Nadir and off‐nadir hyperspectral field data: strengths and limitations in estimating grassland biophysical characteristics

In recent years, hyperspectral and multi‐angular approaches for quantifying biophysical characteristics of vegetation have become more widely used. In fact, as both hyperspectral and multi‐angle reflectance decrease the level of noise on retrieved geophysical parameter values, they increase their reliability by also reducing the saturation problem of the relationships between vegetation indices and biophysical characteristics. To test which is the best methodology in estimating some important biophysical grassland parameters (biomass, total and percent biomass nitrogen content, phytomass and its total and percent nitrogen content), nadir and off‐nadir measurements were carried out, three times during the vegetative period of 2004, in a permanent flat meadow located in the experimental farm of the University of Padua, Italy. The two approaches and the broad band vegetation indices calculated using Landsat bands were compared considering both the best determination coefficients of five vegetation indices, calculated with the two analysis, and through a partial least squares regression using different spectral regions measured at different angles as predictive variables. Using nadir data the red edge region was the most useful for the prediction of biophysical variables, especially phytomass, but also nitrogen content. The off‐nadir data did not provide any significance differences in results to that of data obtained in nadir view but both methods seem to be better adapted to describe biophysical parameters of vegetation than the use of broad band vegetation indices