Articles

Multitemporal Landsat TM data for stand-based management in northern US forest ecosystems

---

Abstract

Multitemporal  Landsat Thematic Mapper (TM) imagery was processed and forest canopy change  information was extracted over a 415 km2 area in north central Minnesota. The study covered two-,  four-, and six-year disturbance monitoring cycles. At the pixel level, the  detection of canopy depletion and increment dynamics over the six-year  interval was 94% accurate. The stand-based accuracy was 714 out of 759 (also  94%). Here change was detected in at least 50% of the total number of pixels  that made up the individual stands. This paper specifically discusses the  information content of multitemporal multispectral satellite data from the perspective  of traditional stand management. Cross-verification using contingency  matrices and geographical information system (GIS) modelling, combined with  intensive field verification, has shown that the majority of the so-called  "change classification errors" were not errors at all, but a  powerful source of substand information that can significantly impact  sustainable resource management. While the digital disturbance monitoring  methodology as developed for this study does not perform its task at the  stand level, stand data on change can be readily extracted with the  additional benefit of explicit substand information being made available to  the resource manager.

How to Cite:

Coppin, P., (1994) “Multitemporal Landsat TM data for stand-based management in northern US forest ecosystems”, Silva Gandavensis 59. doi: https://doi.org/10.21825/sg.v59i0.872