Abstract. Model intercomparison studies are carried out to test and compare the simulated outputs of various model setups over the same study domain. The Great Lakes region is such a domain of high public interest as it not only resembles a challenging region to model with its transboundary location, strong lake effects, and regions of strong human impact but is also one of the most densely populated areas in the USA and Canada. This study brought together a wide range of researchers setting up their models of choice in a highly standardized experimental setup using the same geophysical datasets, forcings, common routing product, and locations of performance evaluation across the 1×106 km2 study domain. The study comprises 13 models covering a wide range of model types from machine-learning-based, basin-wise, subbasin-based, and gridded models that are either locally or globally calibrated or calibrated for one of each of the six predefined regions of the watershed. Unlike most hydrologically focused model intercomparisons, this study not only compares models regarding their capability to simulate streamflow (Q) but also evaluates the quality of simulated actual evapotranspiration (AET), surface soil moisture (SSM), and snow water equivalent (SWE). The latter three outputs are compared against gridded reference datasets. The comparisons are performed in two ways – either by aggregating model outputs and the reference to basin level or by regridding all model outputs to the reference grid and comparing the model simulations at each grid-cell. The main results of this study are as follows:
 The comparison of models regarding streamflow reveals the superior quality of the machine-learning-based model in the performance of all experiments; even for the most challenging spatiotemporal validation, the machine learning (ML) model outperforms any other physically based model. While the locally calibrated models lead to good performance in calibration and temporal validation (even outperforming several regionally calibrated models), they lose performance when they are transferred to locations that the model has not been calibrated on. This is likely to be improved with more advanced strategies to transfer these models in space. The regionally calibrated models – while losing less performance in spatial and spatiotemporal validation than locally calibrated models – exhibit low performances in highly regulated and urban areas and agricultural regions in the USA. Comparisons of additional model outputs (AET, SSM, and SWE) against gridded reference datasets show that aggregating model outputs and the reference dataset to the basin scale can lead to different conclusions than a comparison at the native grid scale. The latter is deemed preferable, especially for variables with large spatial variability such as SWE. A multi-objective-based analysis of the model performances across all variables (Q, AET, SSM, and SWE) reveals overall well-performing locally calibrated models (i.e., HYMOD2-lumped) and regionally calibrated models (i.e., MESH-SVS-Raven and GEM-Hydro-Watroute) due to varying reasons. The machine-learning-based model was not included here as it is not set up to simulate AET, SSM, and SWE. All basin-aggregated model outputs and observations for the model variables evaluated in this study are available on an interactive website that enables users to visualize results and download the data and model outputs.


/pdf/the-great-lakes-runoff-intercomparison-project-phase-4-the-23fb1xsz.pdf

The Great Lakes Runoff Intercomparison Project Phase 4: the Great Lakes (GRIP-GL)

Abstract. Deep learning methods have frequently outperformed conceptual hydrologic models in rainfall-runoff modelling. Attempts of investigating such deep learning models internally are being made, but the traceability of model states and processes and their interrelations to model input and output is not yet fully understood. Direct interpretability of mechanistic processes has always been considered an asset of conceptual models that helps to gain system understanding aside of predictability. We introduce hydrologic neural ordinary differential equation (ODE) models that perform as well as state-of-the-art deep learning methods in stream flow prediction while maintaining the ease of interpretability of conceptual hydrologic models. In neural ODEs, internal processes that are represented in differential equations, are substituted by neural networks. Therefore, neural ODE models enable the fusion of deep learning with mechanistic modelling. We demonstrate the basin-specific predictive performance for 569 catchments of the continental United States. For exemplary basins, we analyse the dynamics of states and processes learned by the model-internal neural networks. Finally, we discuss the potential of neural ODE models in hydrology.


/pdf/improving-hydrologic-models-for-predictions-and-process-19un5m8g.pdf

Improving hydrologic models for predictions and process understanding using neural ODEs

It has been proposed that conservation laws might not be beneficial for accurate hydrological modelling due to errors in input (precipitation) and target (streamflow) data (particularly at the event time scale), and this might explain why deep learning models (which are not based on enforcing closure) can out‐perform catchment‐scale conceptual and process‐based models at predicting streamflow. We test this hypothesis with two forcing datasets that disagree in total, long‐term precipitation. We analyse the roll of strictly enforced mass conservation for matching a long‐term mass balance between precipitation input and streamflow output using physics‐informed (mass conserving) machine learning and find that: (1) enforcing closure in the rainfall‐runoff mass balance does appear to harm the overall skill of hydrological models; (2) deep learning models learn to account for spatiotemporally variable biases in data (3) however this ‘closure’ effect accounts for only a small fraction of the difference in predictive skill between deep learning and conceptual models.

On strictly enforced mass conservation constraints for modelling the Rainfall‐Runoff process

Abstract. This study investigates the ability of long short-term memory (LSTM) neural networks to perform streamflow prediction at ungauged basins. A set of state-of-the-art, hydrological model-dependent regionalization methods are applied to 148 catchments in northeast North America and compared to an LSTM model that uses the exact same available data as the hydrological models. While conceptual model-based methods attempt to derive parameterizations at ungauged sites from other similar or nearby catchments, the LSTM model uses all available data in the region to maximize the information content and increase its robustness. Furthermore, by design, the LSTM does not require explicit definition of hydrological processes and derives its own structure from the provided data. The LSTM networks were able to clearly outperform the hydrological models in a leave-one-out cross-validation regionalization setting on most catchments in the study area, with the LSTM model outperforming the hydrological models in 93 % to 97 % of catchments depending on the hydrological model. Furthermore, for up to 78 % of the catchments, the LSTM model was able to predict streamflow more accurately on pseudo-ungauged catchments than hydrological models calibrated on the target data, showing that the LSTM model's structure was better suited to convert the meteorological data and geophysical descriptors into streamflow than the hydrological models even calibrated to those sites in these cases. Furthermore, the LSTM model robustness was tested by varying its hyperparameters, and still outperformed hydrological models in regionalization in almost all cases. Overall, LSTM networks have the potential to change the regionalization research landscape by providing clear improvement pathways over traditional methods in the field of streamflow prediction in ungauged catchments.


/pdf/continuous-streamflow-prediction-in-ungauged-basins-long-2tivu1hg.pdf

Continuous streamflow prediction in ungauged basins: long short-term memory neural networks clearly outperform traditional hydrological models

Predicting streamflow in intermittent rivers and ephemeral streams (IRES), particularly those in climate hotspots such as the headwaters of the Colorado River in Texas, is a necessity for all planning and management endeavors associated with these ubiquitous and valuable surface water resources. In this study, the performance of three deep learning algorithms, namely Convolutional Neural Networks (CNN), Long Short-Term Memory (LSTM), and Self-Attention LSTM models, were evaluated and compared against a baseline Extreme Learning Machine (ELM) model for monthly streamflow prediction in the headwaters of the Texas Colorado River. The predictive performance of the models was assessed over the entire range of flow as well as for capturing the extreme hydrologic events (no-flow events and extreme floods) using a suite of model evaluation metrics. According to the results, the deep learning algorithms, especially the LSTM-based models, outperformed the ELM with respect to all evaluation metrics and offered overall higher accuracy and better stability (more robustness against overfitting). Unlike its deep learning counterparts, the simpler ELM model struggled to capture important components of the IRES flow time-series and failed to offer accurate estimates of the hydrologic extremes. The LSTM model (K.G.E. > 0.7, R2 > 0.75, and r > 0.85), with better evaluation metrics than the ELM and CNN algorithm, and competitive performance to the SA–LSTM model, was identified as an appropriate, effective, and parsimonious streamflow prediction tool for the headwaters of the Colorado River in Texas.

https://www.mdpi.com/2073-4441/14/19/2972/pdf?version=1663841058

Using Deep Learning Algorithms for Intermittent Streamflow Prediction in the Headwaters of the Colorado River, Texas

Abstract. Neural networks have been shown to be extremely effective rainfall-runoff models, where the river discharge is predicted from meteorological inputs. However, the question remains: what have these models learned? Is it possible to extract information about the learned relationships that map inputs to outputs, and do these mappings represent known hydrological concepts? Small-scale experiments have demonstrated that the internal states of long short-term memory networks (LSTMs), a particular neural network architecture predisposed to hydrological modelling, can be interpreted. By extracting the tensors which represent the learned translation from inputs (precipitation, temperature, and potential evapotranspiration) to outputs (discharge), this research seeks to understand what information the LSTM captures about the hydrological system. We assess the hypothesis that the LSTM replicates real-world processes and that we can extract information about these processes from the internal states of the LSTM. We examine the cell-state vector, which represents the memory of the LSTM, and explore the ways in which the LSTM learns to reproduce stores of water, such as soil moisture and snow cover. We use a simple regression approach to map the LSTM state vector to our target stores (soil moisture and snow). Good correlations (R2&gt;0.8) between the probe outputs and the target variables of interest provide evidence that the LSTM contains information that reflects known hydrological processes comparable with the concept of variable-capacity soil moisture stores. The implications of this study are threefold: (1) LSTMs reproduce known hydrological processes. (2) While conceptual models have theoretical assumptions embedded in the model a priori, the LSTM derives these from the data. These learned representations are interpretable by scientists. (3) LSTMs can be used to gain an estimate of intermediate stores of water such as soil moisture. While machine learning interpretability is still a nascent field and our approach reflects a simple technique for exploring what the model has learned, the results are robust to different initial conditions and to a variety of benchmarking experiments. We therefore argue that deep learning approaches can be used to advance our scientific goals as well as our predictive goals. 

Hydrological concept formation inside long short-term memory (LSTM) networks

Wastewater treatment removes water pollutants and wastewater reclamation provides an alternative water supply. It is believed that increasing the rate of wastewater reuse and reclamation can reduce water stress. This study aims at understanding whether reusing more wastewater can help mitigate water stress in China. Through scenario analysis, it is found that the potential for reuse of reclaimed water under a water conservation scenario is only 12-56% of the actual situation, but regional water stress under a water saving scenario is 10-82% lower than the current reality. The results show that a higher amount of reclamation does not necessarily lead to a lower water stress in one region. The potential for wastewater treatment and reuse is determined by return flows, which can reduce water use efficiency and exacerbate water stress. To effectively alleviate water stress, it is important to not only increase wastewater reuse, but also prioritize water conservation. 

The more wastewater reclamation, the less water stress?

Groundwater user groups in Tunisia face severe collective action problems. Aquifer depletion leads to empty wells and farmers&amp;#8217; unwillingness to pay water fees leads to bankrupt user groups &amp;#8211; both disastrous for the many communities that rely on irrigation agriculture for their livelihoods. What conditions or combination of conditions drive water user behaviour in a system that is governed by institutional uncertainty and bounded rationality? What conditions or interventions are effective in avoiding or delaying system collapse? What is the role of social norms, particularly trust and leadership, in overcoming collective action problems? Based on and expanding on the theory of common pool resource governance, this paper ties institutional results to environmental outcomes. The complex common-pool resource system studied here is simulated by an Agent-Based Model (ABM) of groundwater user decision-making. This systematic coupling of social and biophysical data and models offers new insights into simulating dynamic interactions between human behaviour, social norms, and the underlying resource. The project aims to provide a guideline for alternative modes of policy-making and implementation to address the main water governance challenge in Tunisia, i.e. groundwater overexploitation.&amp;#160; 

Governing Common Pool Resources in Fragile Political Systems: Modelling Behaviour, Institutions, and Social-Ecological Dynamics

&amp;lt;p&amp;gt;Austria is well endowed with water resources where the total water demand consumes only 3% of the total water supply. However, with increased water demand from agriculture, domestic and industrial sectors and increased frequency of high heat and drought occurrences from future climate change, several regions can potentially face high water stress for short periods in a year which is detrimental to the economy and the environment. The Seewinkel region located in eastern Austria is one such case where both agriculture and environment are dependent on a reliable quantity of water to be available throughout the year. Apart from precipitation, additional crop water requirement is satisfied using an irrigation infrastructure. Groundwater which is the sole source of irrigation water in the region is pumped based on the water allocations prescribed to each irrigation cooperative in the region. These allocations currently limit groundwater withdrawals for potentially higher agricultural output and maintain the natural Seewinkel wetland supported by groundwater body. With climate change induced water stress in the future, sufficient water table levels cannot be maintained throughout the year whil&amp;amp;#8203;e simultaneously trying to satisfy agricultural and environmental water demand. Our study identifies different cost optimal management strategies to optimally manage the groundwater resources shared between multiple irrigation cooperatives in the region. Some of the strategies include optimal crop-land mix, irrigation technology transitions, building new water supply infrastructure and using financial instruments. Additionally, the study aims to identify the nature of the shared common pool of groundwater to identify the potential of trading water allocation rights between irrigation cooperatives which can lead to efficient water use. The strategies have been identified based on the stakeholder interviews conducted in the region under the WaterStressAT project KR19AC0K17504 funded by the Austrian Climate Research Program twelfth call.&amp;lt;/p&amp;gt; 

Reducing future water stress in the Seewinkel region in Austria: exploring water management opportunities for the sustainable use of the shared groundwater resource.

Regina G. Ziegler

Papers

Hydrological concept formation inside long short-term memory (LSTM) networks

The more wastewater reclamation, the less water stress?

Governing Common Pool Resources in Fragile Political Systems: Modelling Behaviour, Institutions, and Social-Ecological Dynamics

Reducing future water stress in the Seewinkel region in Austria: exploring water management opportunities for the sustainable use of the shared groundwater resource.