As an important part of intertidal ecosystem, mudflat have unique environmental regulation services and ecological benefits such as maintaining the stability of coastline, accelerating material exchange and promoting carbon cycle. Accurate and timely assessment of the current status of intertidal wetlands is crucial for achieving sustainable management goals. This paper uses Google Earth Engine (GEE) cloud computing platform, selects Sentinel-2 dense time series remote sensing images in 2020, integrates the Maximum spectral index composite (MSIC) algorithm and Otsu algorithm to build a multi-level decision tree classification model, and realizes the rapid and automatic extraction of mudflat in the intertidal zone of Australia. The spatial distribution data set of mudflat in Australia's high resolution intertidal zone in 2020 was obtained through vectorization. The extracted mudflat area was 10708.22 km2, the overall accuracy was 95.32%, and the Kappa coefficient was 0.94. The storage format of this dataset is. shp, with a time resolution of years and a spatial resolution of 10m, and a data volume of 87.8M.
With the help of Google Earth Engine (GEE) cloud computing platform, the remote sensing images of Sentinel-2 dense time series in 2020 are selected, and the multi-level decision tree classification model is built by integrating the Maximum spectral index composite (MSIC) algorithm and Otsu algorithm to achieve rapid and automatic extraction of mudflat in the intertidal zone of Australia.
This research is based on GEE's 2020 intensive time series Sentinel-2 Satellite imagery, integrated with MSIC and Otsu algorithms, and mapped the spatial distribution map of Australia's Intertidal zone mudflat in 2020. Calculate the normalized differential vegetation index and the corrected normalized differential water index, and insert them as two new bands of the image into the image set to construct the Sentinel-2 high-quality dense temporal image set. (2) Select the mNDWI and NDVI bands for MSIC image synthesis, and generate the highest and lowest tide images, respectively. Apply the Otsu algorithm to the climax image to obtain water and non water bodies. Since the upper limit of mudflat is artificial shoreline (dike, road, etc.), and it is closely connected with seawater, the maximum water surface of Intertidal zone can be obtained by retaining the idea of the largest water body patch. (3) Using the maximum water surface mask of Intertidal zone to extract the lowest tide image, a low tide image of the maximum tidal submerged area containing seawater, mudflat and Intertidal zone vegetation is obtained. Otsu algorithm is applied to remove high NDVI pixels to avoid the impact of Intertidal zone vegetation on mudflat extraction, and a seawater mudflat image is obtained. Otsu is applied to seawater mudflat image to realize fast, high-precision and robust extraction of mudflat. Through vectorization and superposition analysis, the spatial distribution data of Intertidal zone mudflat in Australia in 2020 is finally obtained.
In the 2020 Australian high-resolution intertidal zone mudflat spatial distribution data set, the extracted mudflat area is 10708.22 km2, the overall accuracy is 95.32%, and the Kappa coefficient is 0.94. The storage format of this dataset is. shp, with a time resolution of years and a spatial resolution of 10m, and a data volume of 87.8 MB.
| # | number | name | type |
| 1 | 42001383 | National Natural Science Foundation of China | |
| 2 | ZR2020QD020 | National Natural Science Foundation of China | |
| 3 | 42101399 | National Natural Science Foundation of China |
This work is licensed under a
Creative
Commons Attribution 4.0 International License.
CSTR:11738.11.NCDC.RS.DB2883.2023
10.12072/ncdc.rs.db2883.2023