Digital and visual analysis of thematic mapper imagery for differentiating old growth from younger spruce-fir stands

• Published in: Remote sensing of environment Vol. 48; no. 3; pp. 291 - 301
• Main Authors: Nel, Elizabeth M., Wessman, Carol A., Veblen, Thomas T.
• Format: Journal Article
• Language: English
• Published: New York, NY Elsevier Inc 01.06.1994; Elsevier Science
• Subjects: ABIES; AGE; Animal, plant and microbial ecology; Applied ecology; ARBOLES; ARBRE; Biological and medical sciences; BOSQUES; COLORADO; Conservation, protection and management of environment and wildlife; EDAD; FORESTS; FORET; Fundamental and applied biological sciences. Psychology; OLD GROWTH FORESTS; Parks, reserves, wildlife conservation. Endangered species: population survey and restocking; PICEA; REMOTE SENSING; SATELLITE IMAGERY; TELEDETECCION; TELEDETECTION; TREES; Rocky Mountains; United States; North America; Colorado; America; Vegetation index; Multispectral remote sensing; Forest management; Digital image; Primary forest; Imagery; Plant conservation
• ISSN: 0034-4257; 1879-0704
• DOI: 10.1016/0034-4257(94)90003-5

The process of old growth conservation would be greatly facilitated if remote sensing could be used to locate America's few remaining old growth forests. The use of multispectral remote sensing in the identification of old growth has been largely confined to efforts in the Pacific Northwest. The purpose of this study was to determine which image processing techniques best discriminate between old growth and younger spruce-fir stands in the southern Rocky Mountains. Various image processing techniques were applied to TM imagery of the Marvine Lakes valley in the Colorado White River National Forest. The study focused on two major structural differences between the forest types: 1) multistoried (old growth) vs. single-storied (young growth) canopies and 2) higher frequency of standing dead in old growth. Field measurements of photosynthetically active radiation confirmed the hypothesis that old growth canopies contain more gaps than the young post-fire stands. Texture analysis, color-ratio techniques, and vegetation indices successfully identified old growth. Evaluation of the success of old growth identification within a geographic information system suggested that digital per-pixel classification was inferior to visual interpretation. Spatial and spectral methods must be used in combination if structurally variable old growth forests are to be detected.