Using Remotely Sensed Imagery to Extend Forest Inventory Field Plot Data to Large Areas

摘要

The United States Department of Agriculture, Forest Service has traditionally produced forest inventory estimates for large areas using forest attribute data obtained from a combination of plots on aerial photographs and measurements of permanent forest inventory field plots. Human interpretation of the aerial photoplots has been used to estimate the area in each of several of forest land strata, while field measurements have been used to obtain tree volume per unit area. The area and volume per unit area estimates have been combined to estimate totals for large forested areas. However, the process of obtaining and interpreting photography is expensive, labour intensive, time consuming, and subjective, and finally only produces sample estimates. This paper reports an investigation of using satellite imagery in conjunction with techniques for relating plot and satellite data to create pixel-based, large area volume maps that are used to obtain large area estimates for mean tree volume per unit area. The effects of bias and imprecision in the estimated plot-satellite data relationships on the bias and imprecision of the large area volume estimates is analysed.