Aboveground Biomass (AGB) is a core indicator for assessing the functionality and quality of grassland ecosystems. However, in long-term, large-scale AGB inversion studies of the Hulunbuir Grassland, the precision of interannual inversion models is often compromised due to sparse or missing sample points in certain years, which hinders the accurate evaluation of dynamic changes in the grassland ecosystem. To address this critical issue, this paper proposes an interannual optimization inversion model based on precision-weighted allocation, aiming to improve AGB inversion accuracy and analyze the spatiotemporal variation characteristics of AGB in Hulunbuir Grassland through model results. Firstly, the vegetation index was calculated using Landsat-5 Level-2 data from 2003, 2004, 2009, and 2010 as the data source. Meteorological data were extracted based on remote sensing image timestamps, and a correlation analysis was conducted with field-measured grassland AGB data. The vegetation indices, temperature, and precipitation data with the highest correlation were selected. The Partial Least Squares Regression (PLSR) algorithm was used to build AGB inversion models for three years. Next, the three-year AGB inversion models were evaluated using simple averaging, weighted averaging linear, and precision-weighted allocation methods. The results indicated that the precision-weighted allocation model was the optimal model. A comparison with field-measured AGB data was conducted to evaluate the model’s accuracy. Finally, the optimal model was used for the long-term AGB inversion of Hulunbuir Grassland, and the spatiotemporal variation characteristics of grassland AGB were analyzed. The results showed that: (1) The vegetation index NDPI, temperature, and precipitation data exhibited high correlations with grassland AGB, with correlation coefficients of 0.67, 0.26, and 0.29, respectively. (2) Among the three models, the precision-weighted allocation regression model provided the best fit (R2 = 0.67), and its accuracy was superior to the other two models in the comparison analysis. (3) The spatial distribution of grassland AGB showed an increasing trend from west to east, which remained consistent in most years, particularly in 1996, 2013, and 2018. However, in 1997 and 2007, the trend showed minimal variation or fluctuations, exhibiting different spatial distribution characteristics. Temporally, except for 1997 and 2007, when the AGB was relatively low (generally below 30 kg/30 m2), and 2019, when the AGB was relatively high (around 65 kg/30 m2), the AGB in other years remained relatively stable, with fluctuations within a range of approximately (45±10) kg/30 m2, without significant deviations. In conclusion, the precision-weighted allocation model successfully addresses the issue of sparse or missing sample points in some years during the inversion of grassland AGB. The findings provide important reference data for accurately estimating long-term, large-scale grassland AGB, carbon storage, and related studies in the Hulunbuir Grassland.