An on-going challenge in thermal remote sensing is temperature
emissivity separation (TES), where measured radiance can vary either
due to emissivity differences or to a change in land surface temperature
(LST). TES is further complicated by the need to remove atmospheric
effects, and precipitable water vapor is widely recognized as the most
significant error source. Grigsby et al. (2015-in this issue) evaluate five
approaches for TES, including the standard MASTER TES algorithm
using VSWIR-derived and scene-estimated water vapor, a single band
inversion using VSWIR water vapor, and a water vapor scaling method
proposed by JPL. Retrieved LST is validated using field measured values
from grape vines in San Joaquin Valley, California. The authors find significant
improvements in TES in both cases using water vapor scaling,
with the most accurate retrievals using per-pixel AVIRIS-derived
water vapor