The use of ensemble-based sensitivity for determining the impact of supplemental observations

Various operational ensemble techniques exist today which estimate a forecast probability distribution. Some of these techniques allow a rapid estimation of how observations, if they were assimilated, would reduce the uncertainty of a chosen forecast aspect. One such method, the ensemble sensitivity method, uses simple linear regression of a forecast response function onto the initial conditions within an ensemble Kalman filter (EnKF) to determine how assimilated observations would reduce the variance of the forecast response function. This straightforward approach allows response functions to be chosen which diagnose significant weather in important locations, such as windstorms or heavy precipitation in populated areas. If the ensemble sensitivity method is employed within the EnKF after routine observations have been assimilated, regions where additional observations would significantly reduce the forecast uncertainty of important weather events can be identified. The nature of these regions of supplemental observations may reveal whether more operational forecasting benefit would be gained by increasing routine observations or using adaptive observing platforms.

Forecast variance reduction fields due to hypothetical observations for various significant weather events in the Pacific Northwest United States are shown to demonstrate the ensemble sensitivity method. An operational 90-member EnKF using the Weather Research and Forecasting (WRF) model at the University of Washington is used to calculate the variance reduction fields. Finally, the universal application of the ensemble sensitivity method and its limitations are discussed.