PREDICTING FOREST HEIGHT FROM IKONOS, LANDSAT AND LiDAR IMAGERY

This paper compares and contrasts predictions of forest height in Sitka spruce (Picea sitchensis) plantations based on mediumresolution Landsat ETM+, high-resolution IKONOS satellite imagery and airborne Light Detection And Ranging (LiDAR) data. The relationship between field-measured height and LiDAR height is linear and highly significant (R 0.98) and so LiDAR height measurements were used to improve the height predictions derived from Landsat and IKONOS data. The results showed that despite the difference in spatial resolution and radiometry between Landsat ETM+ and IKONOS data, the strength of the relationship between field height and predicted height using the green spectral band was very similar, with R values of 0.84 and 0.85 respectively. The inclusion of additional observations taken from the LiDAR data improved the strength of the relationship slightly for the Landsat ETM+ data (R = 0.87), but did not change the relationship for the IKONOS data (R = 0.84). Comparison of the height models derived from the satellite and LiDAR data shows that the optical models provide accurate predictions up to the point of forest canopy closure (10 m) in densely stocked plantations (>2000 stems ha), beyond this point only the LiDAR model is able to provide a reliable estimate of forest height. 1 Corresponding author 1. AIMS This study compares forest height predictions from LiDAR, IKONOS and Landsat ETM+ data for managed Sitka spruce stands. Regression analysis is used to evaluate the quality of predictions from each of these sensors against measured tree heights. It is time consuming and expensive variable to obtain measurements of tree height for large areas of forestry. Therefore, we compare two different empirical models to predict height from multi-spectral IKONOS and Landsat ETM+ satellite image data. The first approach uses only tree height measured in the field as the dependent variable; the second approach uses height data derived from LiDAR to complement the field measurements.