View article

There are many methods that can be used to derive estimates of forest parameters using satellite data. The accuracy of these estimates depends on different factors, such as, how well forest areas can be identified. In practice, difficulties in delineating forest areas from other land use classes might lead to biased estimates. The Swedish National Forest Inventory (NFI) has therefore decided to use a post-stratification approach to combine field data and optical satellite data to derive unbiased estimates of forest parameters over large regions. The objective of this study has been to investigate how much the estimation accuracy for different forest parameters can be improved on a county level by combining field data from the NFI and satellite data using post-stratification compared to use of field data only. Landsat ETM+ and Envisat MERIS images have been used for stratification. The results show that the standard errors for estimates of total stem volume, stem volume for pine, stem volume for spruce, and tree biomass were reduced by 10%-30% on a county level (approximately 1 million ha forest land) by using post-stratification based on Landsat ETM+ data compared to use of field data from the NFI. For stem volume of deciduous trees and the amount of dead wood, the standard deviations were reduced by less than 10%. The stratification based on Envisat MERIS was found to produce estimates that were less accurate than the ones obtained using Landsat ETM+ for all tested forest parameters. However, all estimates based on the Envisat MERIS stratification were found to be more accurate than estimates based only on field data from the NFI.