Simultaneous estimation of fractional cover of photosynthetic and non-photosynthetic vegetation using visible-near infrared satellite imagery
Jia Tian,Zhichao Zhang,William D. Philpot,Qingjiu Tian,Wenfeng Zhan,Yanbiao Xi,Xiaoqiong Wang,Cuicui Zhu
DOI: https://doi.org/10.1016/j.rse.2023.113549
IF: 13.5
2023-01-01
Remote Sensing of Environment
Abstract:In vegetation-abundant ecosystems, accurate fractional cover estimation of multiple targets, including nonphotosynthetic vegetation (NPV), photosynthetic vegetation (PV) and bare soil (BS) is an important indicator of carbon storage and climate change, and a contributing factor to soil and water conservation, wildfire, and drought. Satellite remote sensing began with sensors limited to the visible and near infrared (VNIR, 380 nm to 1000 nm), and advancing to sensors spanning the whole solar domain (350 nm to 2500 nm) over the last five decades. However, NPV is still difficult to distinguish from BS in satellite images covering large spatial areas, since their spectral reflectances are always mixed and have subtly different spectral features. Many spectral indices have been proposed to identify NPV, however, most include a short-wave infrared band which is lacking on many current and historical spaceborne sensors. To fully utilize current and historical airborne/satellite images with only VNIR bands, a novel spectral index - the VNIR spectral shape index (SSI) - is proposed in this paper. Using spectra of PV, NPV, and BS acquired from world-recognized spectral libraries, the SSI was devised using only blue, green, red, and NIR bands. When applied to Sentinel-2 MSI images, the SSI was effective in estimating/mapping the fractional cover of PV, NPV, and BS using the triangular space method. The resulting fractional cover was evaluated in three 3 km x 2 km study areas including one grassland, one woodland, and one cropland. The spectral reflectance collected by spectroradiometer in field work and digital image collect by digital camera mounted on unmanned aerial vehicle (UAV) support that the triangular space method is able to estimate the fractional cover of photosynthetic and non-photosynthetic vegetation simultaneously, and that endmember selection relying solely on satellite images is reasonable. Comparing the abundance of each target in UAV images to corresponding MSI pixels, the SSI index and triangular space method proved effective for fractional cover estimation in varied vegetated ecosystems. For fractional cover evaluation in grassland, woodland, and cropland, the R2 reaching to 0.59, 0.62, and 0.70, respectively. These results demonstrate that the SSI is feasible for multiple sensors, enabling potential large temporal and/or spatial scale applications.