Fig. 8. Performance comparison of the data assimilation system with the
assumed uncertainties alone, the assumed uncertainties combined with the
introduced inflation factor in EnKF, and the quantified uncertainties.
The NRMSE of the updated states in the growing season was shown,
based on different values of
coefficient of variation (CV) of crop model simulations with the CV of
remote sensing predictions fixed at 0.01 (a), different values of CV of
remote sensing predictions with the CV of crop model simulations fixed
at 0.1 (b), and different values of inflation factor in EnKF with the
CVs of crop model simulations and remote sensing predictions fixed at
0.1 and 0.01, respectively (c). The inflation factor was fixed at 1.0
for panels (a) and (b), assuring the assumed uncertainties functioned
alone. The red lines in each panel indicate the NRMSE based on
the estimated uncertainties of crop model simulations and remote sensing
predictions from the proposed Bayesian methodology in this study.W leaf, leaf weight; N leaf,
leaf N content in the canopy; LAI, leaf area index;W above, aboveground biomass;W grain, grain weight;N above, aboveground N content; andN grain, grain N content.