Integrating Landsat TM imagery and See5 decision-tree software for identifying croplands: a case study in Shunyi district, Beijing

Authors:
Jinling Zhao;Dongyan Zhang;Dacheng Wang;Wenjiang Huang
Affiliations:
Beijing Research Center for Information Technology in Agriculture, Beijing Academy of Agriculture and Forestry Sciences, Beijing, P.R. China and Institute of Remote Sensing Applications, Chinese A ...;Beijing Research Center for Information Technology in Agriculture, Beijing Academy of Agriculture and Forestry Sciences, Beijing, P.R. China;Beijing Research Center for Information Technology in Agriculture, Beijing Academy of Agriculture and Forestry Sciences, Beijing, P.R. China;Beijing Research Center for Information Technology in Agriculture, Beijing Academy of Agriculture and Forestry Sciences, Beijing, P.R. China
Venue:
WISM'11 Proceedings of the 2011 international conference on Web information systems and mining - Volume Part I
Year:
2011

Citing 2
Cited 0

C4.5: programs for machine learning

C4.5: programs for machine learning
The application of remote sensing techniques to sugarcane (Saccharum spp. hybrid) production: a review of the literature

International Journal of Remote Sensing

Quantified Score

Hi-index	0.00

Visualization

Abstract

As an important natural resource, cropland plays a key role in ensuring food safety. In this study, an integral method combining Landsat TM imagery and See5 decision-tree software was developed to identify croplands by taking Shunyi District, Beijing as the study area. Considering the specific topographic conditions, vegetation types and variable climate environment as well as growth period of the study area, texture variables, band ratios, digital elevation model (DEM) and its derived slope and aspect were added into decision tree classification. Finally, the cropland distribution map of Shunyi District was derived combining See5 decision tree classification software and NLCD mapping tools integrated in the ERDAS environment. An accuracy evaluation shows that the overall accuracy is 88.48% and 91.85% using GPS sample points and statistical data, separately. The result shows that it is feasible to identify croplands using See5 decision-tree classification tool based on the Landsat TM imagery.