Fig. 1. Vegetation map of the Northeast China and geographic location of study area

Fig. 2. Concept diagram of the causal relationship between meteorological and vegetation dynamics

Fig. 3. Spatial distribution of NDVI changing trend types and the percentage（%） of different types among vegetation types

Fig. 4. Nonlinear test results of NDVI， temperature and precipitation （∆ρ）

Fig. 5. The overall cross-map prediction skill between NDVI and meteorological factors

Fig. 6. Seasonal surrogate test between NDVI and meteorological factors

Fig. 7. Examine cross-map causality between vegetation-meteorological factors in different vegetation covers and analyze the differences in CCM skills （ρ） in different directions

Fig. 8. The best time lags in different vegetation covers

Fig. 9. Convergence cross-mapping skills of NDVI and meteorological factors on grid scale

Fig. 10. Multiview EDM forecast improvement （the two marginal graphics show the column （x） and row （y） averages of the predicted skill improvement， respectively）

Fig. 11. Sensitivity of NDVI to monthly average temperature in different vegetation cover types

Fig. 12. Sensitivity of NDVI to monthly total precipitation in different vegetation cover types

Fig. 13. The overall cross-map prediction skill between temperature and precipitation

Fig. 14. Convergence cross-mapping skills between temperature and precipitation on Grid scale