Instruments for managing seasonal flood risk using climate forecasts.

摘要

Annual Maximum Flood occurrence has hitherto been treated as a random, independent and identically distributed (iid) process. But there is evidence of both spatial organization, expressed as coherent regions of response to climatic indices such as El Nino Southern Oscillation (ENSO) and the Pacific Decadal Oscillation (PDO) [Pizarro and Lall, 2002]; and temporal organization expressed as clustering of 3-5 years to decades in one regime.; I studied the existence regional scale of flood organization, the potential for season ahead and longer climate based peak flow forecasts and risk management considering decadal/interannual regimes to seasonal management of flood control.; The first part of the thesis is devoted to perform a diagnostic analysis of the spatial and temporal scales of climate-flood risk organization. I find that in the western United States there are coherent regions of peak flows response to climate drivers. The main climate drivers for the region are the El Nino Southern Oscillation (ENSO) and the Pacific Decadal Oscillation (PDO). I also find evidence of nonlinearity in the response of peak flows to these climate drivers.; I then present the development of a univariate model for flood risk estimation conditional on current climate state, capable of issuing season-ahead peak flow forecasts. I find that following the correlation structure seen in the diagnostic analysis, there are more than half of the identified stations in the western US were such forecasts can be issued successfully.; Finally, the thesis explores specific management options in the West using climate driven forecasts. First, I look into operational adjustment of reservoir operating rules using seasonal flood and flow forecasts. I find that using season-ahead peak flow and total seasonal inflow forecasts allows improvements in reservoir operations. Through appropriate modifications in the operation, a higher yield can be obtained while not increasing the risk of water supply shortages, or of flood control failures.; Finally, I look into financial flood risk management incorporating climate information. Here, I find two things. First, the use of climate information in the form of the spatial correlation structure of the stations allows insurance companies to reduce premiums while keeping the same expected profits. They also reduce their exposure to catastrophic losses. The temporal structure of varying risks is also explored in the context of managing catastrophe (cat) bonds. We find that considering varying risk levels has an impact on the spot price of cat bonds, and such impact is relevant for the management of portfolios. (Abstract shortened by UMI.)