1. Extending Geostationary Satellite Retrievals from
Observations into Forecasts
Using GOES Sounder Products to Improve Regional Hazardous Weather Forecasts
Ralph A. Petersen : University of Wisconsin – Madison
Robert M Aune : NOAA/NESDIS/STAR - Advanced Satellite Products Branch - Madison, WI

2. Background / Objectives
The overall goal is to provide new tools to help forecasters to
identify areas where convection is (and is not) likely to 3-6 hours
in advance of storm development using moisture and temperature
products from current/future Geostationary satellites.
Due to the perishability of the information, NearCasts
requires numerical approaches that are extremely
fast and notably different from those used in numerical
weather prediction covering longer time periods.
The new Lagrangian approach to objective NearCasting
optimizes the impact and retention of information provided by GOES / SEVIRI
Designed to ingest and preserve extreme vertical and horizontal variations of moisture/temperature observed and updated by GOES
GOES data not currently used in NWP over land
Extremes are under-represented in conventional NWP analyses and forecasts.
NearCasting products include 1-6 hour projections of the GOES multi-layer Moisture & Temperature retrievals, along with predictions of derived areas of convective destabilization.
Existing GOES sounder DPI products used as surrogate for GOES-R ABI
Tests underway using MeteoSat data
Including additional stability indices
Methodology directly transferable to GOES-R and SEVIRI
Higher spatial/temporal resolution of ABI products will improve products further.
The NearCasting System adds a short-range forecast component to GOES observations.
Although the value of GOES IR sounding profiles vanishes after clouds form, the NearCast System anticipates destabilization by dynamically projecting GOES multi-level products to future locations, even after subsequent clouds develop.
When used in combination with other conventional data, the hourly-updated NearCast output can help forecasters identify/isolate areas where severe convection is likely to occur 1-6 hours in advance.
Helps reduce size of warning/watch areas, extend lead times of watches/warnings, increases Probability of Detection and reduces False Alarms (non-occurrence)
Especially useful for hard-to-forecast isolated summer-time convective events
We are developing tools to allow forecasters to:
- Project detailed satellite products into the near future – objectively
- Preserve observed extreme gradients and max/min features
- Retain earlier IR information in areas where clouds later grow
- Provide rapid and timely updates to NWP guidance- Focus on weather events of greatest local need
Use of the Lagrangian approach provide products very quickly and :
- Data can be inserted directly without ‘analysis smoothing’
- Retains maxima and minima and extreme gradients
- Spatial resolution adjusts to available data density
- GOES derived products can be projected forward at full resolution – even as they move into what become ‘data void’ cloudy areas
- Data can be tracked (and aged) through multiple time periods
- Uses all data at time observed – not binned and smoothed

3. How the Lagrangian NearCasts work:
Objectives:
♦Preserve Data Maxima/Minima/Large Gradients
♦Use Geostationary satellite data at Full Resolution
♦Be Fast
Methodology:
The Lagrangian approach first interpolates wind data to locations of full resolution GOES multi-layer moisture & temperature observations
Next, these high-definition data are moved to future locations, using dynamically changing winds with ‘long’ (10 min.) time steps.
. Finally, the moved ‘obs’ values from each layer are then both:
1) Transferred back to an ‘image’ for display of ‘predicted DPIs’,
2) Several parameters are combined to produce derived parameters and
3) Results between layers are compared to obtain various “Stability Indices” that are combined with ‘conventional tools’ to identify mesoscale areas where severe convective will develop -
even after convective clouds appear.
13 April 2006 – 2100 UTC
hPa GOES PW
0 Hour Ob Locations
13 April 2006 – 2100 UTC
hPa GOES PW
3 Hour NearCast Image
Vertical Moisture Gradient (indicating Convective Instability)
( hPa GOES PW hPa GOES PW)
3 Hour NearCast : Valid 0000UTC
The Lagrangian approach determines the future location of atmospheric tracers (in this case mid- and low-level moisture observations from GOES) through simple forward transport of the parcels from one location to the next using a minute time step.
The wind field changes during the NearCast period are determined by solving the Primitive Equations of Motion in their simplest (Total Derivative) form, where accelerations of air parcels are determined from the difference of the Coriolis and Pressure Gradient forces at each parcel location. This eliminates 1) the need to calculate the highly non-linear inertial advective terms in Eulerian Model (fixed grid) form of the equations, 2) the time-step limitations imposed in grid point models (compared to a 10 km Eulerian model which uses sec time steps, the Lagrangian model transporting 10 km GOES moisture products can use minute time steps), and 3) the smoothing of moisture fields which is inherent in advection calculations used in grid-point models.
The key outputs from the NearCast Model are 1) projections of future mid- and lower-level moisture/temperature fields (in the layers observed by GOES) and 2) derived fields identifying areas where Stability Parameters (in the case shown here Convective Instability) indicate that convection is likely (or not likely) to form in the near future. For this example, Convective Instability occurs when lower-level moisture is overlaid by dry air aloft. The process of Convective Destabilization commonly occurs through the differential transport between the lower- and mid- troposphere, a result of both directional and speed shear across the layers. When Convective Instability occurs, lifting of the entire layer can change it from being stable to absolutely unstable over a very short time period. It should be noted that similar differential advection processes produce the major changes to other Stability Indices as well. These types of scenarios are optimal for using GOES IR products, since 1) the convective destabilization occurs in areas which had previously be stable and therefore are very likely to have been free of clouds and 2) it relies in part on one of the best observations from GOES - the location and depth of mid-tropospheric dryness/moisture.
Although it was not included as a project task in last years briefing, a number of additional stability parameters are in the process of being added to the NearCasting system. The choice of which Stability Parameters to include is being made to assure that they reflect those observations that GOES performs best, 2-3 layers of moisture and 5-6 layers of temperature. Candidate indices include: Lifter Index (using both surface and lower-tropospheric temperature/moisture data), Totals index, CAPE, CINH, K Index, as well as Ensembles of Combined Indices. These indices are very similar to those developed (or under consideration) for the MeteoSat GII products.
Actual observations of areas of Convective Instability itself are relative rare using only GOES water vapor observations because, and even when they do occur, they often become quickly obscured by the subsequent formation of convective clouds and cirrus outflow shields. The fact that the NearCast System projects the moisture observations to future locations allows earlier observations taken in stable conditions (before the convection forms) to be projected into the future – providing a picture of the amount of support that will be available for the subsequent convection and whether additional storms are likely to develop under the first cirrus shield.
Verification
Updated Hourly - Full-resolution 10 km data - 10 minute time steps

4. Moving GOES data from Observations to Forecasts
US Example:
Formation of
Strong Pre-Frontal Convection
Vertical Moisture Gradient
( hPa GOES PW -
hPa GOES PW)
7 Analyses Plus 6-Hour NearCast from 1100UTC
10 February 2009
Pseudo-Convective Stability
Verification: Radar/Reports

5. Can the NearCasting approach be applied to SEVIRI data?
Tests were conducted with 2 time periods of retrievals obtained 8 and 6 hours prior to development of the F2/T4 tornado that occurred in Częstochowa, Poland near 16UTC - 20 July 2007.
Full description by Pajek, Iwanski, König and Struzik
Results using only 09UTC retrievals (provided by Konig) shown here
NearCast results valid from 09UTC to 15UTC
Initial Wind and Geopotential data from NCEP 0.5o resolution
Results displayed on 0.25o output grid
NearCasts were made for more variable than in previous US tests
Multi-Layer and Total Precipitable Water
Lower- and Mid-tropospheric parameters:
Temperature
Mixing Ratio
Temperature at LCL
Equivalent Potential Temperature
Several Stability Indices were derived from NearCasts of these primary variables
Note: Apologies for “quality” of graphics - but they get the point across
Currently integrating NearCasts into McIdas-V

6. Lifted Index – 840-480 hPa – From 09Z– F0-06 : Valid 09Z-15Z
Difference between TLCL840 and T480
Observations show:
Weakest Stability South-West of tornado development-
…but…
NearCasts show:
Initial Instability weakens and moves East
Second area of Instability forms to west and moves to tornado site by 15Z

7. True Convective Instability
Vertical Equiv. Pot. Temp. Difference (∂Өe/∂p) – hPa – From 09Z :Valid 09Z – 15Z
True Convective Instability
Convective Instability
Observations show:
Weakest Stability South-West of tornado development
NearCasts show combined effects of differential advection between Warm/Moist air at low levels and Dry/Cool air aloft
Area of greatest Convective Instability moves to tornado site at same time as rapid lapse rate change

8. Summary
Tests show GOES DPI NearCasts can isolate the pre-convective environment for hazardous weather in many US cases
Effect for detecting isolated convection and reducing warning area sizes
Important for predicting various type of Hazardous Convection
Useful in adding detail to Heavy Precipitation Forecasts
GOES Temperature Soundings provide additional information beyond TPW in defining Convective Potential when using Өe
Tests using SEVIRI retrieval positive
Useful in diagnosing the pre-convective environment evolution
Applicable to many forecasting Indices
FUTURE
Beta-test version of NearCasting system available for distribution by mid-October
Major US testing at SPC/NSSL in 2010
Plans for improved graphics using McIDAS-V Ensembles , Consistency, . . .

9. Other Potential Areas of Interest from GOES-R Projects
McIDAS-V
McIDAS-V can be used as an ‘integration platform’ for analyzing/visualizing data and products
Global high-resolution Model-Derived Proxy ABI Datasets Available for Research and Demonstration Activities
Inter-calibration of GOES Sounders and Imagers with Hyperspectral measurements for future GOES-R ABI
Combining GOES-R ABI and polar orbiting advanced IR sounder for improved atmospheric profile
Water Vapor Calibration/Validation/Bias Removal using GPS/TPW
GOES-R AWG Legacy Atmospheric Profile (LAP) algorithm development
Lifted Index (K)
CAPE s
SEVIRI
RAOBs RAOBs RAOBs