Publications

Zhang, HK; Roy, DP; Kovalskyy, V (2016). Optimal Solar Geometry Definition for Global Long-Term Landsat Time-Series Bidirectional Reflectance Normalization. IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, 54(3), 1410-1418.

Abstract
The systematic generation of global Landsat time-series products has been advocated, and studies have suggested the need to minimize Landsat reflectance anisotropy effects. Considering a global year of non-Antarctic Landsat 5 and 7 acquisitions, the scene center solar zenith varied from 22.14 degrees to 89.71 degrees, with a mean of 43.23 degrees. However, a solar zenith angle definition suitable for generation of global bidirectional reflectance normalized Landsat time series has not been considered. In this paper, a solar zenith optimal definition is developed in terms of the following criteria: i) ensuring that it can be modeled for any location and date; ii) ensuring that it is constant or smoothly changing with respect to space and time and has global annual variation no greater than the variation in the observed Landsat solar zenith angles (theta(s)); and iii) minimizing the differences between theta(s) and the solar zenith angle used for normalization. Global coverage Landsat data are used to evaluate four solar zenith angle definitions, namely, fixed solar zenith angles, i.e., (theta(fixed)), the solar zenith at local solar noon, i.e., (theta(solar noon)), the solar zenith at the global mean Landsat overpass time, i.e., (theta(mean overpass time)), and the solar zenith at a latitude-varying local time parameterized by a polynomial function of latitude, i.e., (theta(poly overpass time)), that effectively provides a model of theta(s). The optimal definition for global and long time-series Landsat bidirectional reflectance normalization is (theta(poly overpass time)). The polynomial coefficients are provided so that users may implement Landsat bidirectional reflectance normalization algorithms using this optimal solar geometry.

DOI:
10.1109/TGRS.2015.2480684

ISSN:
0196-2892