skip to main content
US FlagAn official website of the United States government
dot gov icon
Official websites use .gov
A .gov website belongs to an official government organization in the United States.
https lock icon
Secure .gov websites use HTTPS
A lock ( lock ) or https:// means you've safely connected to the .gov website. Share sensitive information only on official, secure websites.

Explore Research Products in the PAR
It may take a few hours for recently added research products to appear in PAR search results.


Title: A Spatiotemporal Interrogation of Hydrologic Drought Model Performance for Machine Learning Model Interpretability
Abstract The predictive accuracy of regional hydrologic models often varies across both time and space. Interpreting relationships between watershed characteristics, hydrologic regimes, and model performance can reveal potential areas for model improvement. In this study, we use machine learning to assess model performance of a regional hydrologic model to forecast the occurrence of streamflow drought. We demonstrate our methodology using a regional long short‐term memory (LSTM) deep learning model developed by the U.S. Geological Survey (USGS) and data from 384 streamgages across the Colorado River Basin region. Performance was assessed by clustering catchments using: (a) physical and climatological catchment attributes, and (b) streamflow drought signatures time series. We examined the association of USGS LSTM model error measures with clusters generated by both approaches to interpret meaningful spatial and temporal information about LSTM model performance. Clustering static catchment attributes identified elevation, degree of streamflow regulation, baseflow contribution, catchment aridity, and drainage area as the most influential attributes to model performance. Clustering gages by their drought signatures revealed that catchments with significant seasonal peak runoff between January and June generally exhibited better model performance. Additionally, a Random Forest classifier was trained to successfully predict LSTM model performance (F1 score of 0.72) based on physical and climatological catchment attributes. Low degree of flow regulation was identified as a key indicator of better LSTM model performance. These findings point to the opportunities for improving the USGS LSTM model performance in future hydrologic drought prediction efforts across regional and CONUS scales.  more » « less
Award ID(s):
2202706 2026431
PAR ID:
10660380
Author(s) / Creator(s):
 ;  ;  ;  
Publisher / Repository:
Water Resources Research
Date Published:
Journal Name:
Water Resources Research
Volume:
61
Issue:
11
ISSN:
0043-1397
Format(s):
Medium: X
Sponsoring Org:
National Science Foundation
More Like this
  1. ; (, Hydrological Processes)
    In dry summer months, stream baseflow sourced from groundwater is essential to support aquatic ecosystems and anthropogenic water use. Hydrologic signatures, or metrics describing unique features of streamflow timeseries, are useful for quantifying and predicting these valuable baseflow and groundwater storage resources across continental scales. Hydrologic signatures can be predicted based on catchment attributes summarising climate and landscape and can be used to characterise baseflow and groundwater processes that cannot be directly measured. While past watershed‐scale studies suggest that landscape attributes are important controls on baseflow and storage processes, recent regional‐to‐global scale modelling studies have instead found that landscape attributes have weaker relationships with hydrologic signatures of these processes than expected compared to climate attributes. In this study, we quantify two landscape attributes, average geologic age and the proportion of catchment area covered by wetlands. We investigate if incorporating these additional predictors into existing large‐sample attribute datasets strengthens continental‐scale, empirical relationships between landscape attributes and hydrologic signatures. We quantify 14 hydrologic signatures related to baseflow and groundwater processes in catchments across the contiguous United States, evaluate the relationships between the new catchment attributes and hydrologic signatures with correlation analysis and use the new attributes to predict hydrologic signatures with random forest models. We found that the average geologic age of catchments was a highly influential predictor of hydrologic signatures, especially for signatures describing baseflow magnitude in catchments, and had greater importance than existing attributes of the subsurface. In contrast, we found that the proportion of wetlands in catchments had limited influence on our hydrologic signature predictions. We recommend incorporating catchment geologic age into large‐sample catchment datasets to improve predictions of baseflow and storage hydrologic signatures and processes across continental scales. 
    more » « less
  2. ; ; ; (, Water Resources Research)
    Abstract Finding similarities between model parameters across different catchments has proved to be challenging. Existing approaches struggle due to catchment heterogeneity and non‐linear dynamics. In particular, attempts to correlate catchment attributes with hydrological responses have failed due to interdependencies among variables and consequent equifinality. Machine Learning (ML), particularly the Long Short‐Term Memory (LSTM) approach, has demonstrated strong predictive and spatial regionalization performance. However, understanding the nature of the regionalization relationships remains difficult. This study proposes a novel approach to partially decouple learning the representation of (a) catchment dynamics by using theHydroLSTMarchitecture and (b) spatial regionalization relationships by using aRandom Forest(RF) clustering approach to learn the relationships between the catchment attributes and dynamics. This coupled approach, calledRegional HydroLSTM, learns a representation of “potential streamflow” using a single cell‐state, while the output gate corrects it to correspond to the temporal context of the current hydrologic regime. RF clusters mediate the relationship between catchment attributes and dynamics, allowing identification of spatially consistent hydrological regions, thereby providing insight into the factors driving spatial and temporal hydrological variability. Results suggest that by combining complementary architectures, we can enhance the interpretability of regional machine learning models in hydrology, offering a new perspective on the “catchment classification” problem. We conclude that an improved understanding of the underlying nature of hydrologic systems can be achieved by careful design of ML architectures to target the specific things we are seeking to learn from the data. 
    more » « less
  3. ; ; (, American Geophysical Union)
    In the Anthropocene, humans have altered the properties and processes of hydrological systems across scales. The extent of human intervention in the landscape limits the utility of traditional hydrological modelling schemes. Since purely hydrological conceptual models no longer fit these systems, hydrologists must integrate key human interventions into conceptual models of human-modified catchments. Despite the advances in analyzing the observed changes within the hydrological cycle using bottom-up (or reductionist) modelling approaches, the aptitude of top-down hydrologic schemes for socio-hydrological system analysis is still untested. Here we show the potential of top-down hydrological modelling human modified watersheds using anthropogenic hydrological signatures. Specifically, we assess the ability of the top-down modelling method in human-modified catchments to improve the representation hydrological signatures (e.g. mean monthly runoff, flow duration curve) while ensuring a sufficient, but not excessive, level of complexity in model formulation. First, we develop new conceptual models which include human hydrological modifications commonly identified in the literature. Then, we link these new features in the conceptual models to features in the hydrological signatures. We apply the proposed methodology to the Lake Mendocino Watershed in Northern California, US. We compare a purely hydrological model developed for this catchment based on natural watershed properties using naturalized streamflow to a hydrological model of the human-modified catchment using observed streamflow. We anticipate that the proposed approach contributes to the development of detection and attribution frameworks for key anthropogenic changes of observed hydrological variability and improved model performance in human-modified catchments. 
    more » « less
  4. ; ; ; ; (, Hydrological Processes)
    ABSTRACT We investigated the influence of landslide deposits on hydrologic connectivity and subsurface water movement in small headwater catchments in the Western Cascades, Oregon, USA. We examined isotopic variations in surface water across multiple catchments, comparing wet and dry periods to assess how antecedent moisture influences hydrologic connectivity and groundwater interactions. Seasonal shifts in δ18O values reveal that hydrologic connectivity increases during wet conditions, resulting in more uniform isotopic signatures across catchments due to enhanced vertical and lateral water movement in the subsurface. In contrast, during dry periods there was greater spatial variability in δ18O, reflecting localised groundwater contributions and reduced connectivity. Notably, some catchments with high proportions of earthflow terrain maintain consistent water isotopic ratios across seasons, suggesting persistent groundwater inputs from landslide deposits. Spatial patterns in δ18O also point to subsurface inter‐catchment flow paths facilitated by landslide deposits. Streamflow measurements during the dry season further support these findings. Catchments underlain by older, stabilised landslide deposits had highly variable unit discharge and frequent periods of flow cessation, consistent with weaker subsurface connectivity and limited water retention. In contrast, catchments draining active earthflows maintained relatively high unit discharges and perennial flow, indicating stronger subsurface linkages and greater potential for water accumulation that sustains both flow and ongoing slope movement. We estimated storage potential within landslide deposits and then used this to estimate catchment storage potential. Catchment storage was negatively correlated to variability in isotopic ratios, indicating an inverse relationship between catchment storage and variability in water sources in both space and time. Overall, our results demonstrate that geomorphic setting—particularly the presence and structure of landslide deposits—can exert strong control on the spatial distribution of hydrologic connectivity in mountain catchments. These insights improve our understanding of how subsurface properties mediate water movement and streamflow resilience under varying climate conditions. 
    more » « less
  5. ; ; ; ; ; ; ; ; ; ; et al (, Frontiers in Water)
    null (Ed.)
    Understanding and predicting catchment responses to a regional disturbance is difficult because catchments are spatially heterogeneous systems that exhibit unique moderating characteristics. Changes in precipitation composition in the Northeastern U.S. is one prominent example, where reduction in wet and dry deposition is hypothesized to have caused increased dissolved organic carbon (DOC) export from many northern hemisphere forested catchments; however, findings from different locations contradict each other. Using shifts in acid deposition as a test case, we illustrate an iterative “process and pattern” approach to investigate the role of catchment characteristics in modulating the steam DOC response. We use a novel dataset that integrates regional and catchment-scale atmospheric deposition data, catchment characteristics and co-located stream Q and stream chemistry data. We use these data to investigate opportunities and limitations of a pattern-to-process approach where we explore regional patterns of reduced acid deposition, catchment characteristics and stream DOC response and specific soil processes at select locations. For pattern investigation, we quantify long-term trends of flow-adjusted DOC concentrations in stream water, along with wet deposition trends in sulfate, for USGS headwater catchments using Seasonal Kendall tests and then compare trend results to catchment attributes. Our investigation of climatic, topographic, and hydrologic catchment attributes vs. directionality of DOC trends suggests soil depth and catchment connectivity as possible modulating factors for DOC concentrations. This informed our process-to-pattern investigation, in which we experimentally simulated increased and decreased acid deposition on soil cores from catchments of contrasting long-term DOC response [Sleepers River Research Watershed (SRRW) for long-term increases in DOC and the Susquehanna Shale Hills Critical Zone Observatory (SSHCZO) for long-term decreases in DOC]. SRRW soils generally released more DOC than SSHCZO soils and losses into recovery solutions were higher. Scanning electron microscope imaging indicates a significant DOC contribution from destabilizing soil aggregates mostly from hydrologically disconnected landscape positions. Results from this work illustrate the value of an iterative process and pattern approach to understand catchment-scale response to regional disturbance and suggest opportunities for further investigations. 
    more » « less
  • Citation Formats
  • MLA
  • APA
  • Chicago
  • BibTeX
  • Text Encoded Conversion
  • XML
  • Save / Share this Record
  • Send to Email