Ralph Petersen 1 and Robert M Aune 2
1 Cooperative Institute for Meteorological Satellite Studies (CIMSS), University of Wisconsin – Madison, Madison,
Wisconsin, Ralph.Petersen@NOAA.GOV
2 NOAA/NESDIS/ORA, Advanced Satellite Products Team, Madison, Wisconsin, Robert.Aune@NOAA.GOV

Short-range forecasts of timing and locations of severe thunderstorms
- especially hard-to-forecast, isolated summer-time convection
- Poor precipitation forecast accuracy in short-range NWP
- Lack of satellite moisture data over land in NWP
- Time delay in getting NWP guidance to forecasters
Dominance of parameterizations over observations in increasingly complex short-range NWP
Excessive smoothing of mesoscale moisture patterns in NWP data assimilation
Smoothing of upper-level dynamics and dryness
Lack of objective tools for use by forecasters in detecting pre-convective environments 1-6 hours in the future

Update/Enhance NWP guidance:
Be Fast ( valid 0-6 hrs in advance )
Be run frequently
 Can avoid ‘computational stability’ issues of longer-range NWP
Use all available data quickly:
“Draw closely” to good data
 Avoid ‘analysis smoothing’ issues
0 1 5 6 hours of longer-range NWP
Be used to anticipate rapidly developing weather events:
“ Perishable ” guidance products – need rapid delivery
 Avoid ‘computational resources’ issues of longer-range NWP
Run Locally? – Few resources needed beyond comms, users easily trained
Focus on the “pre-storm environment”
- Increase Lead Time / Probability Of Detection (POD) and
Reduce False Alarm Rate (FAR)
Goals: - Increase the length of time that forecasters can make good use of quality observations which can supplement NWP guidance for their short range forecasts
- Provide objective tools to help them do this

To increase usefulness, GOES Sounder products are available to forecasters as images
Products currently available include:
- Total column Precipitable Water (TPW)
- 3-layers Precipitable Water (PW)
- Stability Indices, . . .
-
DPI Strengths and Current Limitations
+ Image Displays speed comprehension
GOES 900-700 hPa PW - 20 July 2005 of information in GOES soundings , and
TPW NWP Guess Error Reduction using GOES-12
-
+ but
Data Improve upon Model First Guess
- Products used only as observations, and
- Currently have no predictive component
Data not used in current NWP models
,
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
BIAS(mm)
Vertical Layer
SD(mm)
900-700 hPa
700-300 hPa

0700 UTC
Lower-tropospheric GOES Sounder Derived
Product Imagery (DPI)
And , after initial storms have developed, cirrus blow-off can mask lowerlevel moisture maximum in subsequent DPI satellite products
3 layers of Precipitable Water
1000 UTC
Sfc-900 hPa
900-700 hPa
Forecasters need tools that:
700-300 hPa
Preserve highresolution data
1300 UTC and
Show the future state of the atmosphere
1800 UTC
Small cumulus developing in boundary layer later in day can also mask retrievals
GOES 900-700 hPa Precipitable Water - 20 July 2005

What are the benefits of a Lagrangian NearCasting approach?
Quick (10-25 minute time steps) and minimal resources needed
- Can be used ‘stand-alone’ or to ‘update’ other NWP guidance
DATA DRIVEN
Data used ( and combined ) directly - no ‘analysis smoothing’
– Retains observed maxima and minima and extreme gradients
Spatial resolution adjusts to available data density
GOES products can be projected forward at full resolution
- even as they move into areas where clouds form and subsequent IR observations are no longer available
- Data can be tracked ( aged ) through multiple observation times
New data are added at time observed – not binned / averaged
- Use best aspects of all available data sets e.g., Cloud Drift winds can be combined with surface obs cloud heights to create a ‘partly cloudy’ parcel with good height and good motion e.g., Wind Profiler data can be projected forward and include accelerations

13 April 2006 – 2100 UTC
900-700 hPa GOES PW
0 Hour NearCast
Dry
Lagrangian NearCast
How it works:
Moist
Instead of interpolating randomly spaced moisture observations to a fixed grid (and smooth data) and then using gridded wind data to determine changes of to calculating moisture changes at the fixed grid points, the
Lagrangian approach interpolated wind data to each of the moisture obs

13 April 2006 – 2100 UTC
900-700 hPa GOES PW
0 Hour Ob Locations
Lagrangian NearCast
How it works:
Instead of interpolating randomly spaced moisture observations to a fixed grid (and smoothes data) and then using gridded wind data to determine changes of to calculating moisture changes at the fixed grid points, the
Lagrangian approach interpolated wind data to each of the moisture obs and then moves the 10 km data to new locations, using dynamically changing wind forecasts with ‘long’ (10-15 minute)
.
time steps

13 April 2006 – 2100 UTC
900-700 hPa GOES PW
1 Hour NearCast Obs
Lagrangian NearCast
How it works:
Instead of interpolating randomly spaced moisture observations to a fixed grid (and smoothes data) and then using gridded wind data to determine changes of to calculating moisture changes at the fixed grid points, the
Lagrangian approach interpolated wind data to each of the moisture obs and then moves the 10 km data to new locations, using dynamically changing wind forecasts with ‘long’ (10-15 minute)
.
time steps

13 April 2006 – 2100 UTC
900-700 hPa GOES PW
2 Hour NearCast Obs
Lagrangian NearCast
How it works:
Instead of interpolating randomly spaced moisture observations to a fixed grid (and smoothes data) and then using gridded wind data to determine changes of to calculating moisture changes at the fixed grid points, the
Lagrangian approach interpolated wind data to each of the moisture obs and then moves the 10 km data to new locations, using dynamically changing wind forecasts with ‘long’ (10-15 minute)
.
time steps

13 April 2006 – 2100 UTC
900-700 hPa GOES PW
3 Hour NearCast Obs
Lagrangian NearCast
How it works:
Instead of interpolating randomly spaced moisture observations to a fixed grid (and smoothes data) and then using gridded wind data to determine changes of to calculating moisture changes at the fixed grid points, the
Lagrangian approach interpolated wind data to each of the moisture obs and then moves the 10 km data to new locations, using dynamically changing wind forecasts with ‘long’ (10-15 minute)
.
time steps

10 km data, 10 minute time steps
Dry
Lagrangian NearCast
How it works:
13 April 2006 – 2100 UTC
900-700 hPa GOES PW
3 Hour NearCast Image
Moist
Instead of interpolating randomly spaced moisture observations to a fixed grid (and smoothes data) and then using gridded wind data to determine changes of to calculating moisture changes at the fixed grid points, the
Lagrangian approach interpolated wind data to each of the moisture obs and then moves the 10 km data to new locations, using dynamically changing wind forecasts with ‘long’ (10 minute)
.
time steps
The moved moisture ‘obs’ values are then periodically transferred back to an ‘image’ grid, but only for display.

Determining if the atmosphere is conducive to convective storms
STABLE Vertical Temperature Structure shows warm air over colder air
WHY?
Before we go farther,
If parcels are lifted from either the top or bottom of the inversion layer, they both become cooler than surroundings and sink
A Quick Tutorial
-20 -10 0 10 20 30
Temperature (C)

Determining if the atmosphere is conducive to convective storms
But if sufficient moisture is added to bottom of an otherwise stable layer and the entire layer is lifted , it can become Absolutely Unstable.
Latent heat added
Dry
When the entire layer is lifted with moist bottom and top dry , the inversion
Moist bottom cools less than top and it becomes absolutely unstable – producing auto-convection
-20 -10 0 10 20 30
Temperature (C)

Determining if the atmosphere is conducive to convective storms
Lifted layer can become
Absolutely Unstable.
This will cause the layer to overturn and the moist air will raise very rapidly
When the lifted entire layer with moist bottom and top dry is
, the inversion
Dry
Moist
Requires sequence of low-level moistening, mid/upper-level drying and descent (needed to trap lower-level moisture and prevent convection from occurring too quickly) followed by decreased convergence aloft.
bottom cools less than top and it becomes absolutely unstable – producing auto-convection
-20 -10 0 10 20 30
Temperature (C)
So, we really need to monitor not only the increase of low level moisture, but ,more importantly, to identify areas where
Low-level Moistening and Upper-level Drying will occur simultaneously

A case study was conducted for the hail storm event of 13 April 2006 that caused major property damage across southern Wisconsin
The important questions are both where and when severe convection will occur, and
Where convection will NOT occur?
24 Hr Precip
10 cm Hail
NWP models placed precipitation too far south and west - in area of large-scale moisture convergence

A case study was conducted for the hail storm event of 13 April 2006 that caused major property damage across southern Wisconsin
The important questions are both where and when severe convection will occur, and
Where convection will NOT occur?

Initial Conditions
0 Hour Moisture for 2100UTC
From 13 April 2006 2100UTC
GOES Derived Product Images
900-700 hPa PW 
700-300 PW 
2115 UTC Sat Imagery
Initial
GOES DPIs
Valid
2100 UTC
13 April 2006 – 2100 UTC
700-300 hPa GOES PW
0 Hour NearCast
13 April 2006 – 2100 UTC
900-700 hPa GOES PW
0 Hour NearCast

Initial Conditions
0 Hour Moisture for 2100UTC
From 13 April 2006 2100UTC
GOES Derived Product Images
900-700 hPa PW 
700-300 PW 
2115 UTC Sat Imagery
NAM 03h Forecast of 700-300 hPa and 900-700 hPa PW valid 2100 UTC
Initial
GOES DPIs
Valid
2100 UTC
Comparison with NWP conditions
13 April 2006 – 2100 UTC
700-300 hPa GOES PW
0 Hour NearCast
13 April 2006 – 2100 UTC
900-700 hPa GOES PW
0 Hour NearCast

NearCasts
0 Hour Projection for 2100UTC
From 13 April 2006 2100UTC
GOES Derived Product Images
900-700 hPa PW 
700-300 PW 
2115 UTC Sat Imagery
0 hr
Lagrangian
NearCasts of
GOES DPIs
Valid
2100 UTC
Low-Level
Moistening
13 April 2006 – 2100 UTC
700-300 hPa GOES PW
0 Hour NearCast
13 April 2006 – 2100 UTC
900-700 hPa GOES PW
0 Hour NearCast

NearCasts
2 Hour Projection for 2300UTC
From 13 April 2006 2100UTC
GOES Derived Product Images
900-700 hPa PW 
700-300 PW 
13 April 2006 – 2100 UTC
700-300 hPa GOES PW
2 Hour NearCast
2315 UTC Sat Imagery
2 hr
Lagrangian
NearCasts of
GOES DPIs
Valid
2300 UTC
Low-Level
Moistening
13 April 2006 – 2100 UTC
900-700 hPa GOES PW
2 Hour NearCast

NearCasts
4 Hour Projection for 0100UTC
From 13 April 2006 2100UTC
GOES Derived Product Images
900-700 hPa PW 
700-300 PW 
13 April 2006 – 2100 UTC
700-300 hPa GOES PW
4 Hour NearCast
0115 UTC Sat Imagery
4 hr
Lagrangian
NearCasts of
GOES DPIs
Valid
0100 UTC
Low-Level
Moistening
13 April 2006 – 2100 UTC
900-700 hPa GOES PW
4 Hour NearCast

NearCasts
6Hour Projection for 0300UTC
From 13 April 2006 2100UTC
GOES Derived Product Images
900-700 hPa PW 
700-300 PW 
13 April 2006 – 2100 UTC
700-300 hPa GOES PW
6 Hour NearCast
0315 UTC Sat Imagery
6 hr
Lagrangian
NearCasts of
GOES DPIs
Valid
0300 UTC
Low-Level
Moistening
13 April 2006 – 2100 UTC
900-700 hPa GOES PW
6 Hour NearCast

NearCasts
0 Hour Projection for 2100UTC
From 13 April 2006 2100UTC
GOES Derived Product Images
900-700 hPa PW 
700-300 PW 
13 April 2006 – 2100 UTC
700-300 hPa GOES PW
0 Hour NearCast
2115 UTC Sat Imagery
0 hr
Lagrangian
NearCasts of
GOES DPIs
Valid
2100 UTC
Mid-Level
Drying
13 April 2006 – 2100 UTC
900-700 hPa GOES PW
0 Hour NearCast

NearCasts
2 Hour Projection for 2300UTC
From 13 April 2006 2100UTC
GOES Derived Product Images
900-700 hPa PW 
700-300 PW 
13 April 2006 – 2100 UTC
700-300 hPa GOES PW
2 Hour NearCast
2315 UTC Sat Imagery
2 hr
Lagrangian
NearCasts of
GOES DPIs
Valid
2300 UTC
Mid-Level
Drying
13 April 2006 – 2100 UTC
900-700 hPa GOES PW
2 Hour NearCast

NearCasts
4 Hour Projection for 0100UTC
From 13 April 2006 2100UTC
GOES Derived Product Images
900-700 hPa PW 
700-300 PW 
13 April 2006 – 2100 UTC
700-300 hPa GOES PW
4 Hour NearCast
0115 UTC Sat Imagery
4 hr
Lagrangian
NearCasts of
GOES DPIs
Valid
0100 UTC
Mid-Level
Drying
13 April 2006 – 2100 UTC
900-700 hPa GOES PW
4 Hour NearCast

NearCasts
6Hour Projection for 0300UTC
From 13 April 2006 2100UTC
GOES Derived Product Images
900-700 hPa PW 
700-300 PW 
13 April 2006 – 2100 UTC
700-300 hPa GOES PW
6 Hour NearCast
0315 UTC Sat Imagery
6 hr
Lagrangian
NearCasts of
GOES DPIs
Valid
0300 UTC
Mid-Level
Drying
13 April 2006 – 2100 UTC
900-700 hPa GOES PW
6 Hour NearCast

NearCasts
Validation of rapid moisture change in NE Iowa using GPS TPW data
0 Hour Projection for 2100UTC
From 13 April 2006 2100UTC
GOES Derived Product Images
900-700 hPa PW 
700-300 PW 
2115 UTC Sat Imagery
GOES Observed 900-300 PW = 18+6 = 24
GPS TPW = 25
13 April 2006 – 2100 UTC
700-300 hPa GOES PW
0 Hour NearCast
13 April 2006 – 2100 UTC
900-700 hPa GOES PW
0 Hour NearCast
It this real ?
0 hr
Lagrangian
NearCasts of
GOES DPIs
Valid
2100 UTC
Mid-Level
Drying
Validation of
Local NE Iowa moisture maximum with GPS TPW data

NearCasts
Validation of rapid moisture change in NE Iowa using GPS TPW data
2 Hour Projection for 2300UTC
From 13 April 2006 2100UTC
GOES Derived Product Images
900-700 hPa PW 
700-300 PW 
2315 UTC Sat Imagery
GOES Observed 900-300 PW = 20+5 = 25
GPS TPW= 28
13 April 2006 – 2100 UTC
700-300 hPa GOES PW
2 Hour NearCast
13 April 2006 – 2100 UTC
900-700 hPa GOES PW
2 Hour NearCast
It this real ?
2 hr
Lagrangian
NearCasts of
GOES DPIs
Valid
2300 UTC
Mid-Level
Drying
Validation of
Local NE Iowa moisture maximum with GPS TPW data

NearCasts
Validation of rapid moisture change in NE Iowa using GPS TPW data
4 Hour Projection for 0100UTC
From 13 April 2006 2100UTC
GOES Derived Product Images
900-700 hPa PW 
700-300 PW 
0115 UTC Sat Imagery
GOES Observed 900-300 PW = 22+4 = 26
GPS TPW= 30
13 April 2006 – 2100 UTC
700-300 hPa GOES PW
4 Hour NearCast
13 April 2006 – 2100 UTC
900-700 hPa GOES PW
4 Hour NearCast
It this real ?
4 hr
Lagrangian
NearCasts of
GOES DPIs
Valid
0100 UTC
Mid-Level
Drying
Validation of
Local NE Iowa moisture maximum with GPS TPW data

NearCasts
Validation of rapid moisture change in NE Iowa using GPS TPW data
6Hour Projection for 0300UTC
From 13 April 2006 2100UTC
GOES Derived Product Images
900-700 hPa PW 
700-300 PW 
0315 UTC Sat Imagery
GOES NearCast 900-300 PW = 10 + 3 = 13
GPS TPW = 16  13
13 April 2006 – 2100 UTC
700-300 hPa GOES PW
6 Hour NearCast
13 April 2006 – 2100 UTC
900-700 hPa GOES PW
6 Hour NearCast
It this real ?
6 hr
Lagrangian
NearCasts of
GOES DPIs
Valid
0300 UTC
Mid-Level
Drying
Validation of
Local NE Iowa moisture maximum with GPS TPW data

Vertical Moisture Gradient
(900-700 hPa GOES PW -
700-500 hPa GOES PW)
0 Hour NearCast for 2100UTC
From 13 April 2006 2100UTC
2115 UTC Sat Imagery
13 April 2006 – 2100 UTC
700-300 hPa GOES PW
0 Hour NearCast
13 April 2006 – 2100 UTC
900-700 hPa GOES PW
0 Hour NearCast
Formation of
Convective
Instability
0 hr
Lagrangian
NearCasts of
GOES DPIs
Valid
2100 UTC
Differential
Moisture
Transport
Creates
Vertical
Moisture
Gradients

Vertical Moisture Gradient
(900-700 hPa GOES PW -
700-500 hPa GOES PW)
2 Hour NearCast for 2300UTC
From 13 April 2006 2100UTC
13 April 2006 – 2100 UTC
700-300 hPa GOES PW
2 Hour NearCast
2315 UTC Sat Imagery
13 April 2006 – 2100 UTC
900-700 hPa GOES PW
2 Hour NearCast
Formation of
Convective
Instability
2 hr
Lagrangian
NearCasts of
GOES DPIs
Valid
2100 UTC
Differential
Moisture
Transport
Creates
Vertical
Moisture
Gradients

Vertical Moisture Gradient
(900-700 hPa GOES PW -
700-500 hPa GOES PW)
4 Hour NearCast for 0100UTC
From 13 April 2006 2100UTC
13 April 2006 – 2100 UTC
700-300 hPa GOES PW
4 Hour NearCast
0115 UTC Sat Imagery
13 April 2006 – 2100 UTC
900-700 hPa GOES PW
4 Hour NearCast
Formation of
Convective
Instability
4 hr
Lagrangian
NearCasts of
GOES DPIs
Valid
2100 UTC
Differential
Moisture
Transport
Creates
Vertical
Moisture
Gradients

Vertical Moisture Gradient
(900-700 hPa GOES PW -
700-500 hPa GOES PW)
6 Hour NearCast for 0300UTC
From 13 April 2006 2100UTC
13 April 2006 – 2100 UTC
700-300 hPa GOES PW
6 Hour NearCast
13 April 2006 – 2100 UTC
900-700 hPa GOES PW
6 Hour NearCast
Formation of
Convective
Instability
6 hr
Lagrangian
NearCasts of
GOES DPIs
Valid
2100 UTC
Differential
Moisture
Transport
Creates
Vertical
Moisture
Gradients

- The Lagrangian NearCasts produce useful and frequently updates
From 13 April 2006 2100UTC about the FUTURE stability of the atmosphere
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
A number of additional modifications have been made to
700-300 PW 
2115 UTC Sat Imagery the image products more useful in operations:
Initial
GOES DPIs
Valid
• Eliminate displays in areas without NearCast ‘data’
2100 UTC
• “Look” more like satellite observations
• Add uncertainty into longer NearCast images by increasing smoothing through NearCast period
Also:
• 13 April 2006
Want to show the impact of frequent data updates available
700-300 hPa GOES PW
13 April 2006 – 2100 UTC
900-700 hPa GOES PW
0 Hour NearCast

Run Time > 18 UTC 19 UTC 20 UTC 21 UTC
Valid
Derived Vertical Moisture Difference
18UTC
19UTC

19Z
New
Data

Gray areas indicate regions without trajectory data from last 6 hours of NearCasts
20 UTC

20Z
New
Data

21 UTC

21Z
New
Data

22 UTC
Convectively
Stable
No Data
Convectively
Unstable
Verifying Radar
23 UTC
00 UTC
01 UTC
02 UTC
03 UTC
Areal influence of each predicted trajectory point expands with NearCast length
(Longer-range NearCasts have more uncertainly and should therefore show less detail)

01 UTC
02 UTC
03 UTC
04 UTC
05 UTC
Run Time > 20 UTC 21 UTC 22 UTC 23 UTC
Valid
Derived Vertical Moisture Difference
20 UTC
21 UTC

21Z
New
Data

22 UTC

22Z
New
Data

23 UTC

23Z
New
Data

00 UTC
RUC
6 hr
Cloud
Top
Forecast
Valid 03 UTC
Repeated insertion of hourly GOES data improves and revalidates NearCasts
Convectively
Stable
No Data
Convectively
Unstable
Verifying Radar

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
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
Differential Moisture Transport
Creates
Convective Instability
Stable

Unstable
Dry

Moist
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

Testing planned in NWS offices

- The GOES DPI moisture products provide good data
0 Hour Moisture for 2100UTC
- The Lagrangian NearCasts produce useful and frequently updates
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
GOES Derived Product Images
Next steps:
900-700 hPa PW 
Initial
700-300 PW  24/7 runs
2115 UTC Sat Imagery
GOES DPIs
Need WFO evaluation:
Valid
2100 UTC
Is this a useful forecasting tool?
( If so, when and when not? )
What additional tuning is needed?
What other parameters should be included (LI, CAPE, . . .)
How should it be tested? ( Web-based initially and/or directly to AWIPS? )
13 April 2006 – 2100 UTC
Let’s talk more
0 Hour NearCast
13 April 2006 – 2100 UTC
900-700 hPa GOES PW
0 Hour NearCast

Why we need GOES (not just POES) data for detecting small-scale convection,
An unforecasted mesoscale Derecho which initially moved from MN across south-central WI, decayed and then re-intensified south of Chicago
Comparison of GOES (left) and AIRS (right) data coverage around 0700 UTC 20 July 2006.
Times and lateral limits of AIRS overpasses shown.
For GOES: Details of moisture maximum ( warm colors ) which was initially over Iowa and subsequently moved eastward to support convection over WI and IL are clearly identified in spatially continuous data .
For POES: No AIRS data were available over IA (indicated by white areas), due to combination of:
1) cloud obscurations (e.g., over MN and western IA in later 0900 UTC data), and
2) data gaps between successive orbital paths (e.g., central and eastern IA).
Note: Neither radiosonde nor aircraft moisture data would have been available around 0700 UTC for this area.