An automated approach to mapping corn from Landsat imagery

Computers and Electronics in Agriculture
By: , and 



Most land cover maps generated from Landsat imagery involve classification of a wide variety of land cover types, whereas some studies may only need spatial information on a single cover type. For example, we required a map of corn in order to estimate exposure to agricultural chemicals for an environmental epidemiology study. Traditional classification techniques, which require the collection and processing of costly ground reference data, were not feasible for our application because of the large number of images to be analyzed. We present a new method that has the potential to automate the classification of corn from Landsat satellite imagery, resulting in a more timely product for applications covering large geographical regions. Our approach uses readily available agricultural areal estimates to enable automation of the classification process resulting in a map identifying land cover as ‘highly likely corn,’ ‘likely corn’ or ‘unlikely corn.’ To demonstrate the feasibility of this approach, we produced a map consisting of the three corn likelihood classes using a Landsat image in south central Nebraska. Overall classification accuracy of the map was 92.2% when compared to ground reference data.

Publication type Article
Publication Subtype Journal Article
Title An automated approach to mapping corn from Landsat imagery
Series title Computers and Electronics in Agriculture
DOI 10.1016/j.compag.2003.09.001
Volume 43
Issue 1
Year Published 2004
Language English
Publisher Elsevier
Contributing office(s) Earth Resources Observation and Science (EROS) Center
Description 12 p.
First page 43
Last page 54
Google Analytic Metrics Metrics page
Additional publication details