Comparing Trends in Modeled and Observed Streamflows at Minimally Altered Basins in the United States

Abstract

We compared modeled and observed streamflow trends from 1984 to 2016 using five statistical transfer models and one deterministic, distributed-parameter, process-based model, for 26 flow metrics at 502 basins in the United States that are minimally influenced by development. We also looked at a measure of overall model fit and average bias. A higher percentage of basins, for all models, had relatively low trend differences between modeled and observed mean/median flows than for very high or low flows such as the annual 1-day high and 7-day low flows. Mean-flow metrics also had the largest percentage of basins with relatively good overall model fit and low bias. The five statistical transfer models performed better at more basins than the process-based model. The overall model fit for all models, for mean and/or high flows, was correlated with one or more measures of basin precipitation or aridity. Our study and previous studies generally observed good model performance for high flows up to 90th or 95th percentile flows. However, we found model performance was substantially worse for more extreme flows, including 99th percentile and annual 1-day high flows, indicating the importance of including more extreme high flows in analyses of model performance.

The consequences of neglecting reservoir storage in national‐scale hydrologic models: An appraisal of key streamflow statistics

“…The accuracy of the models used in the current study can be seen in the results of Hodgkins et al (2020) for unregulated basins in the CONUS.…”

“…However, it is only feasible to collect long-term streamflow data on a very small percentage of these stream reaches, and there are many spatial and temporal gaps in observed streamflow records (e.g., Kiang et al, 2013). For ungaged reaches of interest, methods are needed to estimate streamflow; this can be done with either process-based models or statistical-transfer models (Hodgkins et al, 2020).…”

Modeling Functional Flows in California’s Rivers

“…For example, catchment elevation was important in predicting the duration of the 2-years flood, spring recession duration, and dry season duration, but the physical basis for these relationships is less clear. Additional studies that offer robust comparisons between statistical and physically-based models, such as performed by Hodgkins et al (2020), would be helpful for evaluating the benefits and limitations of different hydrologic modeling approaches in predicting functional flows and supporting environmental flow applications.…”

Integrated water resources trend assessments: State of the science, challenges, and opportunities for advancement

“…The effect of land use on surface water quantity trends is often examined using methods that also account for climate and water use. For example, the National Hydrologic Model Precipitation Runoff Modeling System has been use to evaluate the effects of precipitation, air temperature, and land use on changes in streamflow (Hodgkins et al, 2020), and Kemter et al (2023) developed machine‐learning models to better understand the relative influences of climate and land use on streamflow trends at a national scale. A time‐varying Budyko framework and the Mann‐Kendall test have been used to investigate the influence of both climate and land use on hydrology at continental and regional scales (Ayers et al, 2019; Li & Quiring, 2021).…”

“…Generalizing hydrologic trends observed at monitoring locations to unmonitored areas to increase the spatial coverage of trend results typically involves geostatistical/regional (McCabe & Wolock, 2014), process‐based (Hodgkins et al, 2020), or artificial intelligence/machine‐learning (Miller et al, 2018; Rice et al, 2015) models, which first predict daily or monthly streamflow and then derive long‐term trends from these predictions. While process‐based models hold the greatest potential for understanding physical drivers, reproducing observed long‐term trends is often not a focus of model calibration, which may result in poor predictive performance of trends.…”