Flood loss assessment is an important part of urban flood risk management, and the establishment of disaster damage curve is the key of loss assessment. Because of the limitation of data, it is difficult for cities lacking disaster data to fit the damage curve through historical data. Generally, transferring the damage curve among regions is an effective method, but there are problems of data reference and statistical uncertainty. In view of the inland plain cities lacking disaster data, the damage factor of the data reference area can be transferred to the study area by referring to the principle of analogy. Then, based on the optimization principle of the minimum variation coefficient and the maximum beta distribution probability, the citation error and statistical error in the damage factor quotation were reduced, and the relatively accurate damage factor in the study area was obtained, and then the water depth - damage factor curve was established. Thus, a generalized damage curve fitting method for cities lacking disaster data was formed, which took the regional index values as the input parameters and the damage curves as the output results. Finally, taking Zhengzhou City, China as the study area, the loss curves of 10 property types were calculated. Compared with the original data, the average variation coefficient of the optimal scheme set was reduced by 0.1; the probability of the optimal value was increased by 1.39% compared with the average value. In addition, the method test was conducted in Jinan, China, where the data were available, and it was found that the three types of errors were significantly lower than the traditional comparison method of single city as reference object. This study is expected to provide a scientific reference for the establishment of flood damage curve in cities or areas lacking disaster data.
Keywords: Beta distribution; Damage factor; Data mining; Spatial transference; Variation coefficient; Zhengzhou city.
Copyright © 2020 Elsevier B.V. All rights reserved.