The Use Of Statistical Tree Methods On Rice Field Mapping

Abstract

Java, as the largest populated Island in Indonesia, has experienced tremendous agricultural land use conversion, mostly to Industrial and settlement uses. In order to optimize current rice fields, agricultural Intensification techniques have been introduced including seed technology and remote estimation. For the latter, remotely-sensed data plays an Important role which provide updated information for food security. In general, remotely sensed data provide two basic information i.e. spatial extents of current agricultural fields and estimation of yields. This paper discusses first theme of the role using Landsat multi-spectral data coupled with statistical tree analyses. Two test sites were selected covering different land tenure. Although the rate of classification accuracy was Similar, we found that the decision tree approaches consistently provided acceptable accuracy on around 90%.