Scholarlink

<p>We developed a model to predict surface water temperature across U.S. lakes using satellite remote sensing and in situ observations to enhance cyanobacterial harmful algal bloom (cyanoHAB) forecasting. The study focused on Sentinel-3 Ocean and Land Colour Instrument (OLCI) sensor resolved lakes. We developed random forest models using both Landsat-derived and in-situ-measured surface water temperature. Landsat models offered broad spatial and temporal coverage of all OLCI resolved lakes, but they were sensitive to cloud cover and required filtering to minimize error. In contrast, the in situ model represented fewer OLCI resolved lakes, but yielded lower mean absolute error and bias. The models predicted lake surface temperature across the entire calendar year, with best performance (RMSE<sub>applied</sub> = 1.11; bias<sub>applied</sub> = 0.01; MAE<sub>applied</sub> = 0.77) from the in situ model. This approach allowed for the continuous prediction of lake surface temperatures from 1.1 to 31.6&#xa0;°C for unfrozen, open‑water conditions critical for improving the accuracy of cyanoHAB forecasting. A key strength of this study was the use of an extensive dataset and model validation against in situ observations, which improved predictive accuracy throughout the year across all seasons. The predictive model offers a water resource tool for management, ecosystem protection, and public health.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Evaluating satellite and modeled lake surface water temperature across the contiguous United States

  • 期刊论文

  • 发表日期：

    2026年3月11日

查看全文
  • Blake A. Schaeffer,
  • Hannah Ferriby,
  • Wilson Salls,
  • Natalie Reynolds,
  • Jeffrey W. Hollister,
  • Betty Kreakie,
  • Stephen Shivers,
  • Brent Johnson,
  • Olivia Cronin-Golomb,
  • Kate Meyers,
  • Maxwell Beal

摘要

We developed a model to predict surface water temperature across U.S. lakes using satellite remote sensing and in situ observations to enhance cyanobacterial harmful algal bloom (cyanoHAB) forecasting. The study focused on Sentinel-3 Ocean and Land Colour Instrument (OLCI) sensor resolved lakes. We developed random forest models using both Landsat-derived and in-situ-measured surface water temperature. Landsat models offered broad spatial and temporal coverage of all OLCI resolved lakes, but they were sensitive to cloud cover and required filtering to minimize error. In contrast, the in situ model represented fewer OLCI resolved lakes, but yielded lower mean absolute error and bias. The models predicted lake surface temperature across the entire calendar year, with best performance (RMSEapplied = 1.11; biasapplied = 0.01; MAEapplied = 0.77) from the in situ model. This approach allowed for the continuous prediction of lake surface temperatures from 1.1 to 31.6 °C for unfrozen, open‑water conditions critical for improving the accuracy of cyanoHAB forecasting. A key strength of this study was the use of an extensive dataset and model validation against in situ observations, which improved predictive accuracy throughout the year across all seasons. The predictive model offers a water resource tool for management, ecosystem protection, and public health.