Evaluation of daily stream temperature predictions (1979–2021) across the contiguous United States using a spatiotemporal aware machine learning algorithm

Stream temperature controls a variety of physical and biological processes that affect ecosystems, human health, and economic activities. We used 42 years (1979-2021) of data to predict daily summary statistics of stream temperature across >50,000 stream reaches in the contiguous United States using a recurrent graph convolution network. We comprehensively documented the performance – both across all reaches and by stream type (e.g., reservoir or groundwater influence) – as a baseline for future improvement. The model showed reach-level RMSE of <2°C with 90% prediction intervals that contain 90.7% of observations. We also assessed how the model captured variability in ecologically relevant metrics (e.g., R 2 for annual 7-day maximum = 0.76; R 2 for days exceeding 25°C = 0.75). This model does not outperform state-of-the-art machine learning efforts (e.g., RMSE ≤1.5°C) due to a limited input set but does provide the most spatially complete modeling to date to support water availability assessments. • We produced CONUS-extent predictions of daily temperature metrics from 1979-2021 • A spatiotemporally aware deep learning model performed well with RMSE <2°C • Model uncertainty had high observational adherence to 90% prediction intervals • Sites influenced by groundwater and reservoirs displayed higher error • The model captures variability in ecologically relevant temperature metrics