Evaluation of Spectral Vegetation Index Translation Equations for the Development of Long-Term Data Records

Multi-sensor continuity/compatibility of spectral vegetation indices (VIs) is a complicated issue due to differences in both sensor characteristics and product generation algorithms. In this study, we focused on the spectral issue (spectral bandpass differences) and examined various functional forms to approximate the "isoline-based" translation equation of Yoshioka et al. for the normalized difference vegetation index (NDVI). This Yoshioka translation equation was derived based on the physics of atmosphere-vegetation-photon interactions and, thus, is advantageous in accurately inter-relating and translating the NDVI across multiple sensors. The results indicated that a polynomial approximation to the Yoshioka translation equation was an appropriate form, in which the soil-adjusted vegetation index and aerosol optical thickness (AOT) were used as predictor variables. Radiometric variables that would be correlated well with atmospheric contaminations (e.g., AOT) need to be sought for the practical applications of the Yoshioka translation equation.