Estimation Models for Pixel-Scale Coverage of Aquatic Vegetation in Lakes Based on Landsat and Sentinel Data

Aquatic vegetation (AV) plays a vital role in maintaining lake ecosystem balance and regulating carbon cycling. Accurate estimation of AV coverage or area is crucial for assessing lake stability and carbon sequestration potential. Current satellite-based methods for AV coverage estimation rely on co...

Full description

Saved in:
Bibliographic Details
Main Authors: Haitao Qin, Juhua Luo, Ying Xu, Yihao Xin, Qitao Xiao, Yinguo Qiu, Shuying Bai, Hongtao Duan
Format: Article
Language:English
Published: American Association for the Advancement of Science (AAAS) 2025-01-01
Series:Journal of Remote Sensing
Online Access:https://spj.science.org/doi/10.34133/remotesensing.0616
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Aquatic vegetation (AV) plays a vital role in maintaining lake ecosystem balance and regulating carbon cycling. Accurate estimation of AV coverage or area is crucial for assessing lake stability and carbon sequestration potential. Current satellite-based methods for AV coverage estimation rely on conventional binary classification (presence/absence within pixels), failing to quantify fractional coverage proportions critical for enhancing carbon stock assessment accuracy. To overcome this limitation, this study introduces a novel stepwise upscaling approach that integrates multiscale remote sensing data. By combining unmanned aerial vehicle (UAV), Sentinel-2 multispectral imager, and Landsat 8 operational land imager (OLI) data, we developed the first satellite pixel-scale AV coverage estimation model that accurately quantifies both the AV coverage and area at the pixel level. Initially, a set of coverage samples matched with Sentinel-2 satellite pixels (10 m × 10 m) was constructed on the basis of the high-precision AV coverage maps obtained from UAV images. Subsequently, the Sentinel-based model was established on the basis of the samples and corresponding spectral features derived from Sentinel (R2 = 0.95, RMSE = 7.85%, MAE = 5.35%). Similarly, we constructed a set of AV coverage samples matching Landsat satellite pixels (30 m × 30 m) based on a refined coverage map obtained from the Sentinel-based model. Based on these samples and Landsat 8 OLI spectral features, we developed the Landsat-based model (R2 = 0.95, RMSE = 7.85%, MAE = 5.35%). Applying the Landsat-based model, we mapped AV spatiotemporal dynamics in 42 lakes (>50 km2) in the middle and lower reaches of the Yangtze River and Huai River basins (MLY-HRB) from 1990 to 2022. The total AV area in 2022 was 4,896.4 km2, with increasing trends in MLY and decreasing trends in HRB over 3 decades. AV coverage ranged from 1.18% in Lake Chaohu to 66.43% in Lake Dongting. The satellite pixel-scale model developed in this study demonstrates high accuracy, robustness, and scalability. It is expected to provide a reliable framework for ecosystem carbon sink quantification and carbon sequestration potential assessments in global lakes.
ISSN:2694-1589