Combining High Spatial Resolution Lidar Data with Aerial Photography to Automate Individual Tree Measurements

Detailed forest inventory information is critically required for many ecological applications as well as for forest management. However, traditional field measurements, which are labor intensive and time consuming, can provide only a very limited amount of information for large forest areas so that extrapolated estimation of forest characteristics tends to have large errors. With the help of new technologies, we have high spatial resolution remotely sensed data. Since every sensor has its own strengths and weaknesses, one sensor cannot provide all the information we want to acquire. While both small foot print LiDAR data and large scale aerial photographs have the ability to extract individual tree level of biophysical characteristics in large forests, the former do not have spectral information and the latter have problems with automatic extraction of height information because of distortion and heterogeneous tree shapes. In this study we combine LiDAR data and aerial photographs to automate individual tree measurements of South Fork Eel forests in Northern California. First, we generate Canopy Height Model (CHM) from Digital Terrain Model (DTM) and Crown Surface Model (CSM) detect individual tree top locations and tree heights with LiDAR data using morphological analysis. Next, we obtain tree top locations from aerial photography with morphological analysis. Finally, we conducted an accuracy assessment for both LiDAR data and aerial photography to evaluate performance of our methods. The morphological analysis detected tree top locations with a User’s and Producer’s accuracy of 85% and 51% for LiDAR and 55% and 43% for aerial photography.