Table 1 Comparison of random forest regression evaluations modeled by multiple combinations of characteristic variables.

From: Mapping forest aboveground carbon stock of combined stratified sampling and RFRK model with mean annual temperature and precipitation

Combination of characteristic variables

R2

RMSE (t/hm2)

P (%)

rRMSE (%)

Texture characterization (271)

0.856

4.106

90.2

10.4

Remote sensing element (24)

0.861

4.738

88.3

12.0

Remote sensing + Texture characteristics (48)

0.857

5.686

87.2

14.4

Remote sensing + Texture characteristics (15)

0.871

3.948

90.4

10.0

Back to article page