Evaluation of airborne video data for land-cover classification accuracy assessment in an isolated Brazilian forest

Stuart E. Marsh, James L. Walsh, Claudia Sobrevila

Research output: Contribution to journalArticlepeer-review

21 Scopus citations


This research was designed to evaluate the operational utility of airborne bispectral video data for reconnaissance assessment of land-cover variability and to document the viability of the video data for classification accuracy assessment. There are distinct logistical advantages utilizing airborne video data in isolated and environmentally sensitive regions where there is limited preexisting aerial photography and poor infrastructure and where ground accessibility is difficult and expensive. In a study of land-cover characteristic in Mato Grosso, Brazil, video data provided important insights into the variability and transitions in land-cover that could be used to identify targets for field work and training areas for subsequent satellite image (Landsat TM) classifications. Comparisons of Landsat TM classification accuracy assessments derived from the airborne video and standard color photography revealed that comparable results can be achieved at a high statistical level of significance (0.01). These results demonstrated that the video data can provide information for accuracy assessments equivalent to more standard photographic point-sample data acquisition missions with the added benefit of easily acquiringfar more data. Results of the supervised Landsat TM classification coupled with a topographic model produced an overall accuracy of 68% with a Kappa coefficient of 0.60, based upon a priori classes identified through evaluation of the airborne data and a site visit.

Original languageEnglish (US)
Pages (from-to)61-69
Number of pages9
JournalRemote Sensing of Environment
Issue number1
StatePublished - Apr 1994

ASJC Scopus subject areas

  • Soil Science
  • Geology
  • Computers in Earth Sciences


Dive into the research topics of 'Evaluation of airborne video data for land-cover classification accuracy assessment in an isolated Brazilian forest'. Together they form a unique fingerprint.

Cite this