An evaluation of the use of satellite data for monitoring early development of young Sitka spruce plantation forest growth.

Abstract

This paper describes an evaluation of the potential of medium resolution (20–30 m pixel size) satellite data to monitor the growth of Sitka spruce (Picea sitchensis Bong. Carr.) plantations in the UK. The aim is to develop a cost-effective methodology to allow crops to be monitored by remote sensing and, in particular, to provide a mechanism to identify compartments that fail to establish as commercially viable crops. The method used involves predicting forest variables from optical reflectance imagery and field observations using generalized linear modelling. The first stage of this process is to establish the quantitative relationship between forest structure and reflectance data from the Landsat ETM+ and SPOT 4 HRVIR sensors. This was investigated for 25 compartments of Sitka spruce ranging in age from 2 to 17 years. The generalized linear models used showed that mean height is most strongly correlated with reflectance data. There is, however, no relationship between tree density and reflectance data. Interestingly, despite the differences in spatial resolution between Landsat and SPOT data, the predictive models derived were almost identical. Independent field checking of these models indicates that the root mean square error of crop height predictions is 1.5 m. The study suggests that by using repeated satellite surveys, tree growth can be monitored for very low cost per unit area and data can easily be integrated into a GIS (Geographical Information System) allowing foresters to interrogate and visualize the predictions over large areas.