Regional changes in landscape pattern and carbon stores in the interior of British Columbia as determined by satellite imagery.

摘要

The detection and mapping of large scale changes to forested landscapes is increasingly important in ecology and management. I used Landsat TM and MSS imagery to map forest cover in 1992 for a 4.2 million ha area of the interior of British Columbia with an overall classification accuracy of 79%. A combination of unsupervised and supervised classification allowed mapping of forested pixels into closed conifer, semi-open conifer, deciduous, and mixed deciduous and coniferous forest classes. The closed conifer class was further subdivided into old ({dollar}>{dollar}140 years), mature (60-140 years), and young ({dollar}<{dollar}60 years) age classes. Unsupervised classification of a combined image containing the Tasseled Cap indices from 1975 and 1992 imagery allowed accurate mapping of disturbances during the period.; I examined changes in landscape pattern from 1975-1992 by calculating indices that describe overall landscape pattern and that of each cover class. The B.C. biogeoclimatic ecosystem classification map was used as an overlay to calculate landscape pattern and changes by biogeoclimatic zone. Analysis showed that 11.4% of the forested area outside provincial parks was disturbed during the 17-year period. Disturbed areas were consistently much smaller than conifer patches in all biogeoclimatic zones and had a lower percentage of interior area. Conifer patch shape complexity varied between zones, but the size and shape of disturbed areas were similar in all zones. Results indicated the early stages of fragmentation of this landscape.; Models of stand level carbon storage and carbon stores in forest products were used in conjunction with British Columbia Ministry of Forests inventory data to calculate a carbon budget for the Engelmann spruce-subalpine fir forests of the study area. Based on inventory data, the net flux of carbon from these forests to the atmosphere averaged 0.14 Mg ha{dollar}sp{lcub}-1{rcub}{dollar} year{dollar}sp{lcub}-1{rcub}{dollar} between 1975 and 1992. Assuming no harvesting during this period, these forests would have been a net carbon sink averaging 0.39 Mg ha{dollar}sp{lcub}-1{rcub}{dollar} year{dollar}sp{lcub}-1{rcub}.{dollar} Calculating the carbon budget using satellite data showed a slight net carbon sink of 0.15 Mg ha{dollar}sp{lcub}-1{rcub}{dollar} year{dollar}sp{lcub}-1{rcub}{dollar} for 1975-1992. The major difference between satellite- and inventory-based carbon budgets was due to differences in the simulation of the detrital carbon pool and timing of harvests. The satellite-based estimate failed to account for recovering areas harvested before 1975, and assumed a constant harvest rate.