Daily mean land surface temperature (LST) acquired from satellite thermal sensors is crucial for various applications such as global and regional climate change analysis. This study proposed a framework to generate global spatiotemporally seamless daily mean LST products (2003–2019). Validations show that the products outperform the traditional method with satisfying accuracy. Our further analysis reveals that the LST-based global land surface warming rate is 0.029 K yr⁻¹ from 2003 to 2019.