the Sand Engine, an innovative mega-nourishment placed along the Holland coast with 1024 strategically chosen parameter sets. The resulting simulations are analysed using Generalised Likelihood Uncertainty Estimation (GLUE) to attain probability distributions of morphological evolution and its sensitivity to parameter settings. The model simulations all show an alongshore redistribution of sediment resembling what is observed. However, even simulations with similar skill reveal substantial differences in predicted morphologies (same order of magnitude as the predictions’ 90% confidence interval). Our findings suggest that identifying a single optimal parameter set for coastal numerical models might be unrealistic, even for well-defined cases like large-scale coastal interventions, and that an ensemble modeling approach that quantifies parameter uncertainty is likely better suited for studies relying on morphodynamic predictions. Furthermore, we find that the magnitude of the uncertainty induced by the free model parameters is comparable to that resulting from year-to-year variations in wave climate, underscoring the importance of including both sources in uncertainty assessments.