Reliable loss estimation is crucial for flood risk management. As the current standard form of flood loss assessment, it is difficult to fit the Flood Inundated Depth-Loss Rate Function (FILF) due to the lack of historical data in most inland arid and semi-arid plain cities. To address the current trend of increasing flood risk, it has become increasingly important to develop a scientific and reasonable loss assessment function or model for these cities. Therefore, the flood loss rate data of several cities were transferred through amplified characteristic indices to form a loss rate transfer vector of cities lacking disaster data based on the analogy principle. Three-dimensional data processing rules were then set, including the priority sequence of object dimensional variance and the greatest correlation coefficient (CC) of the joint dimension of function and array. Finally, a FILF of cities lacking disaster data was constructed after three-level optimization. The FILF of eight property types was calculated taking Zhengzhou City, China, as the study area. The optimal function and array dimensions were F6 (Biquadratic) and D4-D6, respectively. All CCs exceeded 0.9935, with an average of 0.9971. The joint fitting results also showed that the function dimension was more sensitive to the FILF than the array dimension. The simulated total flood loss of the Jinshui District in 20 years was 2.46 billion yuan, and there was clear spatial disparity in economic loss. This study is expected to resolve the problem of the absence of a loss function in cities or regions lacking data to support urban flood risk management.
Keywords: Beta distribution; Correlation coefficient; Data mining; Space transfer; Three-level optimization.
Copyright © 2021 Elsevier B.V. All rights reserved.