Above Ground Biomass (AGB) is a key indicator of forest ecosystem carbon storage, energy flow changes and biodiversity, and is crucial for climate change research and forest resource management. Fujian Province, as the largest collective forest area in southern China, has abundant forest resources, accurately estimating forest aboveground biomass can lay the foundation for estimating carbon storage and provide decision-making support for achieving the dual carbon goals. Fujian Province is located in a cloudy and rainy subtropical zone with complex terrain and forest types, making it difficult to estimate forest aboveground biomass, estimating forest aboveground biomass using traditional methods is difficult to meet accuracy requirements. In order to improve the accuracy of aboveground forest biomass, this study integrated and comprehensively utilized multi-source remote sensing data such as the latest spaceborne lidar data GEDI, Landsat and Sentinel series satellites. Above all, the GEDI_V27 canopy height product was optimized based on the forest age calculated from Landsat. Then combined with the optimized MGEDI_V27 canopy height product, by establishing an XGBoost biomass inversion model that combined traditional remote sensing features with canopy height, we effectively improved model accuracy,estimated and mapped the aboveground biomass of forests in Fujian Province. The research results showed that: (1) The GEDI canopy height accuracy evaluation result optimized by forest age was R²=0.67, RMSE=2.24m; (2) The recursive feature elimination algorithm was used to optimize the features of the three forest types, and 10 remote sensing features were obtained. Among them, the most important remote sensing features of the three forest types were forest canopy height, and a comparative evaluation was performed on the features including traditional remote sensing features. The results showed that when the canopy height factor was included in the feature construction, the accuracy of the forest AGB regression analysis was significantly improved, confirming that canopy height played a significant role in biomass estimation; (3) The studied forest AGB range in Fujian Province was 0.001--363.331Mg/hm², the overall accuracy evaluation result was R²=0.75, RMSE=17.34 Mg/hm², and the total AGB amount in the province in 2020 was 822 million Mg. The average value was 101.24Mg/hm², reflecting the good ecological quality of Fujian Province. By optimizing the forest canopy height in GEDI and combining it with traditional remote sensing features, the accuracy of forest aboveground biomass modeling can be significantly improved, and it is possible to accurately estimate and monitor forest biomass in Fujian Province. The research results are helpful for the high-precision estimation of aboveground biomass in regional forests, and have certain guiding significance for the assessment of carbon sinks.