Reduction of the Long-Term Inaccuracy from the AVHRR-Based NDVI Data


—This paper investigated the normalized difference vegetation index (NDVI) stability in the NOAA/NESDIS Global Vegetation Index (GVI) data during 1982-2003, which was collected from five NOAA series satellites. An empirical distribution function (EDF) was developed to eliminate the long-term inaccuracy of the NDVI data derived from the AVHRR sensor on NOAA polar orbiting satellite. The instability of data results from orbit degradation as well as from the circuit drifts over the life of a satellite. Degradation of NDVI over time and shifts of NDVI between the satellites were estimated using the China data set, because it includes a wide variety of different ecosystems represented globally. It was found that the data for the years of 1988, 1992, 1993, 1994, 1995 and 2000 are not stable compared to other years because of satellite orbit drift, AVHRR sensor degradation, and satellite technical problems, including satellite electronic and mechanical satellite systems deterioration. The data for NOAA7(1982, 1983), NOAA-9 (1985, 1986), NOAA-11(1989, 1990), NOAA-14(1996, 1997), and NOAA-16 (2001, 2002) were assumed to be standard because the crossing time of satellite over the equator (between 1330 and 1500 hours) maximized the value of the coefficients. These years were considered as the standard years, while in other years the quality of satellite observations significantly deviated from the standard. The deficiency of data for the affected years were normalized or corrected by using the method of EDF and comparing with the standard years. These normalized values were then utilized to estimate new NDVI time series which show significant improvement of NDVI data for the affected years.

DOI: 10.12720/jcm.9.11.821-828

Extracted Key Phrases

10 Figures and Tables

Cite this paper

@article{Rahman2014ReductionOT, title={Reduction of the Long-Term Inaccuracy from the AVHRR-Based NDVI Data}, author={Md. Z. Rahman and Leonid Roytman and M. Nazrul Islam}, journal={JCM}, year={2014}, volume={9}, pages={821-828} }