Geology topics

Scott Steinschneider

Publications and source records attributed to Scott Steinschneider.

7 recordsLinked to original sources

USGS2026

Deep learning error post-processing improves stochastic watershed modeling

Hydrologic extremes, including floods and droughts, pose substantial societal risks that are expected to intensify with climate change. Deterministic watershed models (DWMs) remain a mainstay for modeling these extremes, but lack explicit representation of uncertainty, limiting their utility for risk-informed planning. Stochastic watershed models (SWMs) address this limitation by generating ensembles of streamflow via models of observed DWM residuals. However, most SWMs struggle with the complex dependence between DWM residuals and the underlying hydrologic state, which can complicate stochastic simulations under nonstationary climates. Deep learning (DL) models, whether used as standalone models or post-processors for process-based DWMs, offer a pathway to address this challenge by reducing conditional dependence. In this study, we evaluate SWMs applied to seven models: three process-based models (PRMS, Hymod, and HBV), their hybrid process-DL counterparts, and a pure DL DWM, focusing on daily simulations and extremes under both historical conditions and synthetic climate change scenarios. Results for a case study watershed in Massachusetts show that SWMs applied to hybrid or pure DL DWMs consistently outperform those applied to process-based DWMs. However, an SWM applied to the pure DL model exhibits weaknesses at low flows for this study basin, underscoring the value of hybrid approaches. Extending the analysis across 73 additional basins demonstrates that these improvements are robust and generalizable statewide. This work highlights the potential of a DL-enhanced stochastic watershed modeling framework to advance hydrologic risk prediction under changing climate conditions, offering a scalable methodology for integrating uncertainty into watershed modeling for long-term planning.

Journal of Hydrology↗

USGS2024

Characterizing future streamflows in Massachusetts using stochastic modeling—A pilot study

Communities throughout Massachusetts face the potential effects of climate change, ranging from more extreme rainfall to more pronounced and frequent droughts. Understanding the effects of climate change on hydrology is important to State and community officials to evaluate the potential effects on infrastructure and water systems. To better understand the effects of climate change on hydrology, the U.S. Geological Survey, in partnership with Cornell University and Tufts University, conducted a study in cooperation with the Massachusetts Executive Office of Energy and Environmental Affairs to develop tools for projecting 21st-century climate and hydrologic characteristics in Massachusetts. A stochastic weather generator was developed to project future climatic characteristics for Massachusetts. The stochastic weather generator estimates daily precipitation, minimum temperature, and maximum temperature for 17 warming scenarios (from 0 to 8 degrees Celsius, in 0.5-degree increments). To project future hydrologic characteristics, the stochastic weather generator output data were input to the Precipitation-Watershed Modeling System deterministic watershed model for the Squannacook River watershed, which is the watershed selected as the pilot study location for investigating future hydrologic characteristics. Hydrologic data output from the deterministic watershed model were then input to a stochastic watershed model developed for this study to correct model errors (model errors are often observed in the output from deterministic models at the high- and low-flow extremes). The output from the stochastic watershed model was then used to characterize hydrology for the 17 warming scenarios. For the Squannacook River watershed, the results project more extreme flood and low streamflows under the warming scenarios. Output from the tools allows the characterization of future streamflows for the years 2030, 2050, 2070, and 2090, which expands our understanding of 21st-century climatic and hydrologic risk in Massachusetts. These tools could improve Federal, State, and community officials’ ability to mitigate the effects of climate change over the next several decades.

Massachusetts↗

USGS2023

Stochastic watershed model ensembles for long-range planning: Verification and validation

Deterministic watershed models (DWMs) are used in nearly all hydrologic planning, design, and management activities, yet they cannot generate streamflow ensembles needed for hydrologic risk management (HRM). The stochastic component of DWMs is often ignored in practice, leading to a systematic bias in extreme events. Since traditional stochastic streamflow models used in HRM struggle to account for anthropogenic change, there is a need to convert DWMs into stochastic watershed models (SWMs) to generate ensembles for use in HRM. A DWM can be converted to an SWM using a post-processing (pp) approach to add error to the DWM predictions. Many pp methods advanced in the area of flood forecasting are useful in HRM and for correcting extreme event biases. Selecting a suitable error model for pp is challenging due to nonnormality, skewness, heteroscedasticity, and autocorrelation. We develop a parsimonious pp method based on an autoregressive (AR) model of the logarithm of the ratio of the observations and simulations, which leads to AR model residuals that are approximately symmetric and independent. We document the value of pp for improving flood and low flow frequency analysis and we reintroduce the concepts of verification and validation of stochastic streamflow ensembles to ensure that the SWM can reproduce both statistics it was and was not designed to reproduce, respectively. These concepts are illustrated on a Massachusetts basin using the USGS Precipitation Runoff Modeling System, with an additional analysis indicating the approach may be applicable to 1,225 other sites across the United States.

Massachusetts, New Hampshire↗

USGS2019

Copula theory as a generalized framework for flow-duration curve-based streamflow estimates in ungaged and partially gaged catchments

Flow‐duration curve (FDC) based streamflow estimation methods involve estimating an FDC at an ungaged or partially gaged location and using the time series of nonexceedance probabilities estimated from donor streamgage sites to generate estimates of streamflow. We develop a mathematical framework to illustrate the connection between copulas and prior FDC‐based approaches. The performance of copula methods is compared to several other streamflow estimation methods using a decade of daily streamflow data from 74 sites located within two river basins in the southeast United States with different climate characteristics and physiographic properties. We show that copula approaches: (1) outperform other methods in the limiting case of perfect information with regard to the rank‐based correlation structure and FDCs across the gaging network; (2) provide a hedge against poor performance when donor information becomes sparser and less informative; (3) outperform other methods when used for partially gaged sites with several years of available data; and (4) remain a competitive albeit nondominating method for ungaged sites and partially gaged sites with limited data when realistic error is introduced in the estimation of FDCs and correlations across the gaging network.

Water Resources Research↗

USGS2019

Prediction and inference of flow-duration curves using multi-output neural networks

We develop multi-output neural network models (MNNs) to predict flow-duration curves (FDCs) in 9,203 ungaged locations in the Southeastern United States for six decades between 1950-2009. The model architecture contains multiple response variables in the output layer that correspond to individual quantiles along the FDC. During training, predictions are made for each quantile, and a combined loss function is used for back propagation and parameter updating. The loss function accounts for the covariance between the quantiles and generates physically consistent outputs (i.e., monotonically increasing quantiles with increasing nonexceedance probabilities). We use neural-network dropout to generate posterior-predictive distributions for FDCs, and test model performance under cross validation. Finally, we demonstrate how local surrgotate models, via the Local Interpretable Model-agnostic Explanations (LIME) method, can be used to infer the relation between basin characteristics and the predicted FDCs. Results suggest that MNNs can learn the monotonic relations between adjacent quantiles on an FDC, they result in better predictions than single output neural-network models that predict each quantile independently, and basin characteristics are most useful for predicting smaller quantiles, whereas bias terms from neighboring quantiles are most informative for predicting higher quantiles.

Water Resources Research↗

USGS2018

Drivers of variability in public‐supply water use across the contiguous United States

This study explores the relationship between municipal water use and an array of climate, economic, behavioral, and policy variables across the contiguous U.S. The relationship is explored using Bayesian‐hierarchical regression models for over 2,500 counties, 18 covariates, and three higher‐level grouping variables. Additionally, a second analysis is included for 83 cities where water price and water conservation policy information is available. A hierarchical model using the nine climate regions (product of National Oceanic and Atmospheric Administration) as the higher‐level groups results in the best out‐of‐sample performance, as estimated by the Widely Available Information Criterion, compared to counties grouped by urban continuum classification or primary economic activity. The regression coefficients indicate that the controls on water use are not uniform across the nation: e.g., counties in the Northeast and Northwest climate regions are more sensitive to social variables, whereas counties in the Southwest and East North Central climate regions are more sensitive to environmental variables. For the national city‐level model, it appears that arid cities with a high cost of living and relatively low water bills sell more water per customer, but as with the county‐level model, the effect of each variable depends heavily on where a city is located.

Water Resources Research↗

USGS2018

Efficiently approximating the Pareto frontier: Hydropower dam placement in the Amazon basin

Real–world problems are often not fully characterized by a single optimal solution, as they frequently involve multiple competing objectives; it is therefore important to identify the so-called Pareto frontier, which captures solution trade-offs. We propose a fully polynomial-time approximation scheme based on Dynamic Programming (DP) for computing a polynomially succinct curve that approximates the Pareto frontier to within an arbitrarily small > 0 on treestructured networks. Given a set of objectives, our approximation scheme runs in time polynomial in the size of the instance and 1/. We also propose a Mixed Integer Programming (MIP) scheme to approximate the Pareto frontier. The DP and MIP Pareto frontier approaches have complementary strengths and are surprisingly effective. We provide empirical results showing that our methods outperform other approaches in efficiency and accuracy. Our work is motivated by a problem in computational sustainability concerning the proliferation of hydropower dams throughout the Amazon basin. Our goal is to support decision-makers in evaluating impacted ecosystem services on the full scale of the Amazon basin. Our work is general and can be applied to approximate the Pareto frontier of a variety of multiobjective problems on tree-structured networks.

Conference Paper↗