SATELLITE DATA TIME SERIES FOR FORECASTING, HABIAT MODELLING AND VISUALISATION OF THE MANAGED BOREAL FOREST LANDSCAPE

摘要

Satellite data of Landsat and SPOT type are operationally used in Sweden for nationwide forest mapping; for detection of clear felled areas and for checking the need for deciduous shrub cleaning in young forest plantations. The data are also used by the Sami people for mapping areas of interest for reindeer grazing. The Swedish forest agency has since 1999 acquired a yearly and nationwide set of images. These images will, together with some older Landsat data collected between 1972 and 1998 and similar future data sets, be made available freely over internet. A similar data policy for Landsat data has recently also been introduced by USGS in USA. In this paper, we give three early examples where we explore the utility of using time series of image data for applications that are related to those that are already operational. In the first example, it is shown that the accuracy for satellite data based forest estimates, trained with national forest inventory field plots, is marginally improved when data from a second time point is added. In the second example, it is shown that young forest plantations can be much better characterised by using data from a series of yearly images than by only using the latest image. In the third example, it is illustrated how a forest data base made from a time series of images, can be used as basis for an economic model that simulates future forest actions, given strictly economic priorities and how such simulations can be used to study the potential impacts on other interests, in this case food resources for reindeer.