H43B-1224
Ensemble Hindcasts for the North Fork of the American River, California
This presentation analyzes two sets of ensemble hindcasts for the North Fork of the American River. The North Fork of the American River drains from the Sierra Nevada Mountains and is an unregulated tributary to Folsom reservoir upstream from Sacramento. Two sets of hindcasts were made. One set uses historical climatological precipitation and temperature forcing. The other uses ensemble predictions from a fixed version of the National Centers for Environmental Predict (NCEP) Global Forecast System (GFS) for the first 14 days of lead time. The results of the climatological forcing hindcasts show the relative importance of intial conditions versus future forcing as a function of time of year and forecast lead time. The GFS forecasts substantially improve the skill of the forecasts during most of the first 30 days during the winter precipitation season. But uncertainty in initial conditions clearly limits the reliability of probabilistic forecasts for the first few days. The results show the importance of having a post-processor to correct for bias and spread errors that may occur in the raw model output.
H43B-1225
A statistical estimation of flood risk using a 29-year river discharge simulation over Japan
A statistical approach that considers the bias and uncertainty of models is proposed for interpreting the simulated river discharge as a flood risk. A 29-year simulation was implemented to estimate parameters of the Gumbel distribution for the probability of extreme discharge, and the estimated discharge probability index (DPI) showed good agreement with observed values. Even more strikingly, high DPI in the simulation corresponded to actual flood damage records. This indicates that the real-time simulation of the DPI could potentially provide flood warnings. This paper also suggests an application using the same statistical method for real-time flood risk prediction that overcomes the lack of sufficiently long simulation data through the use of a pre-existing long-term simulation to estimate statistical parameters. A preliminary flood risk prediction that used operational weather forecast data for 2003 and 2004 gave results similar to those of the 29-year simulation for the Typhoon Tokage (T0423) event on October 20, 2004, demonstrating the transferability of the technique to real-time prediction which is differently biased.
H43B-1226
Uncertainty Analysis for a GIS-based Urban Flood Inundation Model
The assumptions, structure and parameters of a model as well as the data modeled can introduce uncertainties into model output, uncertainties that are not always transferred to the consumers of the model output. The objective of this study is to identify and quantify sources of uncertainty in a model; ultimately the goal is to provide uncertainty information in a form that supports decision-making based on model output. The focus of uncertainty analysis presented here was a GIS-based urban flood inundation model. The model consists of two modules: a simple storm-runoff model and a flat-water based inundation model. Sensitivity analyses were conducted for each uncertainty source. Sources considered include: DEM, rainfall data, status of storm water drainage system, antecedent soil moisture conditions, land cover estimation, and limitations inherent in the model. Based on the sensitivity studies an uncertainty index was derived and relative weights established for each component of the overall uncertainty.
H43B-1227
Toward Improving Streamflow Forecasts Using SNODAS Products
As part of the Water 2025 initiative, researchers at the Desert Research Institute in collaboration with the U.S. Bureau of Reclamation are developing and improving water decision support system (DSS) tools to make seasonal streamflow forecasts for management and operations of water resources in the mountainous western United States. Streamflow forecasts in these areas may have errors that are directly related to uncertainties resulting from the lack of direct high resolution snow water equivalent (SWE) measurements. The purpose of this study is to investigate the possibility of improving the accuracy of streamflow forecasts through the use of Snow Data Assimilation System (SNODAS) products, which are high-resolution daily estimates of snow cover and associated hydrologic variables such as SWE and snowmelt runoff that are available for the coterminous United States. To evaluate the benefit of incorporating the SNODAS product into streamflow forecasts, a variety of Ensemble Streamflow Predictions (ESP) are generated using the Precipitation-Runoff Modeling System (PRMS). A series of manual and automatic calibrations of PRMS to different combinations of measured (streamflow) and estimated (SNODAS SWE) hydrologic variables is performed for several watersheds at various scales of spatial resolution. This study, which is embedded in the constant effort to improve streamflow forecasts and hence water operations DSS, shows the potential of using a product such as SNODAS SWE estimates to decrease parameter uncertainty related to snow variables and enhance forecast skills early in the forecast season.
H43B-1228
The Applicability of Confidence Intervals of Quantiles for the Generalized Logistic Distribution
The generalized logistic (GL) distribution has been widely used for frequency analysis. However, there is a little study related to the confidence intervals that indicate the prediction accuracy of distribution for the GL distribution. In this paper, the estimation of the confidence intervals of quantiles for the GL distribution is presented based on the method of moments (MOM), maximum likelihood (ML), and probability weighted moments (PWM) and the asymptotic variances of each quantile estimator are derived as functions of the sample sizes, return periods, and parameters. Monte Carlo simulation experiments are also performed to verify the applicability of the derived confidence intervals of quantile. As the results, the relative bias (RBIAS) and relative root mean square error (RRMSE) of the confidence intervals generally increase as return period increases and reverse as sample size increases. And PWM for estimating the confidence intervals performs better than the other methods in terms of RRMSE when the data is almost symmetric while ML shows the smallest RBIAS and RRMSE when the data is more skewed and sample size is moderately large. The GL model was applied to fit the distribution of annual maximum rainfall data. The results show that there are little differences in the estimated quantiles between ML and PWM while distinct differences in MOM.