HS2.2.2

Hydrological modelling of ungauged catchments, which lack observed streamflow data, is an important practical goal in hydrological sciences. A major challenge is to identify a model structure that reflects the hydrological processes relevant to the catchment of interest.

This study contributes a Bayesian framework for identifying individual model mechanisms (process representations) from flow indices regionalized to the catchment of interest. We extend a method previously introduced for mechanism identification in gauged basins, by formulating the inference equations in the space of (regionalized) flow indices and by accounting for posterior parameter uncertainty. A flexible hydrological model is used to generate candidate mechanisms and model structures, followed by statistical hypothesis testing to identify "dominant" (more a posterior probable) model mechanisms.

The proposed method is illustrated using real data and synthetic experiments based on 92 catchments from northern Spain, from which 16 catchments
are treated as ungauged. 624 hydrological model structures from the flexible framework FUSE are employed.

In real data experiments, the method identifies a dominant mechanism in 27% of 112 trials (processes and catchments). The most identifiable process is routing, whereas the least identifiable processes are percolation and unsaturated zone processes. In synthetic experiments, where "true" mechanisms are known, the reliability of method varies from 60% to 95% depending on the combined regionalization and hydrological error; the probability of making an identification remains stable at around 25%. More broadly, the study contributes perspectives on hydrological mechanism identification under data-scarce conditions; limitations and opportunities for improvement are outlined.

How to cite: Prieto, C., Le Vine, N., Kavetski, D., Fenicia, F., Scheidegger, A., and Vitolo, C.: An exploration of Bayesian identification of dominant hydrological mechanisms in ungauged catchments, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-6212, https://doi.org/10.5194/egusphere-egu22-6212, 2022.

In recent years, the risk of sediment-related disasters has increased due to the increase in heavy rain disasters. Therefore, a technique for more easily diagnosing the internal structure of dikes such as fill dams is desired. We applied machine learning to this task.
In this study, we estimated the correspondence between the hydraulic conductivity distribution and the pressure head distribution of the zoned dike using machine learning. The machine learning method is pix2pix which is derived from generative adversarial networks (GAN). Pix2pix learns the relationship between input and output image. Training datasets were generated by using HYDRUS-2D In the HYDRUS-2D simulation, the dike was divided into three zones, and the seepage analysis was performed by changing the hydraulic conductivity of each zone to various values, and the pressure head distribution in the steady state was obtained.
In the forward problem, most of the results could be estimated accurately. On the hand, it was difficult to estimate the inverse problem because of the ill-posed problem. In the inverse problem, we were able to improve the results by giving the training data a priori information about the hydraulic conductivity.
This approach can be used as a surrogate model for the forward problem and the inverse problem in seepage analysis.

How to cite: Homma, Y., Kuroda, S., and Makino, N.: Surrogate model for seepage analysis of a dike using generative adversarial networks, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-9129, https://doi.org/10.5194/egusphere-egu22-9129, 2022.

The heterogeneity in the soil medium makes it economically impossible to measure the soil characteristics across the whole soil profile. Inverse modelling configures a computer model to simulate the groundwater system and calibrates it to identify the soil characteristics in all grid-cells defined in the model. In classic inverse modelling approaches, residuals between simulated and measured groundwater head time series are converted into a single likelihood function to be maximized; therefore, the information content in the measured data is not fully exploited in those approaches. Moreover, the large number of grid-cells makes inverse modelling an ill-posed optimization problem with more unknown parameters than known measured values, leading to different parameter sets having a similar model performance. Despite recent advances in groundwater calibration practices, such as the regularization approach, there is a considerable room to improve groundwater model calibration using information content in the measured data. Regularization aims to add various sets of information to the calibration to tackle the non-uniqueness problem. However, this approach could associate with a considerable degree of subjectivity and uncertainty. This study seeks a novel approach to extract information content from measured data in the form of physically-based metrics that are often called signatures for improving the identifiability in the groundwater model calibration. Moreover, this study proposes a novel automated powerful inverse modelling strategy using a multi-objective approach to incorporate most of the information content through physically meaningful metrics to obtain more consistent models. The benchmark Freyberg 1988 synthetic case study, in which the true model parameter values are known, is used to demonstrate the applicability and potentials of the proposed framework. The reconstructed Freyberg case study in MODFLOW 2005 showed the framework has the ability to find consistent estimates of the groundwater model parameters.

How to cite: Farokhzad, P., Asadzadeh, M., and Holländer, H.: Multi-objective Inverse Modelling Using Physically based Metrics, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-2990, https://doi.org/10.5194/egusphere-egu22-2990, 2022.

The Lower Mississippi River Alluvial Plain, referred to as the Delta, is an important agricultural region in the southeastern United States. Recent trends in crop type conversion and higher crop yields resulted in increased irrigation demand for surface and groundwater, which can lead to aquifer levels dropping. Estimates of continued increased irrigation adoption are compounded by future climatic estimates suggesting hotter summers with higher unpredictability in precipitation amounts. In these conditions, the long-term sustainability of this system depends on understanding complex surface-groundwater flow interactions at different temporal and spatial scales, and the impacts of agricultural conservation practices on water use. In this study, a description of the development of the integrated AnnAGNPS-MODFLOW technology is provided. The proposed system was evaluated in the Upper Sunflower River watershed, located in the Delta region of Mississippi, to characterize existing conditions through comparison with observed streamflow and well water levels. Additionally, the system was used to evaluate the impact of alternative irrigation and management strategies on water levels in the aquifer at field and watershed scales. The proposed technology provides a management tool critical to understanding and evaluating the impact of agricultural practices, irrigation, and aquifer recharge strategies that are important to sustaining Delta water resources in a changing climate.

How to cite: Momm, H., Bingner, R., Moore, K., and Herring, G.: AnnAGNPS-MODFLOW integration for evaluation of agricultural practice impacts on surface and groundwater resources, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-10752, https://doi.org/10.5194/egusphere-egu22-10752, 2022.

Reliability of hydrological model simulation plays an important role in better understanding of hydrological processes, eco-hydrologic and eco-hydraulic condition of a data scarce catchment with limited baseline data. Due to the lack of baseline hydrological dataset, precise simulation of hydrological output from a hydrological model becomes essential. There has been several studies carried out to calibrate a data scarce catchment using river discharge solely on remote sensing data but they have been limited to the rivers with large width. The current study was carried out in Manipur River basin where the widths of the streams are narrow due to which direct application of remotely sensed data possesses serious challenge. This study attempts to calibrate and validate a comprehensive physically semi-distributed Soil and Water Assessment Tool (SWAT) model in Manipur River basin. The SWAT hydrological model was set-up using elevation, soil, weather, and land use land cover (LULC) dataset. For the calibration and validation of model two different approaches has been applied. In the first approach, river streamflow was generated by using stage data based on stage-discharge curve through the technique of spatial proximity, whereas in the second approach, Moderate Resolution Imaging Spectroradiometer (MODIS) evapotranspiration dataset was used at sub-basin scale to calibrate and validate the SWAT model. In the calibration period, the model returned R² and Kling-Gupta Efficiency (KGE) of 0.78 and 0.73, whereas in the validation period R², KGE was found to be 0.75, 0.71 while using the stage-discharge curve approach. The model performance of R², KGE equals 0.67, 0.41, respectively during calibration and R², KGE equals 0.79, 0.53 respectively was obtained during validation when MODIS evapotranspiration dataset was used. From the modelling result it was observed that the model performance was found to be better when streamflow dataset derived from stage-discharge curve used as compared to the MODIS evapotranspiration dataset. The main reason behind the under performance of the model while using MODIS evapotranspiration dataset was due to the underestimation of evapotranspiration by the SWAT model in the cold-dry season from December to February.

How to cite: Anand, V., Oinam, B., and Wieprecht, S.: Multiple techniques for calibration and validation of SWAT model in an ungauged catchment in Inner Himalayan Ranges., EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-11070, https://doi.org/10.5194/egusphere-egu22-11070, 2022.

Isotopic landscapes or “Isoscapes” are a valuable tool for studying hydro-climatic processes and their impact on water supplies at various spatial scales. These isoscapes are extremely useful because they enable the documentation and visualization of large-scale hydrological processes occurring on a regional, continental, or global scale. This study focuses on surface water isotope data (present study and published data) to interpolate, develop isoscapes of Himalayan basins (Indus, Ganga, and Brahmaputra), and analyze spatial variability from regional to local scale. We use physically based information of three basins from hydro-climatic variables such as actual evapotranspiration (AET), mean annual precipitation (MAP), mean annual runoff (MAR), and runoff coefficient, as well as basin variables such as elevation, slope, aspect, size, and land-use/land-cover (LULC), to develop geographically weighted regression (GWR) models. We identified a systematic spatial pattern in the stable isotopes (δ¹⁸O, δ²H, d-excess) of surface water that can be predicted using a GWR model. In the absence of long-term precipitation isotope records, an increased spatial and temporal sampling of surface water for isotopic isoscapes would significantly aid our understanding of hydrological processes, providing that catchment characteristics are taken into consideration. The GWR models used in this study demonstrated the ability to predict isotopic changes in the context of future climate and land-use change in these three major basins.

Keywords: Stable isotopes, Geographically Weighted Regression (GWR), Isoscapes, Himalayan basins.

How to cite: Dar, T., Jahan, A., Rai, N., Bhat, M. A., and Kumar, S.: Spatial analysis of hydrogen and oxygen stable isotopes (“isoscapes”) in Himalayan basins: improved prediction using Geographically Weighted Regression (GWR) models, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-290, https://doi.org/10.5194/egusphere-egu22-290, 2022.

Improved quantification of water partitioning is needed to inform sustainable, integrated land and water management. Spatially-distributed, process-based ecohydrological models are promising tools for achieving this; however, such models typically have many parameters that require estimation from data. Utilising an extremely rich plot-scale dataset incorporating energy (net radiation, temperature, and latent and sensible heat), hydrological (soil moisture, sap flow and actual evapotranspiration) and vegetation (net primary productivity and structure) components, we investigated which types of observation best constrain the parameters of a complex ecohydrological model (EcH₂O-iso) for simulation of water partitioning. Our experimental site was situated within one of the largest coffee agroforestry systems in Costa Rica and experiences high energy inputs and intense rainfall events, thus adding further complexity to the robust simulation of hydrological processes. A series of calibration exercises were undertaken based on combinations of the different observation types. In each case, 100 behavioural parameter sets were chosen following 100,000 Monte Carlo simulations. The “flux mapping” approach was then used to quantify the percentage contribution made by different simulated fluxes to total model outflows (e.g., contributions of transpiration, soil evaporation and interception evaporation to total evapotranspiration), in order to assess how consistently plot hydrology was simulated by the retained parameter sets. Additionally, PCA analysis of performance metrics (including those for observations not used in the given calibration) was undertaken to reveal how contrasting observation types “pull” the model in different directions and, thus, affect its ability to capture the dynamics of each type simultaneously. From this work, we are able to provide guidance on how different ecohydrological datasets may be optimally combined in model calibration. This has implications not only for reducing uncertainty in modelling studies underpinning land and water management, but also for designing future field campaigns such that collection of the most valuable data can be prioritised.

How to cite: Neill, A., Birkel, C., Boll, J., Maneta, M., Roupsard, O., Benegas, L., and Soulsby, C.: Towards improved simulation of water partitioning: which observations have most value in constraining a spatially-distributed ecohydrological model?, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-11754, https://doi.org/10.5194/egusphere-egu22-11754, 2022.

Soil moisture is a vital land surface variable that influences the terrestrial hydrothermal cycles, modulates the land-atmosphere interactions and provides important predictability for weather and climate forecasting. Numerous efforts have been made to investigate the contribution of meteorological forcings, land surface parameters and land models to the uncertainties or precision of soil moisture modeling through both complex statistical approaches and simple comparative land surface modeling experiments. However, previous research mainly focus on one or two factors and the influence of a specific factor is usually quantified by comparing two different datasets. It still unclear that how much added value the current high resolution forcings, surface parameters and land models have to the soil moisture modeling and whether the results depend on the choice of model, datasets and even study regions.

To address the above issue, we first performed a high resolution (6km) soil moisture modeling over China during 2012~2017 by using the newly developed Conjunctive Surface-Subsurface Process version 2 (CSSPv2) land model forced by high-resolution meteorological forcing and high-resolution soil hydraulic property data. The high-resolution simulation has good performance in representing the observed magnitudes and variations of the rootzone (0~1 m) soil moisture based on >1,500 soil moisture stations, and improves the Kling-Gupta efficiency by 33~118% from the current high-resolution global land reanalysis (e.g., ERA5-Land and GLDASv2.1) and remote sensing based products (e.g., ESA CCI and GLEAMv3.1). In order to quantify the contributions from forcings, parameters and CSSPv2 model, we repeated the simulation by using coarse resolution datasets and different models including three meteorological forcing datasets, two soil hydraulic property datasets and three land models. By comparing 48 sets of experiments, the model and soil parameter are found to contribute more than 50% of the improvements at national scale which indicates necessity of developing high resolution land models and model parameters. On the regional scale, however, the meteorological forcing is shown to has the largest added value over the northwestern and southwestern China while land model is most important for the improvement over southern and eastern China. Further works will analyze the specific physical process in CSSPv2 model that improve the soil moisture simulation which will shed light on the future land model development.

How to cite: Ji, P. and Yuan, X.: The added value of forcing, surface parameter and land model to the high resolution soil moisture modeling in China, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-12945, https://doi.org/10.5194/egusphere-egu22-12945, 2022.

Moisture recycling is an important process in the hydrological system, as well as an important ecosystem service being responsible for more than 10% of precipitation in the majority of terrestrial areas. Changes in land use are known to affect this process, however, detailed understanding on how vegetation characteristics, i.e., plant traits, are seldom included in modeling this important process. To overcome this knowledge gap, we conduct a first order examination of the effect of plant traits on recycling, where we examine how variation in plant traits influences moisture recycling properties (average and standard deviation) in the Amazon Basin. More specifically, we used remotely-sensed estimates of trait values for: Specific Leaf Area (SLA), Leaf Dry Matter Content (LDMC), Leaf Phosphorus Content (LPC), Leaf Nitrogen Content (LNC), as well as information on Normalised Difference Vegetation Index (NDVI), and Leaf Area Index (LAI) for 10 years (2001-2010). We link this data on plant traits to six parameters that relevant for moisture recycling, namely Evapotranspiration (ET), Potential Evapotranspiration (PET), Land Surface Temperature (LST) Day and Night, Soil Moisture (SM) and Vapour Pressure Deficit (VPD). We used multivariate regression to analyse how plant traits explain the variance of moisture recycling parameters and find that NDVI (10- 40%), LAI (10-50%) and SLA (5-20%) exert the strongest effects on moisture recycling parameters suggesting that leaf gas exchange traits are most important in comparison to the other traits. We find, however, that the strength and the directionality of the effect while variable, it matches the expectations: NDVI positively correlates with ET, PET and negatively correlates with SM and LST Night; SLA positively correlates with VPD and LST Day. These results suggest that leaf gas exchange properties operate differently during the day and night-time, likely constrained by SM availability, and are linked with VPD and ET exchanges in the direction expected. We then examined whether these patterns were exacerbated or attenuated at the extremes of plant trait values using quantile regression (5th, 50th and 95th), to find that indeed some relationships became stronger (e.g. NDVI and LST Night, LAI and PET, ET and LST Day), while others became more attenuated (e.g. LPC and VPD, NDVI and ET). Finally, we examined whether the effect of traits would be related to the sub-basin processes due to the found control of SM on trait effects and founnd that nutrient and dry matter traits became more important, mostly for the extremes of trait distributions. These results show a promising first approach to include trait distributions in modeling hydrological processes. Indeed, we find some relationships in the direction expected, exacerbated in some cases at the extremes of trait distributions, and at local scale we show that different processes control hydrological parameters in comparison to the whole basin. While promising, more and better estimates of traits through remote sensing or in situ data acquisition are necessary to gain a better understanding of which traits might need to be managed to maintain this important ecosystem service and to understand its links with the overall hydrological cycle.

How to cite: Nguyen, K. and Santos, M. J.: Examining the effect of plant traits on moisture recycling in the Amazon Basin, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-12666, https://doi.org/10.5194/egusphere-egu22-12666, 2022.