Using hyperspectral vegetation indices to estimate the fraction of photosynthetically active radiation absorbed by corn canopies

• Published in: International journal of remote sensing Vol. 34; no. 24; pp. 8789 - 8802
• Main Authors: Tan, Changwei, Samanta, Arindam, Jin, Xiuliang, Tong, Lu, Ma, Chang, Guo, Wenshan, Knyazikhin, Yuri, Myneni, Ranga B.
• Format: Journal Article
• Language: English
• Published: Abingdon Taylor & Francis 20.12.2013
• Subjects: Animal, plant and microbial ecology; Applied geophysics; Biological and medical sciences; Biomass; Canopies; Corn; Crops; Earth sciences; Earth, ocean, space; Estimates; Exact sciences and technology; Fundamental and applied biological sciences. Psychology; General aspects. Techniques; Internal geophysics; Regression analysis; Spectra; Teledetection and vegetation maps; Vegetation; Monocotyledones; Variable; Vegetation index; accuracy; vegetation; Modeling; Spectral data; saturation; Field; Relation; Gramineae; Limitation; Angiospermae; performances; biomass; Photosynthetically active radiation; Zea mays; Estimation; Trend; Increase; analysis; in situ; Radiation absorption; Regression model; Hyperspectral characteristic; Parameter; Spermatophyta; primary productivity; Canopy(vegetation)
• ISSN: 0143-1161; 1366-5901
• DOI: 10.1080/01431161.2013.853143

The fraction of photosynthetically active radiation (FPAR) absorbed by vegetation - a key parameter in crop biomass and yields as well as net primary productivity models - is critical to guiding crop management activities. However, accurate and reliable estimation of FPAR is often hindered by a paucity of good field-based spectral data, especially for corn crops. Here, we investigate the relationships between the FPAR of corn (Zea mays L.) canopies and vegetation indices (VIs) derived from concurrent in situ hyperspectral measurements in order to develop accurate FPAR estimates. FPAR is most strongly (positively) correlated to the green normalized difference vegetation index (GNDVI) and the scaled normalized difference vegetation index (NDVI*). Both GNDVI and NDVI* increase with FPAR, but GNDVI values stagnate as FPAR values increase beyond 0.75, as previously reported according to the saturation of VIs - such as NDVI - in high biomass areas, which is a major limitation of FPAR-VI models. However, NDVI* shows a declining trend when FPAR values are greater than 0.75. This peculiar VI-FPAR relationship is used to create a piecewise FPAR regression model - the regressor variable is GNDVI for FPAR values less than 0.75, and NDVI* for FPAR values greater than 0.75. Our analysis of model performance shows that the estimation accuracy is higher, by as much as 14%, compared with FPAR prediction models using a single VI. In conclusion, this study highlights the feasibility of utilizing VIs (GNDVI and NDVI*) derived from ground-based spectral data to estimate corn canopy FPAR, using an FPAR estimation model that overcomes limitations imposed by VI saturation at high FPAR values (i.e. in dense vegetation).