Dynamic NearCasting of Severe Convection using Sequences of GOES Moisture Products - Applicability to NASA Aviation Program? - Ralph Petersen 1 and Robert M Aune 2 1 Cooperative Institute for Meteorological Satellite Studies (CIMSS), University of Wisconsin – Madison, Madison, Wisconsin, 2 NOAA/NESDIS/ORA, Advanced Satellite Products Team, Madison, Wisconsin, Improving the utility of GOES products in operational forecasting

What are we trying to improve? Short-range forecasts of timing and locations of severe thunderstorms - especially hard-to-forecast, isolated summer-time convection What are we trying to correct? - Poor precipitation forecast accuracy in very-short-range NWP - Lack of ANY satellite moisture data in NWP over land in US -Excessive smoothing of mesoscale moisture patterns in NWP data assimilation - Smoothing of upper-level dynamics and dryness -Dominance of parameterizations over observations in increasingly complex short-range NWP - Time delay in getting updated NWP guidance to forecasters How? Provide objective tools that preserves GOES observations and help forecasters identify pre-convective environments 1-6 hours in advance

NearCasting Models Should:  Fill the 1-6 hr “Guidance Gap”  Update/Enhance NWP guidance:  Be Fast and updated very frequently  Use ALL available data - quickly: “Draw closely” to good data  Avoid ‘analysis smoothing’ / ‘super-ob-ing’ issues of longer-range NWP  Help anticipate rapidly developing weather events: “Perishable” guidance products – need rapid delivery  Run Locally? – Few resources needed beyond comms, users easily trained  Improve Situational Awareness Initial Focus on detecting the “Pre-Storm Environment” - Increase Lead Time / Probability of Detection (POD) - Increase Lead Time / Probability of Detection (POD) and Reduce False Alarm Rate (FAR) and Reduce False Alarm Rate (FAR) Goals: Provide objective tools that allow forecasters to expand the usefulness of proven GOES Observations in very-short range predictions hours Fill the Gap Between Nowcasting & NWP

A Specific Objective: Expand the benefits of valuable Moisture Information contained in GOES Sounder Derived Product Images (DPI) GOES hPa PW - 20 July 2005 Improvement of GOES DPI over NWP guess 0700 UTC GOES Sounder products images already are available to forecasters Products currently available include: - Total column Precipitable Water (TPW) - Stability Indices,... (Similar to GII) - 3-layers Precipitable Water (PW)... Build upon GOES’ Strengths + Derived Product Images (DPI) of soundings speeds comprehension of information, and + Data Improve upon Model First Guess,

A Specific Objective: Expand the benefits of valuable Moisture Information contained in GOES Sounder Derived Product Images (DPI) GOES hPa PW - 20 July 2005 Improvement of GOES DPI over NWP guess 1800 UTC GOES Sounder products images already are available to forecasters Products currently available include: - Total column Precipitable Water (TPW) - Stability Indices,... (Similar to GII) - 3-layers Precipitable Water (PW)... Build upon GOES’ Strengths + Derived Product Images (DPI) of soundings speeds comprehension of information, and + Data Improve upon Model First Guess, BUT ( Address some operational realities ) DPIs used primarily as observations - No predictive component - 3-layer DPIs not used in current NWP models - DPIs often obscured when needed most - Correctly anticipated cloud developments obscure subsequent IR observations Need to add a DPI predictive capability to close these information gaps

- Uses Observations Directly - No ‘analysis smoothing’ - Retains observed maxima and minima and extreme gradients - GOES data can be projected forward at full resolution - Spatial resolution adjusts to available data density - Moves mid-/low-level features into areas where clouds form - Provides information even after IR obs are obscured by cirrus  Stability Indices derived from projected Temp/Humidity ‘data’ - New data are added at time observed – not binned / averaged - Data can be included (aged) from multiple observation times Why use a Lagrangian approach for NearCasting? Extending a proven diagnostic approach to prediction - Quick - (10-15 minute time steps) and minimal resources needed - Can be used ‘stand-alone’ or to ‘update’ other NWP guidance DATA DRIVEN – Observation ‘tracked’ into the future Moving Observations to Forecasts

Dry Moist 13 April 2006 – 2100 UTC hPa GOES PW 0 Hour NearCast Lagrangian NearCast How it works: Start with a DPI image,

Dry Moist 13 April 2006 – 2100 UTC hPa GOES PW 0 Hour NearCast Lagrangian NearCast How it works: Start with a DPI image, interpolate wind data to each of the moisture obs,

13 April 2006 – 2100 UTC hPa GOES PW 0 Hour Ob Locations Lagrangian NearCast How it works: Start with a DPI image, interpolate wind data to each of the moisture obs, then move the 10 km data to new locations

13 April 2006 – 2100 UTC hPa GOES PW 1 Hour NearCast Obs Lagrangian NearCast How it works: Start with a DPI image, interpolate wind data to each of the moisture obs, then move the 10 km data to new locations ( use ‘long’ (10 minute) time steps and dynamically change wind fields ),

13 April 2006 – 2100 UTC hPa GOES PW 2 Hour NearCast Obs Lagrangian NearCast How it works: Start with a DPI image, interpolate wind data to each of the moisture obs, then move the 10 km data to new locations ( use ‘long’ ( 10 minute ) time steps and dynamically change wind fields ),

13 April 2006 – 2100 UTC hPa GOES PW 3 Hour NearCast Obs Lagrangian NearCast How it works: Start with a DPI image, interpolate wind data to each of the moisture obs, then move the 10 km data to new locations ( use ‘long’ ( 10 minute ) time steps and dynamically change wind fields ),

13 April 2006 – 2100 UTC hPa GOES PW 3 Hour NearCast Dry Moist 10 km data, 10 minute time steps Lagrangian NearCast How it works: Start with a DPI image, interpolate wind data to each of the moisture obs, then move the 10 km data to new locations ( use ‘long’ ( 10 minute ) time steps and dynamically change wind fields ), and finally. transfer the “projected” moisture ‘obs’ back to a regular grid,

13 April 2006 – 2100 UTC hPa GOES PW 3 Hour NearCast Image Dry Moist 10 km data, 10 minute time steps Lagrangian NearCast How it works: Start with a DPI image, interpolate wind data to each of the moisture obs, then moves the 10 km data to new locations ( use ‘long’ ( 10 minute ) time steps and dynamically change wind fields ), and finally. transfer the “projected” moisture ‘obs’ back to a regular grid, but only to create images of the DPI NearCasts.

Proof of Concept: Identify areas with large vertical moisture gradients RAPID convection development often observed under and along edges of mid-level dry bands as they move over localized low-level moisture maxima Moving GOES data from Observations to Forecasts What are we looking for? Use what GOES observes well !

A Example: Hail storm event of 13 April 2006 that caused record property damage across southern Wisconsin The important questions are both Where and When severe convection will develop, and Where convection will NOT occur - even after cirrus blow-off obscures IR observations? NWP models placed precipitation too far south and west - in area of large-scale moisture convergence 10 cm Hail 24 Hr Precip

13 April 2006 – 2100 UTC hPa GOES PW 0 Hour NearCast 13 April 2006 – 2100 UTC hPa GOES PW 0 Hour NearCast 2115 UTC Sat Imagery NAM 03h Forecast of hPa and hPa PW valid 2100 UTC Initial Conditions 0 Hour Moisture for 2100UTC From 13 April UTC GOES Derived Product Images hPa PW  PW  Initial GOES DPIs Valid 2100 UTC NWP has: Moisture in wrong locations Gradients too weak by factor of 4-8 NWP Analysis Comparison Moving GOES data from Observations to Forecasts Comparison with integrated NWP Moisture fields Hail-Storm Path 

13 April 2006 – 2100 UTC hPa GOES PW 0 Hour NearCast 13 April 2006 – 2100 UTC hPa GOES PW 0 Hour NearCast 2115 UTC Sat Imagery NearCasts 0 Hour Projection for 2100UTC From 13 April UTC GOES Derived Product Images hPa PW  PW  Low-Level Moistening Low-Level Moistening 0 hr Lagrangian NearCasts of GOES DPIs Valid 2100 UTC Low-level Moisture Motion Moving GOES data from Observations to Forecasts Hail-Storm Path 

13 April 2006 – 2100 UTC hPa GOES PW 2 Hour NearCast 13 April 2006 – 2100 UTC hPa GOES PW 2 Hour NearCast 2315 UTC Sat Imagery NearCasts 2 Hour Projection for 2300UTC From 13 April UTC GOES Derived Product Images hPa PW  PW  Low-Level Moistening 2 hr Lagrangian NearCasts of GOES DPIs Valid 2300 UTC Low-level Moisture Motion Moving GOES data from Observations to Forecasts Hail-Storm Path 

13 April 2006 – 2100 UTC hPa GOES PW 4 Hour NearCast 13 April 2006 – 2100 UTC hPa GOES PW 4 Hour NearCast 0115 UTC Sat Imagery NearCasts 4 Hour Projection for 0100UTC From 13 April UTC GOES Derived Product Images hPa PW  PW  Low-Level Moistening 4 hr Lagrangian NearCasts of GOES DPIs Valid 0100 UTC Low-level Moisture Motion Moving GOES data from Observations to Forecasts Hail-Storm Path 

13 April 2006 – 2100 UTC hPa GOES PW 6 Hour NearCast 13 April 2006 – 2100 UTC hPa GOES PW 6 Hour NearCast 0315 UTC Sat Imagery NearCasts 6Hour Projection for 0300UTC From 13 April UTC GOES Derived Product Images hPa PW  PW  Low-Level Moistening 6 hr Lagrangian NearCasts of GOES DPIs Valid 0300 UTC Low-level Moisture Motion Moving GOES data from Observations to Forecasts Hail-Storm Path 

13 April 2006 – 2100 UTC hPa GOES PW 0 Hour NearCast 13 April 2006 – 2100 UTC hPa GOES PW 0 Hour NearCast 2115 UTC Sat Imagery NearCasts 0 Hour Projection for 2100UTC From 13 April UTC GOES Derived Product Images hPa PW  PW  Mid-Level Drying Mid-Level Drying 0 hr Lagrangian NearCasts of GOES DPIs Valid 2100 UTC Mid-level Dryness Motion Moving GOES data from Observations to Forecasts Hail-Storm Path 

13 April 2006 – 2100 UTC hPa GOES PW 2 Hour NearCast 13 April 2006 – 2100 UTC hPa GOES PW 2 Hour NearCast 2315 UTC Sat Imagery NearCasts 2 Hour Projection for 2300UTC From 13 April UTC GOES Derived Product Images hPa PW  PW  Mid-Level Drying 2 hr Lagrangian NearCasts of GOES DPIs Valid 2300 UTC Low-level Dryness Motion Moving GOES data from Observations to Forecasts Hail-Storm Path 

13 April 2006 – 2100 UTC hPa GOES PW 4 Hour NearCast 13 April 2006 – 2100 UTC hPa GOES PW 4 Hour NearCast 0115 UTC Sat Imagery NearCasts 4 Hour Projection for 0100UTC From 13 April UTC GOES Derived Product Images hPa PW  PW  Mid-Level Drying 4 hr Lagrangian NearCasts of GOES DPIs Valid 0100 UTC Low-level Dryness Motion Moving GOES data from Observations to Forecasts Hail-Storm Path 

13 April 2006 – 2100 UTC hPa GOES PW 6 Hour NearCast 13 April 2006 – 2100 UTC hPa GOES PW 6 Hour NearCast 0315 UTC Sat Imagery NearCasts 6Hour Projection for 0300UTC From 13 April UTC GOES Derived Product Images hPa PW  PW  Mid-Level Drying 6 hr Lagrangian NearCasts of GOES DPIs Valid 0300 UTC Low-level Dryness Motion Moving GOES data from Observations to Forecasts Hail-Storm Path 

13 April 2006 – 2100 UTC hPa GOES PW 6 Hour NearCast 13 April 2006 – 2100 UTC hPa GOES PW 6 Hour NearCast 0315 UTC Sat Imagery NearCasts 6Hour Projection for 0300UTC From 13 April UTC GOES Derived Product Images hPa PW  PW  Mid-Level Drying 6 hr Lagrangian NearCasts of GOES DPIs Valid 0300 UTC Rapid moisture changes NearCasted in NE Iowa validated by GPS TPW GOES NearCast PW = = 13 GPS TPW = 16  13 Validation of Local NE Iowa moisture maximum with GPS TPW data     Moving GOES data from Observations to Forecasts Hail-Storm Path  Are these Moisture Changes Real ? VALIDATION with GPS TPW

13 April 2006 – 2100 UTC hPa GOES PW 0 Hour NearCast 13 April 2006 – 2100 UTC hPa GOES PW 0 Hour NearCast Vertical Moisture Gradient ( hPa GOES PW hPa GOES PW) 0 Hour NearCast for 2100UTC From 13 April UTC 2115 UTC Sat Imagery 0 hr Lagrangian NearCasts of GOES DPIs Valid 2100 UTC Differential Moisture Transport Creates Vertical Moisture Gradients Formation of Convective Instability Differential Moisture Transport Creates Convective Instability in hours Vertical Moisture Difference. Observed Instability Moving GOES data from Observations to Forecasts Hail-Storm Path 

13 April 2006 – 2100 UTC hPa GOES PW 2 Hour NearCast 13 April 2006 – 2100 UTC hPa GOES PW 2 Hour NearCast 2315 UTC Sat Imagery Vertical Moisture Gradient ( hPa GOES PW hPa GOES PW) 2 Hour NearCast for 2300UTC From 13 April UTC 2 hr Lagrangian NearCasts of GOES DPIs Valid 2100 UTC Formation of Convective Instability Differential Moisture Transport Creates Vertical Moisture Gradients Vertical Moisture Difference. Moving GOES data from Observations to Forecasts NearCast Instability

13 April 2006 – 2100 UTC hPa GOES PW 4 Hour NearCast 13 April 2006 – 2100 UTC hPa GOES PW 4 Hour NearCast 0115 UTC Sat Imagery Vertical Moisture Gradient ( hPa GOES PW hPa GOES PW) 4 Hour NearCast for 0100UTC From 13 April UTC 4 hr Lagrangian NearCasts of GOES DPIs Valid 2100 UTC Formation of Convective Instability Differential Moisture Transport Creates Vertical Moisture Gradients Vertical Moisture Difference. Moving from Observations to ForecastsMoving GOES data from Observations to Forecasts NearCast Instability

13 April 2006 – 2100 UTC hPa GOES PW 6 Hour NearCast 13 April 2006 – 2100 UTC hPa GOES PW 6 Hour NearCast 0315 UTC Sat Imagery Vertical Moisture Gradient ( hPa GOES PW hPa GOES PW) 6 Hour NearCast for 0300UTC From 13 April UTC 0315 UTC Sat Imagery 6 hr Lagrangian NearCasts of GOES DPIs Valid 2100 UTC Formation of Convective Instability Differential Moisture Transport Creates Vertical Moisture Gradients Vertical Moisture Difference. Moving GOES data from Observations to Forecasts NearCast Instability

13 April 2006 – 2100 UTC hPa GOES PW 6 Hour NearCast 13 April 2006 – 2100 UTC hPa GOES PW 6 Hour NearCast 0315 UTC Sat Imagery 6 hr Lagrangian NearCasts of GOES DPIs Valid 2100 UTC Formation of Convective Instability Differential moisture transports into cloudy regions produces small instability areas Vertical Moisture Difference Information Added by NearCasts. Critical Information about Timing and Location of Mesoscale Destabilzation LOST without NearCast Vertical Moisture Gradient ( hPa GOES PW hPa GOES PW) 6 Hour NearCast for 0300UTC From 13 April UTC Moving GOES data from Observations to Forecasts Mid-Level Drying Low-Level Moistening

13 April 2006 – 2100 UTC hPa GOES PW 0 Hour NearCast 13 April 2006 – 2100 UTC hPa GOES PW 0 Hour NearCast Initial Conditions 0 Hour Moisture for 2100UTC From 13 April UTC GOES Derived Product Images hPa PW  PW  2115 UTC Sat Imagery Initial GOES DPIs Valid 2100 UTC Examples like this show that: - The GOES DPI moisture products provide good data (if available) - The Lagrangian NearCasts produce useful and frequently updates about the FUTURE stability of the atmosphere A number of additional modifications have been made to Make the image products more useful in operations: Added Data Obscuration ‘Flags’ (no-info areas on images) Decreased specificity (detail) with time Increased Temporal Consistency (Incorporated information from 5 previous hourly NearCasts into display image)

Run Time > 18 UTC 19 UTC 20 UTC 21 UTC Valid 18UTC 19UTC 20 UTC 21 UTC 22 UTC 23 UTC 00 UTC 01 UTC 02 UTC 03 UTC Convectively Stable Convectively Unstable  19Z New Data  No Data  20Z New Data  21Z New Data  Derived Vertical Moisture Difference Gray areas added to indicate regions without data in NearCasts from last 6 hrs – “Clouds” Verifying Radar

Areal influence of each predicted trajectory point expands with NearCast length to show less detail and to reflect increased uncertainly at longer forecast range Run Time > 18 UTC 19 UTC 20 UTC 21 UTC Valid 18UTC 19UTC 20 UTC 21 UTC 22 UTC 23 UTC 00 UTC 01 UTC 02 UTC 03 UTC Convectively Stable Convectively Unstable  19Z New Data  No Data  20Z New Data  21Z New Data  Derived Vertical Moisture Difference Verifying Radar

Run Time > 18 UTC 19 UTC 20 UTC 21 UTC Valid 18UTC 19UTC 20 UTC 21 UTC 22 UTC 23 UTC 00 UTC 01 UTC 02 UTC 03 UTC Convectively Stable Convectively Unstable  19Z New Data  No Data  20Z New Data  21Z New Data  Derived Vertical Moisture Difference Verifying Radar Repeated insertion of hourly GOES data improves and revalidates previous NearCasts

13 April 2006 – 2100 UTC hPa GOES PW 0 Hour NearCast 13 April 2006 – 2100 UTC hPa GOES PW 0 Hour NearCast Initial Conditions 0 Hour Moisture for 2100UTC From 13 April UTC GOES Derived Product Images hPa PW  PW  2115 UTC Sat Imagery Initial GOES DPIs Valid 2100 UTC Examples like this show that: - The GOES DPI moisture products provide good data (if available) - The Lagrangian NearCasts produce useful and frequently updates about the FUTURE stability of the atmosphere What next?  24/7 Real-Time runs underway - Output in GRIB-II  Will include more GOES derived hazardous weather forecasting parameters – θ e, LI, Cape,...

Choosing which additional parameters (indices) to use is critical Indices MUST be derived from parameters that GOES observes well!!! For GOES sounder and GOES-R ABI, this means: - ~2-3 layers of Moisture (q) and ~5-6 layers of Temperature (T) Information -  Build upon GII experiences:  Candidate Indices will be derived from T/q NearCasts at each observing layer  Lifted Index – Simple – T&Q at 2 layers only - Compares temperature of parcel lifted form boundary layer to 500 hPa temperature (Modify and Re-scale/Calibrate for GOES obs. layers)  Total Index – Simple – T&Q at 2 layers only – Compares 850 hPa T/Q with 500 hPa T and hPa lapse rate info. (Modify and Re-scale/Calibrate for GOES layers)  Convective Instability – Combines T&Q – Compares Equivalent Potential Temperature (θe - Total Thermal Energy) between low- and mid-tropospheric layers – (In-storm buoyancy realized after convection begins– Need to modify and Re-scale/Calibrate for GOES layers)  Convective Available Potential Energy – Uses low-level T&Q with upper-level T profiles - Compare temperature of parcel lifted form boundary layer through ALL levels to tropopause – (Total buoyancy – Modify and Re-scale/Calibrate for GOES layers)  Convective Inhibition – T only - Compares of low/mid-level Temperature – (Total buoyancy – Modify and Re-scale/Calibrate for GOES layers)  K Index – Combines low/mid level T & Q – (Useful for tropical convection – (Modify and Re- scale/Calibrate for GOES layers)  Multiple Index Ensembles - May require tuning for different areas/conditions

13 April 2006 – 2100 UTC hPa GOES PW 0 Hour NearCast 13 April 2006 – 2100 UTC hPa GOES PW 0 Hour NearCast Initial Conditions 0 Hour Moisture for 2100UTC From 13 April UTC GOES Derived Product Images hPa PW  PW  2115 UTC Sat Imagery Initial GOES DPIs Valid 2100 UTC Examples like this show that: - The GOES DPI moisture products provide good data (if available) - The Lagrangian NearCasts produce useful and frequently updates about the FUTURE stability of the atmosphere What next? - 24/7 Real-Time runs underway - Output in GRIB-II  Will include more GOES derived hazardous weather forecasting parameters – θ e, LI, Cape,...  WFO evaluation:Is this a useful forecasting tool? - If so, when and when not? - What additional tuning/parameters are needed? How should it be tested? - Web-based initially - then directly to AWIPS  Testing scheduled at NWS Central region WFOs, AWC, NSSL,...  Planning tests with METEOSAT data as a GOES-R ABI Risk Reduction

- Data can be inserted (combined) directly retaining resolution and extremes - NearCasts retain useful maxima and minima – image cycling preserves old data Forecast Images agree with storm formations and provide accurate/timely guidance Summary – An Objective Lagrangian NearCasting Model - Quick and minimal resources needed - Can be used ‘stand-alone’ or to ‘update’ other NWP guidance DATA DRIVEN at the MESOSCALE Goals met:  Provides objective tool to increase the length of time that forecasters can make good use of detailed GOES moisture data in their short range forecasts (augment smoother NWP output)  Expands value of GOES moisture products from observations to short-range forecasting tools  Evaluations in NWS offices  Options for METEOSAT tests Stable  Unstable Dry  Moist Differential Moisture Transport Creates Convective Instability Questions ? Can this useful for any of the NASA Aviation efforts?  Extend Convective Initiation Nowcasts?  Help with Oceanic Weather, especially Convection and  Convectively Induced Turbulence? (Could use AIRS/IASI products here!)  Other Areas?