Corn Field Identification Accuracy Using Airborne Radar Imagery

Publisher Website
Google Scholar
Abstract
Four channel synthetic aperture radar imagery was interpreted to determine corn field identification accuracies obtainable using single channel, multi-channel and multi-date radar data. Image tone and texture of agricultural fields in the training set were used to derive discrimination criteria which were applied to fields in the testing set. The confusion of other agricultural crops with corn at X-band frequencies and forest cover with corn at L-band frequencies leads to errors of omission and inclusion. Corn field identification accuracies ranged from 58 to 100% using single channel imagery. The use of multi-channel imagery reduces the identification errors and accuracies exceeding 90% were consistently obtained. Individuals lacking radar interpretation experience but familiar with tonal and textural discrimination on airphotos obtained similar results with a short training period. Only two channels (one X-band, one L-band) are needed for the discrimination of corn from all other surface cover types and by late July the corn should be sufficiently taller than all other crops to enable accurate discrimination. Multi-date imagery is thus not necessary to identify corn fields but may aid in the discrimination of the other agricultural cover types found in the study area.

This publication has 3 references indexed in Scilit: