Mapping Mangrove Forests Based on Multi-Tidal High-Resolution Satellite Imagery

Qing Xia, Cheng‐Zhi Qin, Li He, Chong Huang, Fenzhen SuRemote 2018 · DOI

TL;DR

This paper introduces a novel method for accurately mapping mangrove forests, even those submerged by high tides, by leveraging multi-tidal, high-resolution satellite imagery. Their approach, which incorporates a new "submerged mangrove recognition index" (SMRI) and an object-based Support Vector Machine (SVM) classifier, significantly improves mapping accuracy compared to traditional single-tide methods.

What Problem Does This Paper Address?

Mangrove forests are vital coastal ecosystems that provide numerous ecological and economic benefits, yet they are facing significant decline. Accurate mapping is crucial for conservation efforts. However, a major challenge in mapping mangroves is their location in intertidal zones, meaning they are periodically submerged by tides. Traditional remote sensing methods, often relying on single-date imagery, struggle to capture the full extent of mangrove forests, especially when images are taken during high tide. This leads to an underestimation of their distribution. The paper aims to overcome this limitation by developing a method that accounts for the spectral changes caused by tidal inundation.

Methods

The core of the proposed method lies in its ability to differentiate mangroves under varying tidal conditions. Key techniques include:

  • Multi-Tidal Imagery Acquisition: The study utilizes high-resolution satellite imagery captured at both low and high tide. This is crucial for observing the spectral differences caused by water inundation.
  • Submerged Mangrove Recognition Index (SMRI): This novel index is designed to quantify the spectral signature differences of mangroves between high and low tide. It specifically leverages the near-infrared band, which is sensitive to water content and vegetation.
  • Feature Extraction: Beyond the SMRI, the method incorporates:
  • Normalized Difference Vegetation Index (NDVI): A standard index used to identify vegetation.
  • Textural Features: Homogeneity, contrast, entropy, and correlation are extracted using the Grey Level Co-occurrence Matrix (GLCM) to capture spatial patterns within the imagery.
  • Geographic Features: Features like the maximum distance to water are used to help distinguish mangroves from similar land cover types.
  • Object-Based Image Analysis (OBIA) with Support Vector Machine (SVM): The extracted features are fed into an object-based SVM classifier. OBIA segments the image into meaningful objects, reducing "salt-and-pepper" noise and allowing for the incorporation of spatial and textural properties. The radial basis function kernel SVM is employed for its effectiveness.

Datasets

The study uses GF-1 satellite imagery, which provides high spatial resolution (2m for panchromatic, 8m for multispectral). Specifically, two GF-1 images were used:

  • A low-tide image captured on June 2, 2016.
  • A high-tide image captured on August 31, 2017.

Field sample points were collected for training and validation, along with data from Google Earth. A detailed land cover map for 2017 from the Guangxi Shankou National Mangrove Forest Nature Reserve Administration was also used.

Key Insights and Ideas

  • Tidal inundation is a critical factor: The spectral signatures of mangroves change dramatically with tidal levels, making single-date imagery insufficient for accurate mapping.
  • The SMRI is a powerful discriminator: The newly developed SMRI effectively highlights submerged mangroves, which are often missed by traditional methods. This is particularly important for capturing the full extent of mangrove ecosystems.
  • Combining spectral, textural, and geographic features is key: No single feature type is sufficient. The synergistic use of SMRI, NDVI, textural information, and geographic context significantly boosts classification accuracy.
  • Object-based analysis enhances robustness: Segmenting the imagery into objects before classification helps to overcome spectral variability within classes and reduces classification errors.

Key Results

The proposed multi-tidal, object-based SVM method achieved highly satisfactory results:

  • Overall Accuracy: 94%
  • Kappa Coefficient: 0.86

These results were superior to those obtained from object-based SVMs using only single-tide imagery. For instance, a method using only high-tide imagery with SMRI (SVMh + SMRI) achieved 91% overall accuracy and a Kappa of 0.79, while a method using only high-tide imagery without SMRI (SVMh) achieved only 84% overall accuracy and a Kappa of 0.60. The low-tide based methods also showed improvement with SMRI. The study also demonstrated that the SMRI effectively distinguishes slightly submerged mangroves.

Limitations and Future Work

  • Underestimation at extreme high tides: The paper notes that even with the SMRI, mangrove areas that are deeply submerged by water during extreme high tides might still be underestimated, as their spectral signature can become indistinguishable from water.
  • Data availability: Acquiring cloud-free, high-resolution imagery at precise low and high tide moments can be challenging due to satellite revisit cycles.
  • Generalizability: While the method was validated in a specific case study area, further testing in diverse mangrove environments would be beneficial.

The authors suggest that their method provides a robust framework for mapping mangrove forests, and future work could focus on refining the SMRI or exploring other spectral indices to further improve accuracy, especially in challenging tidal conditions.

Summary updated

Abstract

Mangrove forests, which are essential for stabilizing coastal ecosystems, have been suffering from a dramatic decline over the past several decades. Mapping mangrove forests using satellite imagery is an efficient way to provide key data for mangrove forest conservation. Since mangrove forests are periodically submerged by tides, current methods of mapping mangrove forests, which are normally based on single-date, remote-sensing imagery, often underestimate the spatial distribution of mangrove forests, especially when the images used were recorded during high-tide periods. In this paper, we propose a new method of mapping mangrove forests based on multi-tide, high-resolution satellite imagery. In the proposed method, a submerged mangrove recognition index (SMRI), which is based on the differential spectral signature of mangroves under high and low tides from multi-tide, high-resolution satellite imagery, is designed to identify submerged mangrove forests. The proposed method applies the SMRI values, together with textural features extracted from high-resolution imagery and geographical features of mangrove forests, to an object-based support vector machine (SVM) to map mangrove forests. The proposed method was evaluated via a case study with GF-1 images (high-resolution satellites launched by China) in Yulin City, Guangxi Zhuang Autonomous Region of China. The results show that our proposed method achieves satisfactory performance, with a kappa coefficient of 0.86 and an overall accuracy of 94%, which is better than results obtained from object-based SVMs that use only single-date, remote sensing imagery.
BibTeX
@inproceedings{xia2018mapping,
  author = {Qing Xia and Cheng‐Zhi Qin and Li He and Chong Huang and Fenzhen Su},
  title = {{Mapping Mangrove Forests Based on Multi-Tidal High-Resolution Satellite Imagery}},
  booktitle = {Remote},
  year = {2018},
  doi = {10.3390/rs10091343},
}
Figure