L2FHM-OMA: Landsat- and Lidar-based Forest Height Map for Oahu Mountainous Areas

The Gap: Height is a critical variable for understanding the functioning of Hawaiian forests in carbon sequestration, biodiversity conservation, hydrological cycling, etc. Lidar is the most accurate technology for mapping vegetation height. However, the existing airborne lidar data over Oahu (collected in 2013) miss the majority of the Koʻolau Range and a part of the Waianae Range, due to persistent cloud and complex topography.

The Solution: We combined island-wide Landsat-8 cloud-free mosaic and 2013 lidar data to map vegetation height over Oahu’s mountainous areas (OMA). We first estimated vegetation height over lidar coverage areas, then used the data over the overlap areas to create a model of predicting lidar-based height from Landsat imagery, and finally predicted height over mountains using the model and the Landsat-8 mosaic.

The Product: We created a 30-m resolution forest height map that covers the mountainous areas in Oahu by subsetting the predicted heights over forests mapped by USGS.

Click Here to Download the Map

The Findings: The mean and standard deviation of OMA forest height are 12.7 ± 5. 1 m with a maximum of ~50 m. Much more analysis is waited to be done with this map. However, an immediate observation we had is that the tall trees mostly occur at lower and medium elevations.

Click the Video Below to See the Highlighted Red Box Area

The Disclaimer: Our error analysis indicated that the predicted heights 1) have a RMSE of 5.8 m and 2) tend to overestimate heights for short trees and underestimate tall trees. The data are provided AS IS. We make no warranty of any kind, express or implied, concerning this information, including but not limited to any warranties of merchantability or fitness for any particular purpose. We assume no responsibility or legal liability concerning the Data’s accuracy, reliability, completeness, timeliness, or usefulness. Please acknowledge HawaiiView and PI Qi Chen if you use the data.

%d bloggers like this: