H51B-01
Recurrence and scaling of extreme events with power-law interarrivals
Extreme value theory is used to predict the recurrence of extreme geologic events and is a key factor in risk mitigation. Stochastic representation of both annual and partial duration series is typically based on the Poisson process. Maxima, minima, and threshold exceedances are represented by a best-fit probability distribution but are assumed to be separated by fixed or exponentially distributed interarrival times. These models are robust because all steady state renewal processes with finite-mean waiting times converge to the Poisson Process as the timescale grows. Limiting distributions governing the maximum value for a properly normalized set of independent and identically distributed random variables are the Type I, II and III extreme value distributions. Inappropriate use of these classical extreme value models for data with power-law interarrivals results in under- estimation of recurrence intervals. However, heavy-tailed interarrivals have been observed in ephemeral streamflows, storm origins, raindrop release and arrival on the ground, and earthquakes. By evaluating max processes in the context of continuous time random walks, we obtain the probability distribution governing extreme events with power-law interarrivals. These limiting distributions are valid over all time scales.
H51B-02
Regional Flood Frequency Estimation at Ungauged Sites in the Balsas River Basin, Mexico
This work presents an adaptation of some regional estimation approaches to southern climates and an application of regional frequency analysis to the Balsas River Basin located in Mexico. Three approaches are used in this study for the delineation of homogeneous regions: the first one is the cluster analysis approach which leads to fixed hydrologic regions, the second one is the canonical correlation analysis approach (Ouarda et al., 2001) which allows the determination of hydrologic neighborhoods that are specific to the site of interest, and the third one is a revised version of the canonical correlation analysis approach that is free of parameters to optimize and which can be automated easily. The two versions of the canonical correlation analysis approach allow also to identify the variables to use during the step of regional estimation. Regional estimation is carried out based on a multiple regression approach. A data set of 29 stations from several Mexican River Basins in and around the Balsas region is used to identify the advantages and weaknesses of each method of delineation of homogeneous regions and to demonstrate the usefulness of these types of regional approaches. Results indicate clearly the advantages of the neighborhood type of approach and the superiority of the two canonical correlation analysis based methods. Results demonstrate also the robustness of these methods through their application to a real world case study with a relatively limited number of stations.
H51B-03
Flood Hazard Map for Masachapa Urban Area and Maravilla River Flood Plain, Nicaragua
A flooding hazard study was realized 2004-2006 through technical cooperation of Japan International Cooperation Agency (JICA) with INETER, upon the request of the Government of Nicaragua. The resulting hazard map shows the inundated area for a 200 years return period flood in a small riverbasin (65 square kilometers) near the Nicaraguan Pacific Coast. This area is affected by heavy rainfall induced by tropical storms in the Caribbean Sea and the Pacific Ocean. The map results from a study performed by a Nicaraguan-Japanese team that developed a flood hazard approach for a small river basin with scarce hydrometeorological data to be replicated to other basins. The team elaborated 25, 50,100 and 200 year return period flood hydrographs to be applied to a bidimensional mathematic model. The simulation results were transformed into a cartographic map using GIS. The map was handed out to Central, Departmental and Municipal authorities for their use.