 This study aims to estimate the saturation values for different vegetation types in landsat imagery and explore approaches to improve forest above ground biomass, AGB, estimation. The results show that pine forest and mixed forest have the highest AGB saturation values, while bamboo forest and shrub have the lowest. Stratification based on either vegetation types or slope aspects provided smaller root mean squared errors than non stratification, with the most accurate estimation being obtained when stratifying both vegetation types and slope aspects. The study highlights the importance of stratification in mitigating the data saturation problem and improving AGB estimation. This article was authored by Panpan Zhao, Deng Xinglu, Guangxing Wang, and others.