Premium
Non‐linear growth of cumulative flood losses with time
Author(s) -
Pisarenko V. F.
Publication year - 1998
Publication title -
hydrological processes
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.222
H-Index - 161
eISSN - 1099-1085
pISSN - 0885-6087
DOI - 10.1002/(sici)1099-1085(19980315)12:3<461::aid-hyp584>3.0.co;2-l
Subject(s) - cumulative distribution function , flood myth , statistics , distribution (mathematics) , limiting , environmental science , mathematics , econometrics , probability density function , mathematical analysis , geography , archaeology , mechanical engineering , engineering
Statistical data over the past 30 years show that the cumulative sum of losses caused by floods S ( t ) has been increasing with time approximately as t 1·3 , i.e. faster than the linear growth expected for a stationary process. (Losses are evaluated by the number of homeless caused by floods, since these data are the most systematically reported.) At the same time, the factors determining flood losses (the rate of floods and single loss distribution) appear to be stationary over the period of observation. An explanation of this paradox is suggested based on a heavy‐tail distribution function of losses, i.e. a distribution function with infinite expectation value. The proposed stochastic model predicts a faster than linear growth of the cumulative losses until some limiting time, which corresponds to the recurrence period of the maximal possible single loss. Similar pseudo‐non‐stationary effects can be observed for other types of catastrophes and hydrological characteristics with heavy‐tail distributions © 1998 John Wiley & Sons, Ltd.