Aboveground forest biomass based on OLSR and an ANN model integrating LiDAR and optical data in a mountainous region of China. Issue 15 (3rd August 2019)

• Record Type: Journal Article
• Authors: Dong, Lixin; Tang, Shihao; Min, Min; Veroustraete, Frank; Cheng, Jie
• Is Part Of: International journal of remote sensing. Volume 40:Issue 15(2019)
• Page Start: 6059
• Page End: 6083
• Publication Date: 2019-08-03
• Subjects: Remote sensing -- Periodicals; Télédétection -- Périodiques; 621.3678
• DOI: 10.1080/01431161.2019.1587201
• Languages: English
• ISSNs: 0143-1161

Abstract

ABSTRACT: Aboveground forest biomass ( B agf ) and height of forest canopy ( H fc ) are of great significance for the determination of carbon sources and sinks, carbon cycling and global change research. In this paper, B agf of coniferous and broadleaf forest in the Chinese Three Gorges region is estimated by integrating light detection and ranging (LiDAR) and Landsat derived data. For a better B agf estimation, a synergetic extrapolation method for regional H fc is explored based on a specific relationship between LiDAR footprint H fc and optical data such as vegetation index (VI), leaf area index (LAI) and forest vegetation cover (FVC). Then, an ordinary least squares regression (OLSR) and a back propagation neural network (BP-NN) model for regional B agf estimation from synergetic LiDAR and optical data are developed and compared. Validation results show that the OLSR can achieve higher accuracy of H fc estimation for all forest types ( R 2 = 0.751, Root mean square error (RMSE) = 5.74 m). The OLSR estimated B agf shows a good agreement with field measurements. The accuracy of regional B agf estimated by the BP-NN model (RMSE = 12.23 t ha –1 ) is superior to that estimated by the OLSR method (RMSE = 17.77 t ha –1 ) especially in areas with complex topography.