H21H-01 08:00h
Hydrological Ensemble Prediction Experiment (HEPEX)
Ensemble forecast techniques are beginning to be used for hydrological prediction by operational hydrological services throughout the world. These techniques are attractive because they allow effects of a wide range of sources of uncertainty on hydrological forecasts to be accounted for. Forecasting should not only offer an estimate of the most probable future state of a system, but also provide an estimate of the range of possible outcomes. Indeed, users are often more concerned with having a quantitative estimate of the probability that catastrophic effects may occur, than with knowing the most probable future state. Not only does ensemble prediction in hydrology offer a general approach to probabilistic prediction; it offers an approach to improve hydrological forecast accuracy as well. The main objective of HEPEX is to bring the international hydrological community together with the meteorological community to demonstrate how to produce reliable hydrological ensemble forecasts that can be used with confidence to make decisions that have important consequences for the economy and for public health and safety. Representatives of operational hydrological services and operational water resources agencies are participating in helping to define and execute the project. This objective can be achieved if the meteorological, hydrological and water resources communities understand the key challenges they face and work together both to couple currently available forecast tools and to improve the current quality of available systems. This poster reports on the status of the project and planned future activities.
H21H-02 08:15h
Navigating a Path Toward Operational, Short-term, Ensemble Based, Probablistic Streamflow Forecasts
The National Weather Service (NWS) has federal responsibility for issuing public flood warnings in the United States. Additionally, the NWS has been engaged in longer range water resources forecasts for many years, particularly in the Western U.S. In the past twenty years, longer range forecasts have increasingly incorporated ensemble techniques. Ensemble techniques are attractive because they allow a great deal of flexibility, both temporally and in content. This technique also provides for the influence of additional forcings (i.e. ENSO), through either pre or post processing techniques. More recently, attention has turned to the use of ensemble techniques in the short-term streamflow forecasting process. While considerably more difficult, the development of reliable short-term probabilistic streamflow forecasts has clear application and value for many NWS customers and partners. During flood episodes, expensive mitigation actions are initialed or withheld and critical reservoir management decisions are made in the absence of uncertainty and risk information. Limited emergency services resources and the optimal use of water resources facilities necessitates the development of a risk-based decision making process. The development of reliable short-term probabilistic streamflow forecasts are an essential ingredient in the decision making process. This paper addresses the utility of short-term ensemble streamflow forecasts and the considerations that must be addressed as techniques and operational capabilities are developed. Verification and validation information are discussed from both a scientific and customer perspective. Education and training related to the interpretation and use of ensemble products are also addressed.
H21H-03 INVITED 08:30h
Uncertainty in Ensemble Streamflow Forecasting
A common solution to the streamflow forecasting problem is to run a hydrologic (or land-surface) model up to the start of the forecast period to estimate basin initial conditions, and then run the model into the future, with an ensemble of forecast inputs, to produce an ensemble of forecasted streamflow. The River Forecast Centers in the United States have used this Ensemble Streamflow Prediction (ESP) method for many years. The skill of the ESP approach depends on : (1) the hydrologic/land-surface model (including model driving data, model parameters, and model structure), (2) methods for updating basin initial conditions at the start of the forecast period; and (3) the local-scale weather/climate forecasts. This presentation summarizes our research on all three aspects of the streamflow-forecasting problem.
H21H-04 INVITED 08:45h
Seasonal to Interannual Hydroclimatic Prediction: From Identification of Dynamics to Multi-Attribute Forecasts
Dynamical and Statistical Models for seasonal to interannual forecasts of key hydroclimatic state variables have been explored in recent years. Many authors report success based on typical performance metrics. Thus, a casual external observer may feel that we are at the verge of a breakthrough in hydrologic prediction, and hence in water resource management. This talk explores this notion, with particular regard to the multi-scale (time and space) nature of hydrologic fluxes, and of the management variables and styles that the water resources community has become accustomed to. A conceptual framework for the nascent predictive science of hydroclimatology is developed and exemplified. Aspects of the dynamics that need to be understood, and a unifying estimation/inference framework are proposed.
H21H-05 09:00h
Overview of Hydrometeorologic Forecasting Procedures at BC Hydro
Energy utility companies must balance production from limited sources with increasing demand from industrial, business, and residential consumers. The utility planning process requires a balanced, efficient, and effective distribution of energy from source to consumer. Therefore utility planners must consider the impact of weather on energy production and consumption. Hydro-electric companies should be particularly tuned to weather because their source of energy is water, and water supply depends on precipitation. BC Hydro operates as the largest hydro-electric company in western Canada, managing over 30 reservoirs within the province of British Columbia, and generating electricity for 1.6 million people. BC Hydro relies on weather forecasts of watershed precipitation and temperature to drive hydrologic reservoir inflow models and of urban temperatures to meet energy demand requirements. Operations and planning specialists in the company rely on current, value-added weather forecasts for extreme high-inflow events, daily reservoir operations planning, and long-term water resource management. Weather plays a dominant role for BC Hydro financial planners in terms of sensitive economic responses. For example, a two percent change in hydropower generation, due in large part to annual precipitation patterns, results in an annual net change of \$50 million in earnings. A five percent change in temperature produces a \$5 million change in yearly earnings. On a daily basis, significant precipitation events or temperature extremes involve potential profit/loss decisions in the tens of thousands of dollars worth of power generation. These factors are in addition to environmental and societal costs that must be considered equally as part of a triple bottom line reporting structure. BC Hydro water resource managers require improved meteorological information from recent advancements in numerical weather prediction. At BC Hydro, methods of providing meteorological forecast data are changing as new downscaling and ensemble techniques evolve to improve environmental information supplied to water managers.
H21H-06 09:15h
Implementing Probabilistic Climate Outlooks Within a Seasonal Hydrologic Forecast System
We describe a method for implementing the Climate Prediction Center's probabilistic climate outlooks within an experimental ensemble hydrologic forecast system. The forecast system has been developed for the portion of the U.S. and the Canadian portion of the Columbia River basin west of the Continental Divide. Forecasts are made at lead times of six months to a year, using the Variable Infiltration Capacity (VIC) macroscale hydrology model implemented at 1/8 degree spatial resolution. Ensemble forecasts are derived from historical resampling, global climate model ensemble forecasts, and most recently the CPC Official Seasonal Outlooks, formulated as probability of exceedence (POE) distributions for temperature and precipitation, using the climate division as the spatial unit. Our current approach for downscaling the CPC POE forecasts involves a resampling step (called the "Schaake Shuffle", as described in a recent paper by M. Clark and others) that is applied to create a 13-member ensemble (of monthly timeseries of precipitation and temperature for each Climate Division) that exhibits plausible spatial and temporal structure. A subsequent resampling step is applied to disaggregate timeseries from a monthly to a daily timestep, and from the climate division to the 1/8 degree hydrologic model grid scale. In this talk, we address two questions: 1) whether the downscaling approach reproduces the observational mean and variability for streamflow, and for spatial temperature and precipitation fields; and 2) whether bias-correction of the climate division POE forecasts is a necessary step in the downscaling process. These questions were investigated through the application of the downscaling approach to the CPC retrospective observational climate division dataset, and via comparisons with a climate division unit dataset aggregated from the authors' PRISM-corrected retrospective climatology.
H21H-07 09:30h
ENSEMBLE METHODS FOR FLOOD FORECASTING DEVELOPED IN THE FLOODRELIEF PROJECT.
To have real value, real-time flood management decisions must be based on an understanding of the uncertainties and associated risks. It is therefore critical for effective flood management tools to provide reliable estimates of the forecast uncertainty. Only by quantifying the inherent uncertainties involved in flood forecasting can effective real-time flood management and warning be carried out. Forecast uncertainty requires the estimation of the uncertainties associated with both the hydrological model inputs (precipitation observations and forecasts), model structure, parameterisation and calibration, and methodologies that predict how the uncertainties from different sources propagate through the hydrological and hydraulic system. Ensemble-based approaches are attractive because they allow effects of a wide range of uncertainties to be incorporated. Within the EU 5th framework project FLOODRELIEF, two complementary ensemble-based approaches are being developed to address the issue of handling and quantifying forecasting and modelling uncertainties. A general stochastic framework based on the Ensemble Kalman Filter (Evensen, 1994) has been developed for flood forecast modelling. The Kalman filter provides a natural framework for determining how the different sources of uncertainty propagate through the hydrological and hydraulic models and to reduce forecast uncertainty via data assimilation of real-time observations. An evaluation of this framework is presented for two case studies, US NWS study catchment, the Blue river basin and the UK FLOODRELIEF study catchment, the Welland and Glen. The results of this evaluation highlight the fact that one of the major outstanding problems in uncertainty estimation is the characterisation of the sources of uncertainty. The second approach is the development of an internet-based decision support system to provide highly accessible real-time flood management tools. This decision support system has been designed together with forecasting end-users to support ensemble forecasting. Ensemble forecasting using different forecasting inputs provides an alternative method of estimating uncertainty. For example rainfall forecasts using meso-scale meteorological forecasts, downscaled meteorological forecasts, radar forecasts, best case and worst case forecasts can be used by operational forecasters to models to estimate an uncertainty range. In this manner a direct and intuitive estimate of forecast uncertainties can be achieved to address the issue of how ensemble results can be communicated to flood managers and decision-makers.
http://projects.dhi.dk/floodrelief/
H21H-08 09:45h
The Forecast Paradigm is Changing - What Role Will the Forecaster Play?
The use of Numerical Weather Prediction (NWP) is now the fundamental starting point for all weather forecasts with valid times beyond 12 hours and in many cases 6 hours. This is a testament to the skill of numerical predictions today. Forecasters are inherently deterministic in there thinking and have been taught to provide specific forecast parameters or a small range in the forecast parameter with little emphasis as to quantifying the inherent uncertainty associated with any forecast. As such one of the main forecaster challenges today is to chose the "right" model for a given weather event. Most weather forecast offices have access to a multitude of numerical model solutions, some times dozens, especially when ensemble members are considered. For a forecaster at a Weather Forecast Office (WFO), it usually means choosing a model that has had the best track record over the most recent series of weather events. Some of the real challenges for the forecaster is maintaining an awareness of model bias since models change on the order of yearly instead of every 2 to 5 years as was true in the past. The forecaster must also be aware that depictions of mesoscale phenomena by high resolution numerical guidance may in fact simulate previously observed local phenomena but in fact be totally in error as the larger scale forcing may be in error. This is becoming a very big challenge for the WFO forecaster in the era of producing digital forecasts (Glahn and Ruth, WRF 2002). Here the forecaster is asked to produce 2.5 to 5 km sensible weather grids over their forecast area of responsibility at temporal frequencies of from 1 hour to 12-hour durations out through 7 days. There is no attempt to denote either the reliability of these forecasts or quantify the uncertainty associated with the forecasts. In fact each forecaster is required to collaborate with his/her neighboring offices to adjust the numbers to reduce incoherency of the particular forecast parameter. This philosophy flies directly in the face of two recent US Weather Research Program position papers on warm and cool season QPF studies, which proposes that probabilistic QPFs be the focus of the next 5 to 10 years and that ensembles should play a key role in this transition. We appear to have reached a plateau in our skill level in such forecasts as QPF that we may see only very minor improvements if the deterministic approach is continued. It must be recognized that if the forecast process transitions to a more probabilistic methodology, that the role of the forecaster, both at national centers and at the local office will change. In the digital forecast era, when statistical post-processing of ensemble output will provide both deterministic and reliable probabilistic forecasts, there is a question as to exactly what role, if any, the forecaster will have in producing value added sensible weather grids beyond forecast valid times of say 12 hours for use by the general public and for input to hydrologic models. This paper will try to provoke some thoughts on these issues.