Estimating leaf area and above-ground biomass of forest regeneration areas using a corrected normalized difference vegetation index
|Authors:||Tommy L. Coleman, James H. Miller, Bruce R. Zutter|
|Station:||Southern Research Station|
|Source:||In: Proceedings of the ASPRS/ASCM/RT 92 Convention - Monitoring and Mapping Global Change. Vol. 4. Remote Sensing and Data Acquisition. 3-8 August 1992. Washington, D.C. [Bethesda, MD]: Am. Sot. Photogram. & Rem. Sens.: 214-230.|
AbstractThe objective of this study was to investigate the regression relations between vegetation indices derived from remotely-sensed data of single and mixed forest regeneration plots. Loblolly pine (Pinus taeda L.) seedlings, sweelgum (Liquidambar styraciflua L.) seedlings and broomsedge (Andropogon virginicus L.) grass were arranged in a factorial combination addition series experiment and replicated four limes in a randomized complete block design. The remotely-sensed data were obtained using the Barnes Multiband Modular Radiometer (MMR) and a 35 mm camera between 26 June and 6 July, 1990. The normalized difference vegetation index (NDVI) and the ratio index (RI) were computed using the bidirectional reflectance factors obtained with the MMR sensor. Color 35 mm slides were used to independently estimate percent vegetative cover and percent bare soil exposed to the field of view of the radiometer. These data were used lo generate a corrected normalized difference vegetation index (CNDVI) which alleviated the effect of bare soil exposed lo the radiometer.
The relationships among the vegetation indices (Us) and leaf area (LA) and above-ground biomass (810) in the single and mixed forest regeneration plots were quite variable. The best relationship between LA and BIO achieved among the spectrally-derived vegetation indices was obrained with the CNDVI, which generated prediction equations that accounted for over 82 percent of the variability from the single loblolly and sweetgum plots. The relationships and amount of variability explained among the VIs and LA and 810 decreased in the plots that contained two or more species. These data show the importance of correcting for bare Soit interference IO spectrally-derived VIs.