Desert vegetation-habitat complexes mapping using Gaofen-1 WFV (wide field of view) time series images in Minqin County, China

Abstract

Feedbacks between vegetation and associated habitats play a crucial role in the arid and semi-arid dryland ecosystem for maintaining the stability and sustainability. Due to the high spatial and temporal heterogeneous sparse vegetation with the associated habitats in dryland system, it is a challenge to map their interactions and feedbacks for environment management and policy decision. This paper attempted to develop an algorithm using endmembers (EMs) fraction series unmixed from Gaofen-1 (GF-1) WFV (wide field of view) finer time series images for mapping desert vegetation-habitat complexes as vegetation function groups with associated habitat. The time series of EMs, including green vegetation (GV), sand land (SL), saline land (SA), dark surface (DA) at 16 m subpixel level, derived from Multiple Endmember Spectral Mixture Analysis (MESMA), were combined to obtain classification knowledge describing the interactions and feedbacks between vegetation and habitat, and organized with decision tree (DT). According to the similarity of the interactions and feedbacks in the desert vegetation-habitat complexes, this paper further identified their potential of assessing the status of ecosystems (i.e., land degradation). The results show that the finer time series of EMs with satisfied spatial resolution can discern the sparse vegetation and the associated habitats with an overall accuracy of 83.91%, and help understanding degradation processes (i.e., sandification and local salinization) in the study area.