©2017. American Geophysical Union. All Rights Reserved. In this study, high-quality soil moisture data derived from the Soil Moisture Active Passive (SMAP) satellite measurements are evaluated from a perspective of improving the estimation of the global gross primary production (GPP) using a process-based ecosystem model, namely, the Boreal Ecosystem Productivity Simulator (BEPS). The SMAP soil moisture data are assimilated into BEPS using an ensemble Kalman filter. The correlation coefficient (r) between simulated GPP from the sunlit leaves and Sun-induced chlorophyll fluorescence (SIF) measured by Global Ozone Monitoring Experiment-2 is used as an indicator to evaluate the performance of the GPP simulation. Areas with SMAP data in low quality (i.e., forests), or with SIF in low magnitude (e.g., deserts), or both are excluded from the analysis. With the assimilated SMAP data, the r value is enhanced for Africa, Asia, and North America by 0.016, 0.013, and 0.013, respectively (p < 0.05). Significant improvement in r appears in single-cropping agricultural land where the irrigation is not considered in the model but well captured by SMAP (e.g., 0.09 in North America, p < 0.05). With the assimilation of SMAP, areas with weak model performances are identified in double or triple cropping cropland (e.g., part of North China Plain) and/or mountainous area (e.g., Spain and Turkey). The correlation coefficient is enhanced by 0.01 in global average for shrub, grass, and cropland. This enhancement is small and insignificant because nonwater-stressed areas are included.