The effect of topographic corrections on trellis crop data
When ground reflectance is measured from a remotely sensed image there can be radiometric distortions due to variations in the underlying topography. The slope and aspect of the ground will alter the perceived reflectance. There have been several ‘topographic correction’ methods developed to remove the radiometric distortion caused by variations in the underlying topography. To a large extent the literature tends to focus on forests in areas with high topographic relief. This study investigates how applicable these methods are for vineyards in moderate relief areas.
Several topographic corrections were applied to high spatial resolution images of a vineyard. Vines were planted in distinct rows, commonly along contours. The corrections were evaluated by looking at variation within classes and separability between classes. Of these assessment criteria the c-correction worked best, but it provided only a small improvement overall. The corrections performed better on the soil class than the vine canopy class.
Work is currently in progress to evaluate the ability of topographic correction methods to improve the statistical correlation between remotely sensed vine canopy reflectance and proximal vine and fruit composition measurements. Initial results from this study will be reported and possible modifications to standard topographic modelling methods discussed.