The fusion of multiple scale data indicates that the carbon sink function of the Qinghai-Tibet Plateau is substantial

Table 1 Comparison of four NPP models

NPP	MODIS	GLOPEM	CASA	Mean	Random forest	Difference (%)
r	0.38	0.24	0.23	0.28	0.42	+ 49.00
RMSE	258.62	323.99	264.37	282.33	230.60	+ 18.32
PBIAS	− 28.09	− 54.62	− 35.33	− 39.35	− 25.74	+ 34.58
MAPE	88.61	105.08	81.18	91.62	68.41	+ 25.33

r is the correlation coefficient, RMSE is the root mean square error, PBIAS is the percentage of deviation, MAPE is the average absolute percentage error, Mean is the average value of NPP simulated by MODIS, GLOPEM and CASA models, and Difference is the percentage of difference between Random Forest and Average

ISSN: 1750-0680