High-Resolution Mapping Changes in the Invasion of Spartina Alterniflora in the Yellow River Delta

Abstract

The propagation of the invasive Spartina alterniflora (S. alterniflora) has seriously affected the health of coastal wetland ecosystems in China and thus requires an urgent response. In this research, we construct a feature vector set containing phenological and other time-series features based on the Google Earth Engine (GEE) platform by combining dense time-series images from the Sentinel-1 and Sentinel-2 satellites. We obtained the dataset of the annual distribution of S. alterniflora in the Yellow River Delta (YRD) from 2016 to 2021 by developing an object-oriented random forest classification model. The results show that S. alterniflora has different phenological features from other wetland plants that played an important role in its classification based on the images. A combination of multiple phenological and temporal features improved the classification accuracy of S. alterniflora (multi-year average OA: 95.38%; UA: 95.01%; PA: 95.17%). Our results suggest that from 2016 to 2021, the growth rate of the area occupied by S. alterniflora was 2.17 km 2 per year, and a new patch of the S. alterniflora appeared in the south of the study area in 2018. The work here provides scientific data to support the monitoring and control of the invasive S. alterniflora at a large scale.

Potential of Sample Migration and Explainable Machine Learning Model for Monitoring Spatiotemporal Changes of Wetland Plant Communities

“…Traditional spectral image interpretation, based on features such as color, texture, and shape, can effectively obtain samples of land cover types with significant differences. Some studies attempt to identify > REPLACE THIS LINE WITH YOUR MANUSCRIPT ID NUMBER (DOUBLE-CLICK HERE TO EDIT) < seasonal differences from multi-temporal remote sensing images to improve the discrimination between land cover types with similar spectral features for more accurate visual interpretation [16]. In addition to using multi-temporal remote sensing images, a pixel-based annual NDVI time-series curve was incorporated as a reference, to facilitate identification of different wetland plant communities.…”

“…This is attributed to the diverse features of different sensors, which allow community types to be distinguished from different perspectives. For large-scale classification of plant communities, integrating optical and radar data, such as the Sentinel-1/2, offers numerous advantages [15], [16]. These include cost-effectiveness compared to studies based on hyperspectral data [17], richer feature information than studies based on a single data source [18], and effective mitigation of the impact of clouds and cloud shadows [19].…”

“…The RF algorithm is widely used because of its powerful and flexible ability to handle complex data with varying dimensions and non-linearity [29], [30]. In recent years, numerous studies in terms of wetland mapping across different scales and species have been carried out based on the RF classifier [16], [20], [31]- [34], which has fully proved its effectiveness and robustness. However, all these studies lack deep analysis of the decision logic and feature importance of the RF classifier, which is detrimental to reviewing whether the model output results are reasonable.…”

Mapping Invasive Aquatic Plants in Sentinel-2 Images Using Convolutional Neural Networks Trained With Spectral Indices

“…The authors in [21] discriminated between water hyacinth and water primrose in Sentinel-2 images by using RF and nine spectral vegetation and water indices (NDVI, NDAVI, WAVI, SAVI, NDVIRe2, NDVIRe3, NDWI, NDII and MNDWI). The authors in [22] mapped Spartina alterniflora using Sentinel-2 and Sentinel-1 data, SAR vegetation indices, and seven spectral indices: NDVI, EVI, NDWI, land surface water index (LSWI), and automated water extraction index (AWEI). To the best of our knowledge, no studies have combined deep learning classifiers and spectral indices for detection and mapping of aquatic invasive plants.…”