Toward a 4D Cube of the Atmosphere via Data Assimilation Kelvin Droegemeier University of Oklahoma 13 August 2009
Bringing all the Data Together: Assimilation Old School – Graphically overlay different types of data (the GIS approach) Old School – Graphically overlay different types of data (the GIS approach)
Modern Approach – Assemble a variety of data sets into a single, coherent, dynamically consistent picture – data assimilation Modern Approach – Assemble a variety of data sets into a single, coherent, dynamically consistent picture – data assimilation Bringing all the Data Together: Assimilation
Data Assimilation Data Assimilation System Radars Radial Wind, Reflectivity Other Observations A Bit of Everything Some Places Forecast Model Output All Variables, But From a Forecast 3D Gridded Analysis That Contains all Variables, is Dynamically Consistent, and has Minimum Global Error w/r/t the Observations
Detecting Weather Hazards 3D Gridded Analysis That Contains all Variables, is Dynamically Consistent, and has Minimum Global Error w/r/t the Observations Detection Algorithms Applied to Gridded Fields Features and Relationships
WSR-88D WSR-88D Algorithms Application: Traditional Use of Radar Data for Detecting Weather Hazards TDWR TDWR Algorithms WDSS ITWS
The Problem: Where is the Real Tornado?
Observed Reflectivity Assimilated Reflectivity (ensemble Kalman Filter) Retrieved Temperature R. Fritchie, K. Droegemeier, M. Xue, M. Tong
Observed Reflectivity Assimilated Reflectivity (ensemble Kalman Filter) Retrieved Pressure R. Fritchie, K. Droegemeier, M. Xue, M. Tong
10 Virtual 4D Weather Cube Virtual 4D Weather Cube 4 th dimension time Hazard Observation 0 – 15 mins 15-60mins hrs Aviation weather information in 3 dimensions ( latitude/longitude/height)
Real Time Wind Analysis (400 m grid spacing)
Numerical Prediction 3D Gridded Analysis That Contains all Variables, is Dynamically Consistent, and has Minimum Global Error w/r/t the Observations Detection Algorithms Applied to Gridded Fields Features and Relationships Forecast Models
Prediction: March 2000 Fort Worth Tornado
Tornado Local TV Station Radar
NWS 12-hr Computer Forecast Valid at 6 pm CDT (near tornado time) No Explicit Evidence of Precipitation in North Texas
Reality Was Quite Different!
6 pm 7 pm8 pm Radar Fcst With Radar Data 2 hr 3 hr 4 hr Xue et al. (2003) Fort Worth
Fcst w/o Radar Data 2 hr 3 hr 4 hr Radar 6 pm 7 pm8 pm Fort Worth
Observation-Based Statistical Nowcasting (smart echo extrapolation)
Comparing Model- and Observation- Based/Statistical Nowcasting Approaches Numerical Prediction with Radar Data Assimilation
As a Forecaster Worried About This Reality… 7 pm
As a Forecaster Worried About This Reality… How Much Trust Would You Place in This Model Forecast? 3 hr 7 pm
Actual Radar
Ensemble Member #1 Ensemble Member #2 Ensemble Member #3 Ensemble Member #4 Control Forecast Actual Radar
Probability of Intense Precipitation Model Forecast Radar Observations
Research to Operational Practice: NOAA Hazardous Weather Test Bed n Experimental Forecasts Since 2005 n High Resolution Ensembles n High Resolution Deterministic n Dynamically Adaptive/On Demand
Composite Reflectivity 18 UTC on 24 May 2007 Observed 21 hr, 2 km Grid Forecast Xue et al. (2008)
21 hr, 4 km Grid Spacing Ensemble Forecasts Mean Spread Observed 2 km Grid Xue et al. (2008)
21 hr, 4 km Grid Spacing Ensemble Forecasts Prob Ref > 35 dBZSpaghetti Observed 2 km Grid Xue et al. (2008)
Application to CCFP
Centers of On-Demand Forecast Grids Launched at NCSA During 2007 Spring Experiment Launched automatically in response to hazardous weather messages (tornado watches, mesoscale discussions) Launched based on forecaster guidance Graphic Courtesy Jay Alameda and Al Rossi, NCSA
The Value of Adaptation: Forecaster- Initiated Predictions on 7 June 2007 Brewster et al. (2008) Radar Observations Standard 20-hr Forecast 5 hr LEAD Dynamic Forecast
Real Time Testing Today 1 km grid, 9-hour Forecast