Abstract
ABSTRACTNormalized difference vegetation index (NDVI) has been widely applied for monitoring vegetation dynamics. However, NDVI values are known to be profoundly affected by various external factors. In this study, the variation of NDVI values and trends among the several long-term NDVI datasets with resolution of 1, 4 and 8 km were assessed to understand the differences between the available datasets. The assessment items were 1) Pearson’s correlation coefficient, 2) trend map and breakpoint spatial similarities and 3) comparison of NDVI from Landsat and Flux tower in 2007–2015. The comparison revealed a maximum correlation coefficient of 0.67 among NDVI datasets and average spatial similarity of 37.2% among the trend maps estimated from NDVI datasets. Furthermore, there was a possibility of having significantly opposite trends between two trend maps from different NDVI products. Comparisons with NDVI from vegetation pixel in Landsat 5 TM and 8 OLI resulted in the R2 between 0.06 and 0.68 and RMSE of 0.07–0.2, while comparison with NDVI from flux tower data yielded the RMSE of 0.04–0.41, although the R2 was relatively weak at 0–0.18. Our study highlights the possibility of differences among NDVI datasets, and suggests that these differences should be reconciled especially in time-series analysis.
Highlights
Monitoring and measuring the vegetation over large areas using satellite remote sensing has frequently been performed
Pairwise analysis of normalized difference vegetation index (NDVI) datasets collected by different satellites and prepared for comparison in the 7, 24 and 60 months periods in the spatial resolution categories of 1, 4 and 8 km had Pearson correlation coefficients that were positive but ranged from weak to moderate (Table 6)
The weakest correlation coefficient (0.27) at this spatial resolution was for the comparison of MERIS and AQUA
Summary
Monitoring and measuring the vegetation over large areas using satellite remote sensing has frequently been performed. To correlate values obtained by satellite imaging with actual vegetation characteristics, vegetation indices were developed as spectral transformations that mostly utilize the red and infrared bands, which are sensitive to leaf pigments. NDVI has been viewed as a proxy for vegetation greenness and has been widely used for predicting biophysical properties of vegetation such as the leaf area index (LAI) (Turner, Cohen, Kennedy, Fassnacht, & Briggs, 1999; Wang, Adiku, Tenhunen, & Granier, 2005) and the fraction of photosynthesis active radiation (FAPAR) (Myneni & Williams, 1994) across different vegetation landscapes. NDVI has been successfully applied, the values are known to be sensitive to external factors such as soil and canopy background, atmospheric disturbances, and sun-sensor viewing angle (Baret & Guyot, 1991; Epiphanio & Huete, 1995; Myneni & Williams, 1994; Roujean, Leroy, & Deschamps, 1992). Application of NDVI in areas with dense vegetation results in signal saturation, for LAI values above 2–3 (Carlson & Ripley, 1997), and makes it less likely to provide accurate information of biophysical features vegetation and phenology in this particular area
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
