Measurement Error in Remotely Sensed Optical Data

Affected by measurement error, at-sensor radiance has inherent variation over time, space and spectrum, violating the stationary data assumption in regression-based applications in earth observation. These measurement errors may bias the relationships of remotely sensed data with natural resources such as forests. A short time series of Landsat 8 Collection 2 Level 2 data is generated based on forest sample plots, as replicate measurements of the remote sensing system. Using this data, we propose an approach to correcting the biased effect of remotely sensed data for improving the precision of forest inventory.