Relative spectral mixture analysis - A multitemporal index of total vegetation cover

A new multitemporal technique is presented that allows monitoring of vegetation dynamics in coarse multispectral remote sensing data. This technique, relative spectral mixture analysis (RSMA), provides information about the amount of green vegetation (GV), nonphotosynthetic vegetation (NPV) plus litter, and snow relative to a reference time. The RSMA indices of specific ground components are defined so that they are positive when the fractional cover of a ground component is greater than that at the reference time and negative when the fractional cover is less than that at the reference time. The rationale for the new technique and its mathematical underpinnings are discussed. Example RSMA timeseries from the southern-central United States are presented based on four years of MODIS MOD43 nadir BRDF adjusted reflectance (NBAR) data. This timeseries shows that the RSMA GV index, X-GV, is robust in the presence of snow. Spectral simulations show that X-GV is also robust with different soil backgrounds. The RSMA index of NPV/Iitter cover, X-NPV/(litter), provides information about the dynamics of the nonphotosynthetic portion of organic matter at or above the surface. The RSMA index of total vegetation plus litter, X-TV, provides information about the dynamics of the non-soil/non-snow portion of ground cover. Because it Mirrors the bare ground cover, X-TV may be particularly useful in remote sensing applications aimed at the study of soil erosion. (c) 2006 Published by Elsevier Inc.