Poster: UNCERTAINTY of HYDROLOGIC EVENTS under DAKOTA’S CHANGING CONDITIONS: RESEARCH PLAN (original) (raw)
Related papers
UNCERTAINTY OF HYDROLOGIC EVENTS UNDER SOUTH DAKOTA'S CHANGING CONDITIONS: A RESEARCH AGENDA
Widespread flooding across South Dakota in 2011 has spurred a new look at the institutional, regulatory, and mathematical models used to manage the Upper Missouri River Basin as it affects all aspects of life in South Dakota. An SD EPSCoR planning grant was awarded to a team of local, national and inter-national researchers, who produced a strategy to create a research infrastructure with the goal of developing conceptual and mathematical models to understand and describe the uncertainty of hydrological events (HE) across South Dakota.
To study, describe, assess and communicate the risk associated with hydro-logic events (HE) such as flooding or drought, one must clarify the concept of uncertainty. The uncertainty in hydrological and environmental modeling has been considered for some time (Beven 1993; 2002; Walker et al. 2003; Brown 2004; 2010) and the need for a general theory of uncertainty was introduced by Lofty Zadeh (2005). To move from uncertainty as a property for informational exchange in engineering (Zadeh 2005), decision support systems (Walker et al. 2003) and mathematical theories (Dubois and Prade 2010) to the uncertainty for HE, one must consider that uncertainty has to be part of scientific knowledge and communication. Keith Beven lifted the consideration from errors in data and model generalization (Beven 1993) to a learning process (Beven 2007); I agree with him and see this learning as part of a more general process of commu-nication. To consider learning in a wider approach, I define the syste...
Review-Artificial Intelligence Based Modelling of Hydrological Processes
2012
Hydrological processes such as runoff and contaminant transport are usually affected by various complex interrelated variables. Moreover, uncertainties in variables estimate are the common stamp of these processes. Due to this complex nature, Physical modeling of any hydrological system requires availability of large, accurate and detailed data related to all influencing variables, which are not always available due to financial and technical constraints. This may lead to deficiencies in model’s performance which in turn, negatively affect hydrological planning and policy drawing. To address these shortcomings, artificial intelligence (AI) based techniques have been recently used as alternative tools to traditional physical hydrological models. These techniques have been proved to be successful and effective in tackling wide spectrum of challenging hydrological processes. This article is intended to serve as an introductory review of application of two AI techniques namely, artifici...
Uncertainty Recognition and Quantification of Hydrologic Prediction
2005
We propose a methodology to identify prediction uncertainty through recognizing and quantifying the different uncertainty sources in a hydrologic model. Statistical second moment is used as a measure of uncertainty; also an index which originated from Nash coefficient of efficiency named Model Structure Indicating Index (MSII) is proposed to quantify model structure uncertainty. The results show that MSII can well reflect the goodness of model structure, while a larger value of MSII indicating a poorer structure of hydrologic model. The index can be used as a tool for implementing model quantitative comparison (selection).
Rethink hydrologic modeling framework with AI integrating multi-processes across scales
2021
The predictability of the current earth system modeling is hampered by some critical scientific gaps, including the difficulty of capturing processes and subgrid-processes across scales, mismatch of data and model resolutions, inconsistency of system and subsystem complexities, and lack of coupling with the human system. A proposed novel hydrologic modeling framework will identify a set of AI technologies to construct hybrid models to translate across spatiotemporal scales and complexities and address resolution mismatches, incorporate data-driven causal inference and learning to explore interactions and feedbacks among processes, and develop coupling with the human system by leveraging large amount of earth and human system data. We expect that the new modeling framework will significantly improve the predictability of coupled hydrologic, terrestrial, and biogeochemical processes and outcomes.
Advances in Hydrologic Forecasts and Water Resources Management
Water
The impacts of climate change on water resources management as well as the increasing severe natural disasters over the last decades have caught global attention. Reliable and accurate hydrological forecasts are essential for efficient water resources management and the mitigation of natural disasters. While the notorious nonlinear hydrological processes make accurate forecasts a very challenging task, it requires advanced techniques to build accurate forecast models and reliable management systems. One of the newest techniques for modelling complex systems is artificial intelligence (AI). AI can replicate the way humans learn and has the great capability to efficiently extract crucial information from large amounts of data to solve complex problems. The fourteen research papers published in this Special Issue contribute significantly to the uncertainty assessment of operational hydrologic forecasting under changing environmental conditions and the promotion of water resources manag...
Science-integrated Artificial-intelligence for Flooding and precipitation Extremes (SAFE)
2021
FOCAL AREA(s) • Primary: 3 ("Insights gleaned from complex data … physicsor knowledge-guided AI") • Secondary: 2 ("Predictive modeling through the use of AI … hierarchy of models") This white paper focuses on methods that blend AI and science (e.g., physics, biogeochemistry) by (a) guiding AI cost functions, trajectories and representations with science knowledge and/or with context-specific data-driven insights in a Bayesian-inspired framework; (b) framing AI models in the context of physics-informed dynamic, causal networks; (c) merging AI-enhanced science models with science-guided explainable AI; (d) focusing on statistics/processes related to extremes and translations to risks in a changing world; & (e) identifying model parameterizations or components that must be improved to minimize risks and add maximum value to stakeholders. SCIENCE CHALLENGE A grand challenge [1-10] in hydrologic science is to understand why signals of climate change and variability, which are often visible in precipitation extremes at aggregate scales, are not consistently observed in the case of extreme flooding. However, a solution to this challenge may prove elusive unless the water cycle is viewed in an integrative manner. Thus, for riverine flooding, while Hortonian (infiltration excess) runoff may have stronger correlation with precipitation extremes and hence perhaps to warming trends or climate oscillators, Dunne (saturation excess) runoff may have a more complex relationships with time series of precipitation and with evaporation and transpiration, but rain-on-snow and snowmelt events may depend on land-surface and atmospheric temperatures. Atmospheric rivers [11] and tropical cyclones [12] lead to precipitation or flooding and are impacted by climate. Flooding assessments need to consider long-term baselines [13], evolving risk factors [14], coupled natural-human systems [15-16], and novel adaptation such as nature-inspired design [17-18]. RATIONALE The urgency of stakeholder needs pertaining to precipitation and flooding extremes requires transformative advances in long-standing challenges within earth systems sciences. Challenges for which emerging AI solutions have started to make a difference include (a) cloud physics and subgrid processes [19-24]; (b) spatiotemporal patterns and dependencies [25-29; 44]; (c) climate oscillations and teleconnections with regional hydrology [30-35]; (d) pattern search across multiple ensembles [35-36]; and (e) statistical downscaling [37-40; 45]. The AI solutions cited have ranged from machine learning (including Deep Learning), network-based approaches, and
Improving the Use of Hydrologic Probabilistic and Deterministic Information in Decision-Making
Bulletin of the American Meteorological Society, 2021
Uncertainty is everywhere and understanding how individuals understand and use forecast information to make decisions given varying levels of certainty is crucial for effectively communicating risks and weather hazards. To advance prior research about how various audiences use and understand probabilistic and deterministic hydrologic forecast information, a social science study involving multiple scenario-based focus groups and surveys at four locations (Eureka, California; Gunnison, Colorado; Durango, Colorado; Owego, New York) across the United States was conducted with professionals and residents. Focusing on the Hydrologic Ensemble Forecast System, the Advanced Hydrologic Prediction Service, and briefings, this research investigated how users tolerate divergence in probabilistic and deterministic forecasts and how deterministic and probabilistic river level forecasts can be presented simultaneously without causing confusion. This study found that probabilistic forecasts introduc...
Geoscientific Model Development, 2021
This paper presents Shyft, a novel hydrologic modeling software for streamflow forecasting targeted for use in hydropower production environments and research. The software enables rapid development and implementation in operational settings and the capability to perform distributed hydrologic modeling with multiple model and forcing configurations. Multiple models may be built up through the creation of hydrologic algorithms from a library of wellknown routines or through the creation of new routines, each defined for processes such as evapotranspiration, snow accumulation and melt, and soil water response. Key to the design of Shyft is an application programming interface (API) that provides access to all components of the framework (including the individual hydrologic routines) via Python, while maintaining high computational performance as the algorithms are implemented in modern C++. The API allows for rapid exploration of different model configurations and selection of an optimal forecast model. Several different methods may be aggregated and composed, allowing direct intercomparison of models and algorithms. In order to provide enterprise-level software, strong focus is given to computational efficiency, code quality, documentation, and test coverage. Shyft is released open-source under the GNU Lesser General Public License v3.0 and available at https: //gitlab.com/shyft-os (last access: 22 November 2020), facilitating effective cooperation between core developers, industry, and research institutions.