Tidal creeks play a vital role in influencing geospatial evolution and marsh ecological communities in coastal landscapes. However, evaluating the geospatial characteristics of numerous creeks across a site and understanding their ecological relationships pose significant challenges due to the labor-intensive nature of manual delineation from imagery. Traditional methods rely on manual annotation in GIS interfaces, which is slow and tedious. This study explores the application of Attention-based Dense U-Net (ADU-Net), a deep learning image segmentation model, for automatically classifying creek pixels in high-resolution (0.5 m) orthorectified aerial imagery in coastal Georgia, USA. We observed that ADU-Net achieved an outstanding F1 score of in identifying creek pixels, demonstrating its ability in tidal creek mapping. The study highlights the potential of deep learning models for automated tidal creek mapping, opening avenues for future investigations into the role of creeks in marshes’ response to environmental changes.
19 March 2026 in IEEE Transactions on Geoscience and Remote Sensing
Chlorophyll-a (Chl-a) concentration is a key indicator of water quality and phytoplankton biomass, used to monitor algal blooms in coastal and offshore waters. Traditional Chl-a prediction methods involve a two-step process: atmospheric correction to convert top-of-the-atmosphere (TOA) data to water-leaving radiance, followed by a machine learning or band-ratio algorithm. However, atmospheric correction is challenging in optically complex waters, reducing prediction accuracy. This study introduces a novel single-step approach that directly predicts Chl-a from TOA data using machine learning, bypassing the need for atmospheric correction. An attention-based data-driven atmospheric compensation model is also applied, incorporating contextual features such as sea surface pressure and sun geometry to enhance predictions. The experiments used real-world satellite data collected over the Chesapeake Bay and the Florida coast. The experimental results in the Florida regions demonstrate that using a multilayer perceptron (MLP) with TOA data achieves a logarithmic mean absolute error (Log MAE) of 1.830, outperforming both the Log MAE of 1.843 obtained with Level-2 reflectance data and Log MAE of 4.087 from the Chlorophyll 3-band algorithm (OC3V). The data-driven atmospheric compensation model improves this further, achieving a Log MAE of 1.745. A similar trend is observed in the Chesapeake Bay region data experiments. A key advantage of TOA data is that the model can predict Chl-a in regions previously excluded in the Level-2 reflectance data due to Level-2 quality checks, thus providing broader coverage.