Crop Inventories with Radar

Publisher Website
Google Scholar
Abstract
Results of an investigation of radar data as employed in an automated crop inventory model are presented. Taking a Monte Carlo approach, radar imagery was simulated and subjected to Bayesian classification techniques to determine those radar parameters most suitable for crop species identification. A radar operating at about 14 GHz with a VV antenna configuration viewing the target in the 45° to 55° angle of incidence range seems most suitable for crop identification purposes. Studies have indicated, however, that the temporal dimension (multi-data imagery) must also be employed for successful crop identification. For example, by revisiting a field four times in 30 days with a dual polarized 14.2 GHz system, it appears feasible to classify agricultural fields with better than 90 percent accuracy. Comparison of results using radars of other configurations are also presented for varying field revisit periods.

This publication has 1 reference indexed in Scilit: