Review on applications of remote sensing in urban flood modeling

Urban flooding is serious in China. More than 360 cities have suffered rainstorm floods from 2007 to 2013. Floods in Beijing, Guangzhou, Jinan, et al. have caused serious casualties. In recent years, sea-views in cities have repeated here and there, and rainstorm induced flood has gradually become a chronic illness in many large and medium-sized cities in China. Simulation model for floods is a key technology for urban flood control and land use planning, and play a crucial role in the construction of urban water conservancy projects and infrastructures such as urban drainage system and sponge city. With the continuous and rapid development of the economics and society, urban flood risk warning and management requires a higher demand on the efficiency and elaboration of flood simulation. Obviously, the degree of elaboration on flood simulation highly depends on the temporal and spatial resolution of the basic input data. Rainfall data with smaller time intervals will present more comprehensive rainfall characteristics and further reflect the characteristics of flood processes. Meanwhile, higher resolution terrain data can reflect subtle ground features of urban surface such as buildings, curbs, bridges, etc. By summarizing and analyzing the literature on the researches of hydrological modeling with the utilization of remote sensing data, the application of remote sensing data (including meteorological and hydrological data, and surface information based on remote sensing) in flood simulation is reviewed. Frequently used remote sensing data sources with relatively high precision such as LiDAR (light imaging, detection, and ranging), InSAR (interferometric synthetic aperture radar), oblique photography and hyperspectral remote sensing as well as their applications are presented. Advanced technologies and methods (airborne LiDAR and UAV Oblique Photography) producing basic remote sensing information used in urban flood modeling are presented. Finally, an example is given to illustrate the extraction framework of the basic information of urban flood simulation based on remote sensing, and the prospect to apply remote sensing technology in urban flood simulation is also presented. It is pointed out that flood simulation will be more effective and refined with the support of massive multi-source data in the future. At the same time, rich data sources also bring the following challenges and opportunities: (1) Rapidly increasing amount of high-resolution remote sensing data poses greater challenges to data storage and management, while high-performance distributed storage technologies and data processing technologies can provide solutions for this objective. (2) Massive data and the relatively low processing and utilization capacity will cause a huge waste of resources. Based on data analysis methods such as machine learning and data assimilation, massive multi-source data will be integrated to form a comprehensive database for urban flood modeling, which can provide scientific and technological support for modern urban construction and management. (3) Urban flood simulation with high resolution is extremely computational intensive that the time consumption will exponentially increase. However, this is unacceptable for rapid and efficient urban flood forecasting and early warning. High-performance computing technology such as GPU-based parallelization on CUDA platform will greatly benefit for urban flood simulation.