首页|Examining Accuracy Assessment for an Agricultural Land Use Classification

Examining Accuracy Assessment for an Agricultural Land Use Classification

扫码查看
原文链接
  • NETL
Reporting the classification accuracy for a remote sensing-derived map can be problematic， even when well established methods of assessment are used。 Classification of agricultural crop types represents a relatively straightforward classification problem， in that the classes are distinct (compared to natural vegetation classes)， and the spatial area that they occupy is unambiguous。 Despite this， the way in which ground truth data for model training and validation are used can have a great impact on the accuracy of the information produced， as well as how the accuracy is assessed。 This study examines the impact of model training and validation data on the reporting of classification accuracies for agricultural land use applications。

crop inventorydecision tree classificationaccuracy assessmentagricultureLANDSATSPOTASAR

C. Champagne、H. McNairn、J. Shang、B. Daneshfar

展开 >

Eastern Cereal and Oilseed Research Centre, Agriculture Canada, 960 Carling Avenue, Ottawa, ON K1A-0C6, Canada

National Land and Water Information Service, Agriculture Canada, 960 Carling Avenue, Ottawa, ON K1A-0C6, Canada

International Symposium on Remote Sensing of Environment

San Jose(CR)

Sustainable Development thorough Global Earth Observations

5-8

2007