H12B-01 INVITED
Back to the Future -Precipitation Extremes, Climate Variability, Environmental Planning and Adaptation
--"The last major climatic oscillation peak was about 1856, or 74 years ago. Practically all of our important railroad and public highway work has been done since that time. Most of our parks systems driveways, and roads of all type for auto travel, in the various States, have been completed within the past 30 years, namely, beginning at the very lowest point of our climatic swing (1900-1910). There is every reason to believe, therefore, as the next 20 years comes on apace, we will witness considerable damage to work done during the past regime of weather."-- Schuman, 1931 At the beginning of the 21st century, as at the beginning of the 20th century, the fundamental question is whether the nation is more prepared for natural disasters today than it was eight decades ago. Indeed, the question is whether the best science, engineering and policy tools are in place to prepare for and respond to extreme events. Changes in the risk and magnitude of extreme precipitation events rank among the most studied impacts, and indicators (symptoms) of climatic variations. Extreme precipitation translates generally into extreme flooding, landslides, collapse of lifeline infrastructure, and the breakdown of public health services among others. In approaching the problem of quantifying the risk and magnitude of extreme precipitation events, there are two major challenges: 1) it is difficult to characterize "observed" (20th century) conditions due to the lack of long-term observations – i.e., short and incomplete historical records; and 2) it is difficult to characterize "predicted" (21st century) conditions due to the lack of skill of precipitation forecasts at spatial and temporal scales meaningful for impact studies, and the short-duration of climate model simulations themselves. The first challenge translates in estimating the probability of occurrence (rare) and magnitude (very large) of events that may have not happened yet. The second challenge is that of quantifying uncertainty and separating climatic variability and change from model error. Nonstationarity and persistence at multiple scales confound the problem. From an economics perspective, the unprecedented success of environmental control and "conservation" in the 20th century, present another yet challenge in terms of social expectations and human development, including the right to sustainable (high) quality of life. In this presentation, we illustrate these challenges by considering first the estimation of Probable Maximum Precipitation, an engineering design criterion typically used in dam design, and examine how it varies spatially across the continental US according to physiographic region and as a function of climate regime. Second, we explore the spatial and temporal scales that link climate variability to macroscale environmental planning, and the notion of place-based adaptive riskgrade analysis.
H12B-02 INVITED
Comparison of Statistical and Physically Based Modelling of Flash Floods in the Southwestern U.S.
Large runoff events in the arid and semi-arid southwestern U.S. are generally a result of intense, short- duration precipitation events that frequently result in considerable property damage and loss of life. Short- duration flow and volume frequency relationships for these watersheds are derived from precipitation characteristics in the course of performing rainfall-runoff analyses. The accuracy of these models of flash floods is not well understood because few ephemeral channels are gauged and, high-temporal resolution observed hydrographs are rare. In this study, archived data for nine ephemeral channels in the Las Vegas region were collected and digitized so that recurrence statistics could be calculated and hydrograph characteristics could be compared with those generated with rainfall-runoff models. Although the rainfall- runoff models tended to spread the hydrograph volume over a longer time and produce a rising limb with smaller slope than that of the observed hydrograph, estimates of the 5-, 10-, and 50- year peak flow and volume were typically within the 90% confidence interval estimated statistically from archived data.
H12B-03 INVITED
Extreme climate event trends: The data mining and evaluation of the A1FI scenario for 2000-2100
We will discuss the implications and resulting alterations of the hydrologic cycle as Earth climate evolves from 2000-2100. Climate simulations based on the assumptions implicit in the A1F1 scenario for the period 2000- 2100 using CCSM3 are analyzed. In particular, we will assess the changes in the surface latent and sensible heat energy budget, the Indian regional water budgets including trends in the timing and duration of the Indian monsoon and the resulting impacts on mean river flow and hydroelectric power generation potential. These analyses will also be examined within the context of heat index, droughts, floods and related estimates of societal robustness and resiliency. We will interpret these new A1F1 results within the context of the previous climate simulations based on the SRES A2 and B1 scenarios forced with land cover and atmospheric CO2. Analyses of historical records in the context of the Indian Monsoon Rainfall (IMR) have suggested an evolving relation of IMR with natural climate variability caused by El Nino events. We will report on the combined effects of natural climate variability and global warming on IMR and assess the trend of extreme rain and temperature events in a warming environment.
H12B-04 INVITED
Spatial Hierarchical Modeling of Precipitation Extremes from a Regional Climate Model
The goal of this work is to characterize the extreme precipitation simulated by a regional climate model (RCM) over its spatial domain. For this purpose, we develop a Bayesian hierarchical model. The model assumes the parameters of the extreme value distribution are characterized by a latent spatial process. We simultaneously model the data from both a control and future run of the RCM which allows for easy inference about projected change. Additionally, this hierarchical model is the first to spatially model the shape parameter which characterizes the nature of the distribution's tail. Our hierarchical model shows that for the winter season, the RCM indicates a general increase in 100-year precipitation return levels for most of the study region. For the summer season, the RCM surprisingly indicates a significant decrease in the 100-year precipitation return level.
H12B-05 INVITED
Extreme Storm Intensities Versus Extreme Storm Patterns: The Role of Scale and Variability in Aggregate Risk and Extreme Economic Losses
The aggregate economic loss from a storm is a sum of correlated random variables, where each separate
loss depends mostly on the local intensity of the loss-causing factor (e.g. wind, runoff, etc.), while the total
outcome of an event (and particularly events at the tail of aggregate loss distribution) is dominated by the
spatial clustering, area coverage, and overlap between the storm and the exposure locations of interest. This
is particularly important when the area of interest is large (e.g. US, Western Europe, China) and the use of a
loss estimation approach based on a simple correlation model (random field, multivariate distribution) could
far overestimate the aggregate economic loss by neglecting the restricted spatial extent of storms. At the
other end of the scale, the lack of local variability in large-scale models could introduce significant local
biases of extreme loss, particularly in heavily urbanized areas.
The so-called "event driven" approach to
catastrophe modeling is one solution for these problems. In this approach, storm activity recorded during the
last several decades is perturbed to obtain tens of thousands of simulated storm events (and particularly the
extreme ones) under the assumption of stationary climate conditions. Although this particular approach has
improved greatly since the early 1990s, there remains room for improvement in both the theoretical and
practical aspects of aggregate economic risk modeling.
The purpose of this talk is to give examples of the
above problems, to outline the relevant challenges inherent in this kind of modeling, and to present some
partial solutions currently being implemented in contemporary aggregate risk modeling.
H12B-06
The drop-like nature of rainfall: statistical properties of drop-like quantities and connections to the classic fux-like view
A comprehensive understanding of the rainfall phenomenon cannot ignore its drop-like nature and its statistical properties. Using observed properties of the sequences of inter-drop time intervals and drop diameters, 1) we review the results on "classical" flux-like quantities like drought duration, rain duration, and rain intensity, 2) we propose a new non-subjective definition of rain event.
H12B-07
Seasonal and Regional Variations of U.S. Trends in Extreme Precipitation Frequency
Numerous studies have documented increases in U.S. heavy precipitation during the latter part of the 20th Century. Recent studies have also revealed that event frequencies were quite high early in the 20th Century, nearly as high as in the 1980s and 1990s. This suggests that natural variability may be quite large and perhaps the recent increases in the U.S. have a large natural component. The meteorological reasons behind the observed major decadal-scale variations in heavy precipitation have not been investigated. Have there been secular changes in the frequency, intensity, and other characteristics of the meteorological phenomena producing heavy precipitation? Can we attribute these changes to hemispheric or global trends in circulation, SSTs, etc.? Are the recent increases primarily a result of increases in atmospheric water vapor concentrations? Heavy precipitation events occur in a variety of meteorological situations/types that are seasonally and regionally variable. The seasonal and regional differences in trends can provide important insights into possible causes of decadal-scale variations in the frequency and intensity of extreme events. An investigation of such variations has revealed that in those cases where regional monthly trends are statistically significant, the trends are overwhelmingly upward. The great majority of these statistically significant upward monthly trends occur in the warm season (May-October), with the most widespread increases occurring in August. The central part of the U.S. from the Gulf Coast northward into the Great Lakes in particular has experienced statistically significant increases in many warm season months. The timing and locations of the observed increases suggest that a variety of phenomena could be contributing, including tropical cyclones, mesoscale convective systems, extratropical cyclones, and increased water vapor transport from the Gulf of Mexico/western Atlantic. More detailed investigation is required to unravel the causes.
H12B-08
Parameterization of Compartmental Models for Nutrient Transport From River Basins
A key aspect of modeling the transport of solutes in a river basin is how the model handles the retention processes along the distribution of flow paths. One way to imitate the natural delay is to describe the retention thorough an exchange between the main channel and a transient storage zone. The retention characteristics and hence the mathematical description of the exchange, will be crucial for the models ability to describe different hydrological events (e.g. variation in the water flow) correctly. This study explores a methodology to translate parameters from a 1-dimensional advection-dispersion network model with transient storage to a compartmental model, with special emphasis on effects of the river network and parameter variability in the watershed. Temporal moments expressions for both models are used as a theoretical framework for translating transport properties of the more complex model to the compartmental model. The fundamental constructions of the models are compared through derivation of the temporal moments associated with a unit pulse and convolution methods that allows consideration of parameter heterogeneity. Numerical discretisation of the models are done in order to investigate the model error and to study the criteria for resolution scales and the importance of individual parameters, e.g. water flow.