How spatial resolution mediates canopy spectral diversity as a proxy for marsh plant diversity
Spectral reflectance variations comprehensively capture differences in the biochemical composition and morphological characteristics among plant species, making them a promising approach for monitoring and estimating plant diversity. However, the relationship between spectral reflectance and plant d...
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2025-12-01
|
| Series: | Ecological Informatics |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S1574954125002626 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Spectral reflectance variations comprehensively capture differences in the biochemical composition and morphological characteristics among plant species, making them a promising approach for monitoring and estimating plant diversity. However, the relationship between spectral reflectance and plant diversity is influenced by multiple factors and remains inherently unstable. Spatial resolution is one of the key factors driving the spatial heterogeneity of spectral information. Currently, it remains unclear how spatial resolution influences the spectral-plant diversity relationship in marshes and what the optimal resolution is for establishing significant correlations. This study focuses on typical marshes in Northeast China, using multispectral data acquired from unmanned aerial vehicle (UAV) at spatial resolutions ranging from 5 cm to 40 cm. Downsampling and upsampling algorithms were applied to resample the spectral data at 5 cm and 40 cm resolutions, generating datasets that cover the entire range from 5 cm to 40 cm. Spectral diversity (SD) indices, including the mean and standard deviation of KNDVI, MTCI, NDREI, and NDVI, were evaluated for their ability to predict plant species diversity across varying spatial resolutions and data sources. Results show that the predictive ability of vegetation indices (VIs) significantly declines as spatial resolution decreases to 40 cm. The optimal spatial resolution for predicting plant diversity varies among different VIs, but VIs calculated from the same spectral bands consistently show similar predictive trends. Notably, MTCI at a 10 cm resolution achieved the highest predictive accuracy for species richness (R2adj = 0.48), the Shannon-Wiener index (R2adj = 0.46), and the Gini-Simpson index (R2adj = 0.43). Furthermore, resampling methods were found to produce lower accuracy in estimating species diversity compared to UAV data acquired on-site. These findings emphasize the importance of selecting appropriate spatial resolutions and SD metrics to enhance the accuracy of remote sensing-based biodiversity prediction models. |
|---|---|
| ISSN: | 1574-9541 |