Abstract: Using Integrated remote sensing technology (high-resolution satellite imagery, multi-spectral and synthetic aperture radar images) and big data analysis technology (social media sentiment analysis, multi-source data fusion, machine learning modeling), a set of spatial and temporal dynamic monitoring and evaluation framework for urban flood disasters in Sichuan Basin was constructed, and specific cases such as Chengdu flood disaster in 2018 were taken as research objects to verify the feasibility of the technology. The results show that: (1) The application of remote sensing technology has realized the accurate identification of flood boundary, hydrological parameter inversion and inundation depth estimation. (2) The introduction of big data analysis technology, on the one hand, realizes real-time capture of social media data. Combining natural language processing and sentiment analysis, the disaster information and the psychological state of the affected people are dynamically tracked, and the rescue strategy is optimzed. On the other hand, by integrating multi-source data such as meteorology, transportation and remote sensing, a spatio-temporal dynamic monitoring and early warning model of disaster information is constructed, which improves the accuracy of disaster prediction by 15%. (3) The Mean Absolute Error(MAE) of the disaster loss prediction model based on machine learning is as low as 0.08, and the semantic segmentation accuracy of automatic information extraction technology is 95%, which realizes the rapid assessment and response of disaster. (4) The interactive visualization platform integrates multi-source data and promotes multi-party collaborative rescue. The verification case shows that the direct economic loss is reduced by 30% and the recovery time is shortened by 50%.