Estimating the contribution of historical precipitation changes to increasing flooding damages in the United States
Abstract
Flooding causes billions of dollars in damage each year in the United States, and damages from flooding have been increasing over time. We investigate whether historical precipitation trends have likely contributed to increasing flood damages. We quantify the relationship between precipitation and flood damages using observations of monthly precipitation and flood damages for all states in the continental United States between 1960 and 2017. We model this relationship using fixed-effects regression analyses that account for (i) time-invariant differences in precipitation and flood damages between states, (ii) year-to-year differences common to all states, and (iii) state-specific seasonality in precipitation and flooding. We also account for changes wealth specific to each state. By including these fixed-effects, we isolate the effect of precipitation on flood damages from confounding factors such as exposure and vulnerability. Using the panel regression, we find a significant, log-linear relationship between flood damage and precipitation, with the strength of the relationship varying by geographic region. We use the results of the regression analyses as a framework to understand the impact of historical precipitation changes. Precipitation has likely increased flood damages for multiple states, with some states showing expected damages in 2017 more than 5 times the damages in 1960 due to changes in the upper tail of the precipitation distribution. Additionally, we estimate the cumulative effect of observed precipitation changes on total damages within the study period. We find that precipitation changes have very likely increased total national damages, with a median estimate of 80 billion in flood damages due to increased precipitation over 1960 to 2017 (95% confidence range of 30-100 billion). Comparison of the observed precipitation trends with the Historical and Natural climate model simulations from the Coupled Model Intercomparison Project 5 (CMIP5) reveals that the pattern of observed precipitation trends is significantly more likely under the historical forcing compared to the natural forcing.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2019
- Bibcode:
- 2019AGUFMGC23A..02D
- Keywords:
-
- 1630 Impacts of global change;
- GLOBAL CHANGE;
- 1694 Instruments and techniques;
- GLOBAL CHANGE;
- 1803 Anthropogenic effects;
- HYDROLOGY;
- 4313 Extreme events;
- NATURAL HAZARDS