Improved Spectral Water Index Combined with Otsu Algorithm to Extract Muddy Coastline Data

Tang, Wei; Zhao, Chengyi; Lin, Jing; Jiao, Caixia; Zheng, Guanghui; Zhu, Jianting; Pan, Xishan; Han, Xue

doi:10.3390/w14060855

Cited by 12 publications

(10 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…The Otsu algorithm (Otsu, 1979) uses the histogram of the synthetic feature (the colored area in Figure 4a) as input and finds a threshold value that classifies the pixels of the synthetic feature as vegetated or unvegetated (Figure 4b) while maximizing the interclass variance between vegetated and unvegetated pixels. Although this method has been successfully applied to the extraction of treelines and coastlines (He et al., 2020; Tang et al., 2022; Wang et al., 2017), the Otsu algorithm does not consider the spatial texture that dictates a grassline, that is, a sharp transition from vegetation to non‐vegetation. Therefore, the boundary pixels detected by the Otsu algorithm between vegetated and unvegetated pixels indicate the approximate position of the grassline rather than its precise position.…”

Section: Methodsmentioning

confidence: 99%

A remote sensing method for mapping alpine grasslines based on graph‐cut

Liu,

Chen,

Shen

et al. 2023

Global Change Biology

View full text Add to dashboard Cite

Climate change has induced substantial shifts in vegetation boundaries such as alpine treelines and shrublines, with widespread ecological and climatic influences. However, spatial and temporal changes in the upper elevational limit of alpine grasslands (“alpine grasslines”) are still poorly understood due to lack of field observations and remote sensing estimates. In this study, taking the Tibetan Plateau as an example, we propose a novel method for automatically identifying alpine grasslines from multi‐source remote sensing data and determining their positions at 30‐m spatial resolution. We first identified 2895 mountains potentially having alpine grasslines. On each mountain, we identified a narrow area around the upper elevational limit of alpine grasslands where the alpine grassline was potentially located. Then, we used linear discriminant analysis to adaptively generate from Landsat reflectance features a synthetic feature that maximized the difference between vegetated and unvegetated pixels in each of these areas. After that, we designed a graph‐cut algorithm to integrate the advantages of the Otsu and Canny approaches, which was used to determine the precise position of the alpine grassline from the synthetic feature image. Validation against alpine grasslines visually interpreted from a large number of high‐spatial‐resolution images showed a high level of accuracy (R2, .99 and .98; mean absolute error, 22.6 and 36.2 m, vs. drone and PlanetScope images, respectively). Across the Tibetan Plateau, the alpine grassline elevation ranged from 4038 to 5380 m (5th–95th percentile), lower in the northeast and southeast and higher in the southwest. This study provides a method for remotely sensing alpine grasslines for the first‐time at large scale and lays a foundation for investigating their responses to climate change.

show abstract

Section: Methodsmentioning

confidence: 99%

A remote sensing method for mapping alpine grasslines based on graph‐cut

Liu,

Chen,

Shen

et al. 2023

Global Change Biology

View full text Add to dashboard Cite

show abstract

“…In addition, Yaru and Jixian [45] also used radar data as a Gray-Level Co-Occurrence Matrix. For the use of the Otsu method on NDWI itself, Syamani et al [46] used Otsu and NDWI in areas with mangroves, and Tang et al [47] used Otsu in the muddy area. Tang et al [47] also compared it to other indices besides NDWI, such as IWI, EWI, and AWEI.…”

Section: Thresholdingmentioning

confidence: 99%

Thirty years of change of Pekalongan coastline based on Landsat imagery

Chulafak¹,

Khomarudin²,

Ardha³

et al. 2022

IOP Conf. Ser.: Earth Environ. Sci.

View full text Add to dashboard Cite

The northern coastal area of Java Island is where many economic and industrial centers are located. Many big cities are located in this area, such as Jakarta, Semarang, and Surabaya. One of the problems in this area is the occurrence of tidal flooding. One of those affected by the tidal flood is Pekalongan, where the city also experienced land subsidence which is more severe than other big cities in Indonesia. One of the impacts of tidal flooding is the occurrence of coastal abrasion, where land areas are increasingly disappearing. This research objective is to see how big the coastline changes occurred in the Pekalongan area for 30 years through remote sensing data. Landsat image was used to see changes in coastlines from 1988 to 2021 on an annual basis. NDWI (Normalized Difference Water Index) was used to clarify the difference between land and water areas. Moreover, to reduce subjectivity, Otsu’s method was used for thresholding. The results show that most Pekalongan area is subject to abrasion, with the most severe conditions up to around 232 m. In addition, there are also problems related to tidal flooding, whose conditions are increasingly widespread every year.

show abstract

“…This method is simple, but only applicable to flat areas, and has been replaced by the multi-band inter-spectral analysis method. The classical multiband methods include Normalized Difference Water Index (NDWI) 7-9 , Modified Normalized Difference Water Index (MNDWI) [10][11][12] , and Revised Normalized Difference Water Index (RNDWI) [13][14][15] .NDWI uses near-infrared and visible bands to construct the index formula, which better suppresses the vegetation information, but buildings, shadows, mountains, and clouds all have a large impact on the results, so NDWI is only suitable for flat areas 16,17 . Based on NDWI, infrared bands are added to obtain MNDWI enhances the feature differences between water bodies and buildings, but it is still disturbed by mountains and shadows 18 .…”

mentioning

confidence: 99%

mentioning

confidence: 99%

A method for extracting small water bodies based on DEM and remote sensing images

Sun,

2024

Sci Rep

View full text Add to dashboard Cite

Accurate and comprehensive extraction of water bodies facilitates the study of watershed changes and water quality monitoring and also provides a reference for ecological evolution. When extracting tributaries and small water bodies based on remote sensing images, there are problems such as uncertain threshold selection, more edge voids, and extraction difficulties. The improved water body extraction method (Combination Normalized Difference Water Index), fuses Digital Elevation Model and remote sensing images. The river network is extracted from the Digital Elevation Model and the remote sensing image is normalized. After fusion, threshold comparison is performed, and the threshold of the river network location is used as the basis for connecting the broken water bodies, it is compared and verified with the normalized thresholds to achieve accurate identification and complete extraction of water bodies. Taking the water bodies in Xinyang City as an example, the overall accuracy of the confusion matrix was the highest, with an overall accuracy of 93.0% compared with the traditional index method. Combined with water bodies in Zhengzhou City for model validation, it has high applicability. Whether the water body is large or small, the extraction results are complete, with no voids or broken, and the accuracy is high.

show abstract

Improved Spectral Water Index Combined with Otsu Algorithm to Extract Muddy Coastline Data

Cited by 12 publications

References 15 publications

A remote sensing method for mapping alpine grasslines based on graph‐cut

A remote sensing method for mapping alpine grasslines based on graph‐cut

Thirty years of change of Pekalongan coastline based on Landsat imagery

A method for extracting small water bodies based on DEM and remote sensing images

Contact Info

Product

Resources

About