Abstract

Land Use Mapping Using GIS and Remote Sensing Technique
Track: Imagery and Raster
Authors: Ramesh Gautam, Simon Eching

Scientists are facing challenges to accurately map crop varieties on a field basis due to crop rotations, fallowing and multiple croppings on the fields.

In this research, an innovative approach was developed to classify various crops in Stanislaus County using a decision tree as well as Bayesian based classification algorithms. By integrating both techniques, a methodology was developed to differentiate orchards from non-orchards, and then, map other crop varieties in subsequent classification steps. National Agricultural Imagery Program (NAIP) digital aerial photographs (1x1 m resolution in visible and NIR bands) and LANDSAT multispectral satellite images were processed to classify crop types. The crop accuracy assessment shows an overall classification accuracy of 85% for classifying orchards from non-orchards crops. Similarly, Bayesian classification approach yields an overall classification accuracy of 90% to classify summer crops like corn, melons, dry beans, mixed pasture, etc.

Key words: ArcGIS, LCRAS, decision tree classification, maximum likelihood classification