Scholarlink

<p>High-throughput in situ sensing is critical for resolving the dynamic complexity of aquatic environments. Yet, achieving a balance between performance, cost and simplicity—particularly in excitation–emission matrix (EEM) fluorescence spectrometers—remains a formidable challenge. Here we reveal that high-resolution EEM spectra exhibit intrinsic redundancy and autocorrelation, and show that broadband excitation preserves essential spectral information more effectively than conventional narrowband methods at reduced sampling rates. We further demonstrate that EEM data lie on a low-dimensional manifold, enabling algorithms that reconstruct high-resolution spectra from broadband light-emitting diode (LED) excitation. Leveraging these insights, we introduce Minifluor, a custom-built portable EEM sensor, coupled with a deep generative decoding model, D4DM. Extensive validation against benchtop instruments, alongside autonomous surface vehicle measurements in an urban river, confirms the accuracy and robustness of this approach for real-time aquatic monitoring. By integrating theoretical understanding, hardware innovation and algorithmic reconstruction, this work establishes a scalable pipeline for environmental sensing and water quality management.</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

A miniaturized low-cost excitation–emission matrix sensor for aquatic environmental monitoring with preserved spectral fidelity

  • 期刊论文

  • 发表日期：

    2026年2月24日

查看全文
  • Meng Du,
  • Guan-Lin Chen,
  • Min-Jie Tong,
  • Zhuo Chen,
  • Chao-Hai Gu,
  • Chen Qian,
  • Han-Qing Yu

摘要

High-throughput in situ sensing is critical for resolving the dynamic complexity of aquatic environments. Yet, achieving a balance between performance, cost and simplicity—particularly in excitation–emission matrix (EEM) fluorescence spectrometers—remains a formidable challenge. Here we reveal that high-resolution EEM spectra exhibit intrinsic redundancy and autocorrelation, and show that broadband excitation preserves essential spectral information more effectively than conventional narrowband methods at reduced sampling rates. We further demonstrate that EEM data lie on a low-dimensional manifold, enabling algorithms that reconstruct high-resolution spectra from broadband light-emitting diode (LED) excitation. Leveraging these insights, we introduce Minifluor, a custom-built portable EEM sensor, coupled with a deep generative decoding model, D4DM. Extensive validation against benchtop instruments, alongside autonomous surface vehicle measurements in an urban river, confirms the accuracy and robustness of this approach for real-time aquatic monitoring. By integrating theoretical understanding, hardware innovation and algorithmic reconstruction, this work establishes a scalable pipeline for environmental sensing and water quality management.