GEOSCAN Search Results: Fastlink


TitleLand Cover of the BOREAS Region from AVHRR and LANDSAT Data
AuthorCihlar, J; Beaubien, J; Xiao, Q; Chen, J M; Li, Z
SourceCanadian Journal of Remote Sensing 23, 2, 1997 p. 163-175,
Alt SeriesEarth Sciences Sector, Contribution Series 20042143
PublisherInforma UK Limited
Mediapaper; on-line; digital
File formatpdf
Released2014 07 31
AbstractThe objective of this study was to characterize the distribution of land cover types in the BOREAS Region. Multitemporal AVHRR data were obtained during the entire 1993 growing season and processed to yield seasonal means for 1 km² pixels in channels 1 and 2, the Normalized Difference Vegetation Index (NDVI), and the area under the NDVI curve. Two different AVHRR classification procedures were employed to identify >30 cover types in a 1.44x10 km² area encompassing the BOREAS Region, using an IGBP-compatible hierarchical classification legend. The accuracy of the AVHRR classification was evaluated qualitatively and quantitatively through a comparison with Landsat Thematic Mapper (TM) images. The composition of land cover types within the AVHRR pixels was quantified using classifications of the two BOREAS study areas derived from TM images. It was found that the two AVHRR classifications provide closely similar estimates (within 0.5%-1.4% of the total area) of area proportions for the individual classes, the accuracy depending somewhat on the level within the classification hierarchy. Consistently with a previous study, the absolute AVHRR classification accuracy was fairly low when all pixels were considered, but was high (>80%) when only AVHRR pixels containing mostly one cover type considered. A combination of AVHRR and TM data could be used to quantify the effect of mixed cover type were considered. A combination of AVHRR and TM data could be used to quantify the effect of mixed cover types within a pixel. Within the 1.44x10 km² region, 32% was identified as coniferous forest, 11% as mixed forest, and 10% was water; other classes occupied <5% each. The study indicated that the two AVHRR classification methodologies provided consistent results and that using different methods may be an effective strategy for increasing the consistency and robustness of area estimates. Improvements were identified in data and methods which are needed to optimize area estimates of the individual classes at the regional and national levels.

Date modified: