Modelling Forest Aboveground Biomass Based on GF-3 Dual-Polarized and WorldView-3 Data: A Case Study in Datong National Wetland Park, China
Table 3
Regression models for estimating aboveground biomass (AGB) using observed AGB as the dependent variable and vegetation indexes (VIs) and radar backscatter data as independent variables.
Data source
Model type
Variable
Best model
F-value
R2
RMSE /ha
Relative RMSE (%)
WorldView-3
VIs model
NDVI
AGB = 2.467 × e6.417 × NDVI
7.17
0.76
43.74
30.12
DVI
AGB = 117.39 − 0.14 × DVI
6.87
0.37
—
31.90
RDVI
AGB = 1.06 × e1.04 × RDVI
7.34
0.73
45.68
32.73
RVI
AGB = 32.7 × e0.066 × RVI
5.74
0.43
—
34.80
GF-3
Backscatter coefficient model
HH
AGB = 10.34 × e0.157 × HH
7.47
0.823
36.37
29.14
HV
AGB = 11.469 × e0.173 × HV
7.63
0.836
30.87
26.74
HH & HV
AGB = 0.67 HV + 9.87 HH + 4.01
7.21
0.81
35.13
28.74
—
Combination model
HV & NDVI
AGB = 0.34 HV + 4.67 NDVI + 5.37
7.9
0.861
26.72
19.36
Note. Confidence is at the 0.05 level and F is the joint hypothesis test (≥7.09). Variable. Pixel value of the parameter maps. The RMSE of RVI and DVI in the VIs model is low; thus, it is not shown.