Enhancing the predictive performance of remote sensing for ecological variables of tidal flats using encoded features from a deep learning model

ABSTRACT

Tidal flats are among the ecologically richest areas of the world where sediment composition (e.g. median grain size and silt content) and the macrozoobenthic presence play an important role in the health of the ecosystem. Regular monitoring of environmental and ecological variables is essential for sustainable management of the area. While monitoring based on field sampling is very time-consuming, the predictive performance of these variables using satellite images is low due to the spectral homogeneity over these regions. We tested a novel approach that uses features from a variational autoencoder (VAE) model to enhance the predictive performance of remote sensing images for environmental and ecological variables of tidal flats. The model was trained using the Sentinel-2 spectral bands to reproduce the input images, and during this process, the VAE model represents important information on the tidal flats within its layer structure. The information in the layers of the trained model was extracted to form features with identical spatial coverage to the spectral bands. The features and the spectral bands together form the input to random forest models to predict field observations of the sediment characteristics such as median grain size and silt content, as well as the macrozoobenthic biomass and species richness. The maximum prediction accuracy of feature-based maps was close to 62% for the sediment characteristics and 37% for benthic fauna indices. The encoded features improved the prediction accuracy of the random forest regressor model by 15% points on average in comparison to using just the spectral bands. Our method enhances the predictive performance of remote sensing, in particular the spatiotemporal dynamics in median grain size and silt content of the sediment thereby contributing to better-informed management of coastal ecosystems.

KEYWORDS:

• Convolutional neural network
• autoencoder model
• sediment characteristics
• random forest regression
• Macrozoobenthos
• Wadden Sea

Acknowledgments

This research is funded by the Netherlands Organisation for Scientific Research (NWO), Looking from space to lower levels of the food web in Wadden systems, ref nr.: ALWGO.2018.023 to which all authors are involved. The other team members are Aziza Saud Al Adhubi, Roeland Bom, Jaime Pitarch and Henk van der Veer. With big thanks to the current SIBES core team for collecting and processing the thousands of samples. In alphabetical order: Leo Boogert, Thomas de Brabander, Anne Dekinga, Sander Holthuijsen, Job ten Horn, Loran Kleine Schaars, Jeroen Kooijman, Anita Koolhaas, Franka Lotze, Simone Miguel, Luc de Monte, Dennis Mosk, Amin Niamir, Dana Nolte, Dorien Oude Luttikhuis, Bianka Rasch, Reyhane Roohi, Charlotte Saull, Juan Schiaffi, Marten Tacoma Evaline van Weerlee and Bas de Wit. We also thank all former and current employees and the many volunteers and students who have ensured that the SIBES program has continued in recent years. The RV Navicula was essential for collecting the samples and in particular we thank the current crew, Wim Jan Boon, Klaas Jan Daalder, Bram Fey, Hendrik Jan Lokhorsten and Hein de Vries. SIBES is currently financed by the Nederlandsche Aardolie Maatschappij NAM, Rijkswaterstaat RWS and the Royal NIOZ. We thank Ton Markus for preparing the figures.

All source code of this research and data will be made available at -https://github.com/logsanand/Featureextraction-VAE

Disclosure statement

No potential conflict of interest was reported by the authors.

Data availability statement

The SIBES data points for the sediment and ecological variables of the entire Dutch Wadden Sea can be obtained from the Royal NIOZ (SIBES 2020–2021 - NIOZ) upon request. The data that supports the findings of this study is available from the corresponding author, [Madhuanand, L], upon reasonable request. All related codes will be made available in GitHub once the manuscript is accepted.

Supplemental data

Supplemental data for this article can be accessed online at https://doi.org/10.1080/15481603.2022.2163048.

Additional information

Funding

The work was supported by the nederlandse organisatie voor wetenschappelijk onderzoek [ALWGO.2018.023].