Publications

Zhu, WB; Jia, SF; Lv, AF (2019). A Statistical Analysis of the Remotely Sensed Land Surface Temperature-Vegetation Index Method for the Retrieval of Evaporative Fraction Over Grasslands in the Southern Great Plains. IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING, 12(8), 2889-2896.

Abstract
The triangle method based on the spatial relationship between remotely sensed (RS) land surface temperature and vegetation index (TVX) has been widely used for the estimation of evaporative fraction (EF). Combing moderate resolution imaging spectroradiometer and in situ EF observations, the physical constraint and theoretical accuracy of the TVX method is investigated in this paper through statistical analysis. In theory, the physical constraint of the TVX method is mathematically solvable by using EF measurements as inputs. However, our research suggests that nearly 90% of its solutions are meaningless to retrieve the spatial distribution of EF. Instead, we propose an optimization method to retrieve the spatial distribution of EF over a large heterogeneous area from a limited number of in situ EF observations. Results show that the new method using just one site for calibration has an accuracy comparable with those produced by the traditional TVX method. Its robustness is also demonstrated by its applicability under partially cloudy sky conditions. The accuracy of this optimization method increases with the number of sites used for calibration, but there is an upper limit. Besides, when all sites are used for calibration, the accuracy produced represents the upper limit of the accuracy of the TVX method. Thus, the proposed method also provides a perspective to evaluate the theoretical accuracy of RS-based EF models.

DOI:
10.1109/JSTARS.2019.2917183

ISSN:
1939-1404