A Moisture-Insensitive NPV Hyperspectral Index (MINI) for Vegetation Cover Estimation in Coastal Saltmarsh Wetlands

Accurate monitoring of non-photosynthetic vegetation (NPV) in coastal saltmarsh wetlands is crucial for understanding ecosystem functioning and carbon dynamics. Existing hyperspectral NPV indices are used to monitor NPV coverage. However, their stability may be compromised under dynamic moisture conditions influenced by tidal fluctuations. This study developed a moisture-insensitive non-photosynthetic vegetation index (MINI) for reliable estimation of NPV, photosynthetic vegetation (PV), and bare soil (BS) fractional cover in coastal saltmarsh wetlands. Laboratory spectral measurements of NPV, PV, and BS were analyzed under varying moisture conditions. Based on derivative spectral analysis and triangular space methodology, optimal band combinations (621 and 690 nm) were identified to establish the MINI. A triangular feature space formed by Normalized Difference Vegetation Index and MINI was employed to estimate the fractional cover of PV, NPV, and BS. MINI was applied to synthetic spectral mixtures representing a wide range of soil and vegetation moisture conditions, as well as to ZY-1 hyperspectral satellite imagery acquired during 3 phenological stages and tidal levels in the Yellow River Delta. In both cases, MINI outperformed existing NPV hyperspectral indices, with higher accuracy and greater robustness. For simulated data, MINI achieved coefficient of determination (R2) = 0.80 to 0.84, root mean square error (RMSE) = 0.12 to 0.18, and mean absolute deviation (MAD) = 8.1% to 9.6%; for satellite imagery validated against unmanned aerial vehicle observations, MINI maintained R2 = 0.80 to 0.82, RMSE = 8.16% to 10.64%, and MAD = 6.24% to 8.56%. These results highlight MINI’s suitability for estimating fractional cover of NPV, PV, and BS under varying moisture conditions, providing a practical and stable approach for vegetation monitoring in dynamic coastal wetlands.