Error Metrics for Streamflow Model Evaluation

Problem Statement

Streamflow modeling is vital for supporting water resources management and decision-making, including flood hazard assessment, drought monitoring, and water supply planning. However, evaluating whether model outputs are reliable enough is rarely as simple as reporting a single performance score. Different error metrics capture different aspects of model performance, and a model that excels on one metric may fail on another. The specific decision-making context should therefore guide which error metrics are used for model evaluation. To make informed decisions, hydrologists must understand what each metric reveals, what it overlooks, and how to combine complementary metrics to gain a comprehensive understanding of model performance.

Module Overview

This module introduces learners to diagnostic error metrics used in streamflow model performance evaluation. It begins by covering the fundamentals of streamflow modeling evaluation and introducing commonly used diagnostic error metrics. It then explains how to interpret metric results, considering metric strengths and limitations, aggregation, binning, transformations, and uncertainty, as well as the use of visualizations and evaluation software. The module concludes with a real-world learning activity. In the activity, learners will take the role of a water resources analyst at a regional river forecast center tasked with evaluating streamflow simulation and forecast performance for different flood events.

Prerequisites

Basic familiarity with Introductory hydrology (Recommended Hydrolearn Courses: Introduction section of Developing Storm Inflow and Outflow Hydrographs; Introduction to Open Channel Flow)

Basic familiarity with statistics in hydrology (Recommended Hydrolearn Course: Statistics for Water Resources section of Introduction to Floodplain Analysis)

Basic experience running a guided Jupyter notebook.

Basic graph interpretation, including time series plots and scatter plots.

Familiarity with NWM / USGS gauges is helpful but not required (Recommended Hydrolearn Course: National Water Model)

Learning Objectives

1) Given streamflow datasets for both continuous simulations and operational forecasts, learners will be able to explain how and why operational streamflow forecasts are different from continuous streamflow simulations.

2) Given observed, simulated, and predicted streamflow datasets, learners will be able to calculate and interpret selected streamflow error metrics, including bias, accuracy, correlation, composite, event-based, and forecast-specific metrics, to explain what they reveal about model performance.

3) Given metric results and visualizations for real-world scenarios, learners will be able to evaluate streamflow simulation and forecast performance from multiple perspectives and justify which metrics support their assessment.

This will be accomplished through activities within each section. Results from each activity will be recorded in specified results templates. The results templates for each activity can be found at the beginning of each activity. The results templates are organized so that results from one activity can be easily used in subsequent activities.

Course Authors

Dr. Jiao Wang

Postdoctoral scholar, University of California Davis

jiao.wang_hydro@outlook.com

Jenna Brooks

PhD Candidate, Northeastern University

brooks.je@northeastern.edu

Solution Keys

Completed results templates for each learning activity are available and can be requested from the course authors.

Target Audience

The target audience for this module is junior/senior students in water resources, hydrology, hydrogeology, or environmental science courses.

Tools Needed

To complete this module successfully, students will need a computer with internet access and the ability to open CSV and PDF files as well as Jupyter Notebooks.

Expected Effort

The module developers estimate that this module will take 3 hours to complete.

Course Sharing and Adaptation

This course is available for export by clicking the "Export Link" at the top right of this page. You will need a HydroLearn instructor studio account to do this. You will first need to sign up for a hydrolearn.org account, then you should register as an instructor by clicking 'studio.hydrolearn' and requesting course creation permissions.

Recommended Citation

Wang, J., Brooks, J. (2026). Identifying and Applying Error Metrics for Streamflow Modeling Evaluation. CIROH. https://edx.hydrolearn.org/courses/course-v1:CIROH_HydroLearn+CEE720+2026/about

Acknowledgement

Funding for this project was provided by the National Oceanic and Atmospheric Administration (NOAA), awarded to the Cooperative Institute for Research to Operations in Hydrology (CIROH) through the NOAA Cooperative Agreement with The University of Alabama, NA22NWS4320003. We would also like to thank Dr. Katie van Wekhoven for recommending this module topic and serving as a technical guide, informing the scope and content of this module.

CIROH: CEE720 Error Metrics for Streamflow Model Evaluation