Regional Forest Canopy Height Estimation Using Multi-source Remote Sensing Data

In this paper, a method which based on multi-source remote sensing data was designed to estimate forest canopy height. An improved topographic correction model which was used for calculating based-GLAS canopy height was put forward with airborne LiDAR data, since surface topography directly impacts vegetation LiDAR waveforms and vegetation height retrieval. Then, with corrected GLAS footprints as input samples, forest canopy height in different ecological zones was estimated by support vector regression (SVR) model combining with MODIS imagery. Lastly, the accuracy of estimation results was verified using field survey data and airborne LiDAR canopy height. The result show: slope levels in the study area directly impacted based-GLAS canopy height estimation accuracy. The improved topographic correction model could relieve the influence of topography on based-GLAS canopy height estimation, and the RMSE of estimation result ranged from 3.25m to 3.48m. In different ecological zones, SVR model estimation accuracy was stable and the RMSE ranged from 6.41m to 7.56m. Compared with arithmetic mean height, Lorey's height was closest to the estimation result and the average estimated accuracy of 80.3% in different ecological zones. The average estimated accuracy verified with airborne LiDAR metric height was 79.5% and consistence with Lorey's height.