Comparison of methods for estimating fractional cover of photosynthetic and non-photosynthetic vegetation in the otindag sandy land using GF-1 wide-field view data

Abstract

Photosynthetic vegetation (PV) and non-photosynthetic vegetation (NPV) are important ground cover types for desertification monitoring and land management. Hyperspectral remote sensing has been proven effective for separating NPV from bare soil, but few studies determined fractional cover of PV (f pv) and NPV (f npv) using multispectral information. The purpose of this study is to evaluate several spectral unmixing approaches for retrieval of f pv and f npv in the Otindag Sandy Land using GF-1 wide-field view (WFV) data. To deal with endmember variability, pixel-invariant (Spectral Mixture Analysis, SMA) and pixel-variable (Multi-Endmember Spectral Mixture Analysis, MESMA, and Automated Monte Carlo Unmixing Analysis, AutoMCU) endmember selection approaches were applied. Observed fractional cover data from 104 field sites were used for comparison. For f pv, all methods show statistically significant correlations with observed data, among which AutoMCU had the highest performance (R2 = 0.49, RMSE = 0.17), followed by MESMA (R2 = 0.48, RMSE = 0.21), and SMA (R2 = 0.47, RMSE = 0.27). For f npv, MESMA had the lowest performance (R2 = 0.11, RMSE = 0.24) because of coupling effects of the NPV and bare soil endmembers, SMA overestimates f npv (R2 = 0.41, RMSE = 0.20), but is significantly correlated with observed data, and AutoMCU provides the most accurate predictions of f npv (R2 = 0.49, RMSE = 0.09). Thus, the AutoMCU approach is proven to be more effective than SMA and MESMA, and GF-1 WFV data are capable of distinguishing NPV from bare soil in the Otindag Sandy Land.