Discrimination of wheat crop by using co-polarized ratio derived from ASAR data

The paper investigates the potential of the co-polarized HH/VV backscatter ratio at C-band and at high incidence angle to discriminate agricultural canopies characterized by small leaves and vertical stem structure (i.e. cereal crops) from those having a more branching geometry (e.g. tomato) or large leaves (e.g. sugar beet). The analysed data set consists of multi-temporal C-band HH and VV backscatter data acquired in 2006 and 2007 by the Advanced Synthetic Aperture Radar (ASAR) system over an agricultural site located in Southern Italy. The adopted classification scheme is based on a threshold approach, which is firstly assessed on selected fields and then extended over the entire study area. In addition, the analysis assesses the impact on the classification accuracy of two speckle filtering techniques, i.e. spatial and combined temporal-spatial speckle filtering. On test data, results show that the classification accuracy of cereal fields is equal to about 80%. This figure can reach up to 90% if a spatial averaging at field scale is applied.