Towards better utilization of satellite information A forum on how meteorological satellites should

1
Towards better utilization of satellite
informationA forum on how meteorological
satellites should help more with forecasting the
weather29th Annual National Weather
Association Meeting19 October 2004
http//cimss.ssec.wisc.edu/garyw/nwa2004/tbusi-nw
a2004-gsw.ppt

• Gary S. Wade and Robert M. Aune
• NOAA/NESDIS/ORA(STAR/CoRP) ASPT
• Madison, WI

http//cimss.ssec.wisc.edu/aspt/aspthome.html
UW-Madison
2
Forum rationale and agenda
Goal the exchange of ideas, by open discussion
amongst providers and users, that results in
better utilization of satellite data in weather
forecasting.
Topics 1. Overview of wide application of
current meteorological satellite data. 2.
Consideration of attempts to mainstream current
GOES sounding data in (a) forecast offices and
(b) NWP models. 3. Issues of training for and
access to current, and future, satellite systems.
Measure of success as a grass roots weather
community, can we identify any critical action
items that address better use of satellite data ?
3
Aspects of satellite applications general to
specific
Imagery (monitor qualitative/quantitative,
interpret) Winds (measure motions) Soundings
(derive vertical profiles or quantities) Assimilat
ion of above data into numerical models
Storm/cloud detection, synoptic interpretation,
indicators of turbulence or instability,
multi-spectral combinations (true color images
detection of fog, fire, smoke, volcanic ash,
aerosols, snow, ice) Diagnostic wind fields
(steering of tropical storms synoptic dynamics)
Fields of total precipitable water, atmospheric
stability
4
And turning to
5
Traditional visible satellite imagery
1400 UT 22 June 2004
1730 UT 24 July 2004
Examples from GOES-10 of Pacific Northwest
coastal fog - courtesy of Matt Zaffino of KGW-TV,
Portland, Oregon
6
High resolution visible imagery
GOES rapid scan animation from squall line
development on 09 Oct 2001
http//cimss.ssec.wisc.edu/goes/misc/
7
More animation and feature detection
GOES Imager IR window imagery on 22 June 2003,
indicating an enhanced V signature, often
indicative of severe weather.
http//cimss.ssec.wisc.edu/goes/misc/030622/030622
.html
8
(No Transcript)
9
True color multi-spectral imagery
Hurricane Ivan 1850 UTC 15 Sep 2004 Aqua MODIS
Bands 1,4,3
Liam Gumley, SSEC, UW-Madison
http//www.ssec.wisc.edu/gumley/modis_gallery/
10
Landfall of Hurricane Ivan (during eclipse)
GOES-10 and 12 Sounder Cloud Top Pressure
coverage 16 Sep 2004
11
GOES-N/O/PChanges
Timothy J. Schmit NOAA/NESDIS Advanced
Satellite Products Team (ASPT) in collaboration
with the Cooperative Institute for Meteorological
Satellite Studies (CIMSS) and many others
October 2004
UW-Madison
12
Limitations of Current GOES Imagers

• Regional/Hemispheric scan conflicts
• Low spatial resolution
• Missing spectral bands
• Eclipse and related outages

GOES-N/O/P will supply data through the eclipse
periods.The spacecraft batteries are specified to
be large enough to run through eclipse. Shields
have been added to the secondary mirror spiders.
Outages due to Keep Out Zones (KOZ) will be
minimized.
13
Improved radiometrics on GOES-N
The GOES-N instruments will be less noisy. 
Lower patch temperature is the main driver. Other
modifications have been made to improve the noise
performance on both instruments. Imager For
example, using channel 4 (10.5 micron channel) as
the point of comparison, ground test data showed
a patch low NEdT the GOES-N instrument (SN08)
would be 0.05K. (The similar ground test value
for the GOES-12 imager was 0.07K.) Sounder In
general, The GOES-N LW and MW channels show NEN's
that are about 2/3 of the GOES-12 ground test
(Example  LW on GOES-12 ground test was
0.52mW/(m2Srcm-1) compared to the 0.32 of the
GOES-N instrument).  SW channel NEN's will be
about ¾ of the GOES-12. 
14
Improved calibration on GOES-N
Potential reduction in striping to be achieved
through increasing the Imager's scan-mirror's
dwell time on the blackbody from 0.2 sec to 2
sec.  Analysis shows that the blackbody noise
will be reduced by about 13 in Imager channels
3-5, which should improve the precision of their
calibration by approximately that amount and also
reduce the striping by an unknown amount (since
there are a lot of other factors besides
uncorrelated blackbody errors that cause the
striping). This improvement begins with GOES-N
 
15
Improved navigation on GOES-N
16
GOES-O improved spatial resolution of the 13.3
um band
1 km
2 km
4 km
8 km
GOES-O
GOES-M
17
Summary of changes with upcoming GOES-N series
GOES-N/O instrument changes - GOES-N check-out
is upcoming - Better resolution of the 13.3 um
on GOES-O/P GOES-N/O/P bus change - no eclipse
outages, reduced KOZ outages - better
calibration (colder detectors, longer BB) -
better navigation (earth sensor -gt star tracker)
GOES-N/P
GOES-8/12
18
High spatial resolution imagery - MODIS
Snow covered Wisconsin on 30 Jan 2004 (from Terra
at 1655 UT)
Dust across northern Mexico on 04 Mar 2004 (from
Aqua at 1955 UT)
http//www.ssec.wisc.edu/gumley/modis_gallery/
19
Spatial coverage of wind observations
RAOB
Satellite
12 UTC 05 Oct 2004
20
Environmental steering current
Valid at 1500 UTC 05 Oct 2004
http//cimss.ssec.wisc.edu/tropic/real-time/atlant
ic/winds/winds-dlm.html
21
Wind shear analysis
http//cimss.ssec.wisc.edu/tropic/real-time/europe
/winds/winds.html
22
Satellite measurements of tropical storm
intensities
Hurricane Javier from AMSU on NOAA-16 (21 UT 16
Sep 2004)
Warm core seen at 55 GHz indicative of tropical
storm intensity (MSLP)
http//amsu.ssec.wisc.edu/
23
Satellite winds for dynamics diagnosis
Water vapor tracked wind analyses
From 1245 UTC to 1415 UTC on 15 Oct 2004
http//cimss.ssec.wisc.edu/mesoscale_winds/real.ht
ml
24
QuikSCAT low level winds over oceans
http//manati.orbit.nesdis.noaa.gov/quikscat/
25
Satellite cloud products
http//cimss.ssec.wisc.edu/goes/realtime/grtmain.h
tmlimgrcld
26
Satellite derived total ozone
GOES Sounder (02-06 Feb 2001)
http//cimss.ssec.wisc.edu/goes/realtime/grtmain.h
tmlozone
27
Upper level dynamics inferred from satellite
total ozone determination
A tropopause fold is clearly evident in GOES
ozone imagery and in potential vorticity (PV)
derived from the ETA model on 29 September 1999
at 18 UTC. Ozone and PV show high correlation in
this case.
28
Fire detection
Arizona wildfires GOES-8 ABBA
1715 2215 UTC 20 Jun 2002
http//cimss.ssec.wisc.edu/goes/burn/abba.html
29
Detection of volcanic plumes
USING GOES-R TO HELP MONITOR UPPER LEVEL
SO2 Anthony J. Schreiner, Timothy J. Schmit,
Jun Li, Gary P. Ellrod, Mat Gunshor CIMSS
NOAA/NESDIS
Plume
Simulated IR spectrums for normal and SO2
enriched atmosphere and spectral response
functions
Difference and GOES-R ABI SRF
Plume
GOES Sounder Image Difference
Difference and the GOES Sounder SRF
30
Precipitation estimates from satellite
http//www.ssd.noaa.gov/PS/PCPN/
31
Available current GOES Sounder DPI
http//www.orbit.nesdis.noaa.gov/smcd/opdb/goes/so
undings/index.htmlproducts
32
MODIS derived products CIMSS Direct Broadcast

06 UT 15 Oct 2004
http//eosdb.ssec.wisc.edu/modisdirect/
Total precipitable water
33
Questions to address
A satellite that you use has just failed. How
will the service or product that you regularly
provide be impacted ?
34
Spatial, spectral, and temporal coverage from the
GOES-12 Sounder
01 UT 15 Sep 2004 to 19 UT 16 Sep 2004
http//cimss.ssec.wisc.edu/goes/realtime/realtime.
html
Hurricane Ivan approaches
35
Spatial, spectral, and temporal coverage from the
GOES-12 Sounder
01 UT 15 Sep 2004 to 19 UT 16 Sep 2004
http//cimss.ssec.wisc.edu/goes/realtime/realtime.
html
Hurricane Ivan approaches
36
Real-time GOES Sounder retrieval profiles (from
NOAA/NESDIS OPDB)
http//www.orbit.nesdis.noaa.gov/smcd/opdb/goes/so
undings/skewt/html/skewhome.html
37
Available current GOES Sounder DPI
http//www.orbit.nesdis.noaa.gov/smcd/opdb/goes/so
undings/index.htmlproducts
38
Availability of new satellite products
GHCC/SPoRT to NWS SR
http//weather.msfc.nasa.gov/sport/
39
Contrasting available current GOES TPW DPI
NESDIS OPS
SPoRT
NESDIS OP DEV
CIMSS
NESDIS ops - http//www.ssd.noaa.gov/PS/PCPN/pcpn
-na.htmlSNDR
40
Appearing in NWS offices GOES Sounder DPI
Note enhancement tables for Lifted Index
Memphis Derecho case (08 UTC 22 Jul 2003)
http//cimss.ssec.wisc.edu/goes/visit/sounder_enha
ncements.html
41
Appearing in NWS offices GOES Sounder DPI
Note enhancement tables for Lifted Index
Memphis Derecho case (08 UTC 22 Jul 2003)
http//cimss.ssec.wisc.edu/goes/visit/sounder_enha
ncements.html
42
Temperature andWater VaporIR Sounder Staircase
Spectral Resolving Power (?/? ?)
?Resolving Power _at_ 14 ?m
(1200)
HES, GOES-R (2013-)
(1200)
GIFTS (2008-)
(1200/2800)
CrIS / IASI (2006-)
(1200)
AIRS (2002-)
GOES Sounder (1994-) (3-Axis)
(30)
VAS (1980-) 1st Geo Sounder (Spin-Scan)
(30)
(30)
ITPR,VTPR (1972) / HIRS (1978-)
IRIS / SIRS (1969-70) 1st Sounders
(150-300)
BLUE Leo Red Geo
43
New Era Spaceborne High-resolution IR
AIRS/IASI/CrIS (LEO) to GIFTS/HES (GEO)
CrIS
IASI
AIRS/CrIS
GIFTS
CO
CH4
GOES Sounder
CO2
N2O
N2O
H2O
CO2
CO2
O3
H2O
44
Sensitivity to vertical structureof Temperature
and Water Vapor
Vertical Weighting Functions
Pressure (hPa)
45
Detection of Temperature Inversions Possible with
Interferometer
GOES
GOES
Wavenumber (cm-1)
The detection of inversions is critical for
severe weather forecasting. Combined with
improved low-level moisture depiction, critical
ingredients for night-time severe storm
development over the Plains can be monitored.
Knowing if there is an inversion can also help
improve the profiles estimates.
46
Fundamental CIMSS research striving to make
quality real-time GOES Sounder radiance
observations into practical useful information
for weather forecasting
Atmospheric continuity and evolution are clearly
evident in multi-spectral animation.
Where will clouds be? Comparison between
observed imagery (bottom) and forecast imagery
(top) builds confidence in how well the CRAS
model is assimilating retrieved GOES Sounder
cloud and moisture information.
Where will forecast (GFS) moisture need to be
modified, monitoring trends, to provide a better
forecast for convection (as across Texas)?
Differences between retrieved GOES Sounder TPW
and the GFS forecast values are plotted over the
GOES TPW Derived Product Imagery (DPI).

1800 UT 2 Apr 2004
47
Total and layered precipitable water from the
GOES Sounders at 3x3 FOV and SFOV
16 UT 15 Dec 2003
3x3 FOV
Single FOV
Total PW
Total PW
Note that layered scale (0-30mm) is half that of
total (0-60mm).
Mid layer PW pattern seems closest to that of
total high layer PW is most distinct.
High
High
Mid
Mid
Low
Low
16 UT 24 Dec 2003
48
Differentiation of the three vertical layers of
precipitable water from the GOES Sounders
16 UT 15 Dec 2003
Total PW
High layer PW
Mid layer PW
Low layer PW
Note how ovals of relative maximum PW, over US,
shift with height.
See - http//cimss.ssec.wisc.edu/goes/realtime/gr
tmain.htmluspw-3l
49
Example of real-time GOES Sounder Derived Product
Imagery Focused on Wisconsin
GOES total precipitable water (TPW) DPI at 1700
UTC on 25 Aug 2003
GOES (lt40mm) evidently drier than first-guess
GOES (40-50mm) markedly moister than first-guess
(Shouldnt such differences impact forecasts ?)
50
Can there be a quantum step in GOES Sounder
retrieval development?
GIMPAP funded research GOES Retrieval Science
team at CIMSS (Directed by Dr. Jun Li summer
2004)

• Foci of new approaches for retrieval improvement
• - Time continuity (take advantage of high
  temporal resolution)
• Spatial filtering (reduce noise of upper level
  channels)
• Model independent first-guess (realistically
  characterizing surface emissivity as well as
  optimally using radiances to improve the
  first-guess)

Design and implement the needed algorithm
modifications, followed by validation study of
impact on retrieved temperature/moisture fields.
51
Time continuity of water vapor sensitivity to
GOES Sounder radiances
52
Questions to address
A satellite that you use has just failed. How
will the service or product that you regularly
provide be impacted ?
Given that forecasts sometimes miss the mark,
what poor forecasts have you seen where moisture
data were critical ?
53
GOES Observed Imagery versus CRAS Forecast
Imagery(IR window over upper Midwest)
http//cimss.ssec.wisc.edu/goes/realtime/cf/anilat
estgovcf.html
54
GOES Observed Imagery versus CRAS Forecast
Imagery(IR water vapor over NE Pacific)
http//cimss.ssec.wisc.edu/goes/realtime/cf/anilat
estgovcfp.html
55
And turning to
Bob Aune
56
Quantitative Use of Information from
Meteorological Satellites
Robert M. Aune Advanced Satellite Products Branch
NOAA/NESDIS/ORA/CoRP Madison, WI
57
Utilizing Information from Spaced-based Observing
Systems Nowcasting Analysis monitoring Objective
analysis for nowcasting Forecasting
(NWP) Initializing numerical prediction
models Research Applications Model development
Dynamics and physics
58
Utilizing Information from Spaced-based Observing
Systems Nowcasting Analysis monitoring Objective
analysis for nowcasting Forecasting
(NWP) Initializing numerical prediction
models Research Applications Model development
Dynamics and physics
59
High Level IR Wind Speed Bias Met Office global
forecast analysis minus satellite observations
60
GOES-10 IR winds compared to the Met Office
global forecast analysis
61
European Centre for Medium Range
Prediction Daily Monitoring Statistics GOES-10/12
Clear-sky radiances (CSR)
Data is shown for GOES series longwave infrared
and water vapor channels. The CSR observations
received at ECMWF are area averages of those
pixels of the images that were diagnosed as clear
by the data providers (CIMSS/NESDIS). Area
averages are for about 4545 km2.
62
2004101306 GFS OPERATIONAL GOES SOUNDER DATA
DISTRIBUTION CHANNEL 3, CLOUD CLEARED
63
2004101306 GFS OPERATIONAL GOES SOUNDER DATA
DISTRIBUTION CHANNEL 3, CLOUD CORRECTED
64
Are NCEP Models Really Using Satellite
Observations?
Eta-GOES
Total precipitable water differences (mm) NCEP
operational analyses minus retrieved total
precipitable water from the GOES sounders GOES
sounder observations are included in the NCEP
operational data stream.
GFS-GOES
GOES Sounder Locations
65
Why are there differences in the NCEP analyses?
Total precipitable water differences (mm) Eta
analysis minus GFS
analysis valid 00UTC August 15, 2003.
66
Validating EDAS Radiance Assimilation against
GOES Total Precipitable Water Retrievals
PW differences (mm) at the end of
the assimilation cycle (tm00) using two iteration
in the outer loop throughout the entire
assimilation cycle.
PW differences (mm) at the end of
the assimilation cycle (tm00) using one iteration
in the outer loop throughout the entire
assimilation cycle.
Iterations to Solve RTE EDAS 1 GDAS 2 NESDIS
Retrieval 3
67
An Objective Nowcasting Tool that Incorporates
Geostationary Satellite Measurements Robert M.
Aune Advanced Satellite Products
Branch NOAA/NESDIS/ORA/CoRP and Ralph
Petersen Cooperative Institute for Meteorological
Satellite Studies University of Wisconsin, Madison
Project Goals Develop an objective analysis
system for nowcasting that is observation based,
i. e. minimal dependence on forecast
models. Give priority to preserving vertical and
horizontal gradients in the observed fields with
the goal of detecting extreme variations in
atmospheric parameters and identifying the onset
of significant weather events. Must be
computationally efficient to allow fast
dissemination. Be capable of updating
forecast guidance in the near term.
Symposium on Planning, Nowcasting and Forecasting
in the Urban Zone January 12, 2004
68
Analysis of GOES-12 level 2 (.9s-.7s) PW valid
15UTC 04Nov03 after seven analysis updates Upper
left is corresponding GOES sounder image.
Observation fit is shown at right
69
3-hour nowcast of GOES-12 level 2 (.9s-.7s) PW
valid 18UTC 04Nov03 Upper left is corresponding
GOES sounder image. Observation fit is shown at
right
70
Future Issues Future meteorologists will have up
to the minute access to digital atmospheres that
will be as accurate, as the observations used to
build it. The quantity of information defining
the physical and dynamical state of our
atmosphere, collected in near real time, will
become unmanageable the vast majority of these
observations will come from remote sensing
platforms. An observation-based analysis system
could serve as an intelligent data compression
tool, generating detailed analyses that can be
readily transmitted to nowcasters in the
field. Visualization of these data will be also
an important issue.
71
Analysis of Record
The National Weather Service has an immediate and
critical need to produce real-time and
retrospective analyses at high spatial and
temporal resolution in order to verify forecasts
for the National Digital Forecast Database. The
term Analysis of Record (AOR) has been used
provisionally to describe such analyses. The
National Weather Service (NWS) has been directed
to provide digital forecasts out to 7 days, at up
to hourly temporal resolution, on a 5 km grid
the National Digital Forecast Database (NDFD) has
a nominal grid-spacing of 5 km across the United
States and represents a blend of objective
forecast guidance and forecaster edits.
Forecast parameters include temperature,
dew-point temperature, wind, precipitation,
clouds, and weather.
72
Analysis of Record
Version 0.0 Configuration Horizontal
Resolution 5 km Analysis Frequency hourly Late
ncy 30 min Host NCEP/EMC Parameters Temp
erature Dew-point temperature Wind Pre
cipitation Sky cover (satellite
based) Initial Design 20 km RUC Analysis
downscaled to 5 km Add conventional data - 2D
variational Analysis Precipitation radar Sky
cover GOES
73
Utilizing Information from Spaced-based Observing
Systems Nowcasting Analysis monitoring Objective
analysis for nowcasting Forecasting
(NWP) Initializing numerical prediction
models Research Applications Model development
Dynamics and physics
74
How Do Observations Influence Model Initial
Conditions?

• Factors that influence forecast model initial
  conditions
• Model background (previous forecast)
• Observation density (space and time interval)
• Observation type (forward model)
• Instrument error
• Imposed physical and dynamic constraints
• Observation representativeness
• Quality control

Background and observations are combined using an
optimal statistical approach to produce an
initial state that the forecast model will
accept.
75
Radiance Assimilation Sequence Spectral Forecast
Model The assimilation of radiance information
from satellites is not a trivial process. 1.
Transform variables T,q _at_ grid-points
Spectral, timet(0) 2. Forecast Spectral,
timet(n) 3. Inverse transform Spectral T,q _at_
grid-points 4. Horizontal interpolation T,q
Observation points 5. Vertical interpolation
T,q 42 levels 6. Radiative transfer T,q
Tb 7. Minimize cost function Tb Tb 8. Use
adjoint of radiative transfer Tb T,g 9.
Transform variables T,q _at_ grid-points
Spectral, timet(1)
76
Impact of NOAA Polar Orbiting Satellite Data on
NCEP Global Models 12 Year Period
77
ANOMALY CORRELATIONS FOR SIX OPERATIONAL GLOBAL
MODELS 5-day 500 hPa heights, Northern and
Southern Hemispheres
78
Assessing Impact of Satellite Observations On
Regional Models Using Data Denial?
Impact of Remotely Sensed Data on 24-hour Eta
Forecast SSM/I Total precipitable
water GOESPW Three layer precipitable
water TOVCD T soundings above cloud GOESCD IR
cloud drift winds GOESWV Water vapor cloud
drift winds
Impact of In-situ Data on Eta 24- hour Eta
Forecast RAOBM Rawinsonde T and Td ACARM
Aircraft T RAOBW Rawinsonde winds ACARW
Aircraft winds SFCLM Surface land observations



Courtesy of T. Zapotocny, CIMSS
79
3h 20km RUC cloud-top fcst w/ GOES cloud
assimilation
Verification Cloud-top pressure based on NESDIS
product
3h 40km RUC cloud-top fcst No GOES cloud
assimilation
Effect of GOES cloud-top pressure data on 3-h
cloud forecasts from the Rapid Update Cycle
(RUC) Valid 1200 UTC 9 Dec 2001
80
(No Transcript)
81
(No Transcript)
82
(No Transcript)
83
Utilizing Information from Spaced-based Observing
Systems Nowcasting Analysis monitoring Objective
analysis for nowcasting Forecasting
(NWP) Initializing numerical prediction
models Research Applications Model development
Dynamics and physics
84
Created July, 2002

• The JCSDA Mission Goals
• MissionAccelerate and improve the quantitative
  use of research and operational satellite data in
  weather and climate prediction models
• Goals
• Reduce from two years to one year the average
  time for operational implementation of new
  satellite technology
• Increase uses of current satellite data in NWP
  models
• Advance the common NWP models and data
  assimilation infrastructure
• Assess the impacts of data from advanced
  satellite sensors on weather and climate
  predictions

85
JCSDA Partners
Management Oversight Board
86
CIMSS Regional Assimilation System
CRAS
Purpose
To assess the impact of space-based observations
on numerical weather prediction First real-time
CRAS prediction system to use cloud and moisture
products from the GOES sounders (1995) CRAS
development was guided by validation against
satellite observations
87
4th Order Diffusion
The CIMSS Regional Assimilation System (CRAS)
forecast model was designed conserve model
variables and their horizontal and vertical
gradients. Dissipation is minimized. The
moisture transport experiment shown here
illustrates the advantage of using a sixth-order
filter in place of fourth-order horizontal
diffusion.
6th Order Filter
88
48 hour simulation Advection Equation
40 m/sec
4th order diffusion
40 m/sec
6th order filter (Raymond, 1988)
40 m/sec
Exact Solution
89
CRAS Configuration Model Pseudo
non-hydrostatic, explicit moist
physics Grid Limited area, re-locatable Arakawa
C Projection Lambert conformal/polar
stereo Resolution Horizontal 61km to
10km Vertical Sigma, 40 levels, floating
top Platforms AIX, IRIX, Linux Performance 65
minutes on a single 2.2 GHz Intel Xeon
(Linux) (60hr fcst, 151x97x38 grid, dx61 km,
tstep300 sec) Input Observations In
situ RAOBs, surface data, ACARS,
profilers Geostationary 3-layer precipitable
water - GOES-9/10/12 sounders Cloud-top
pressure and effective cloud amount -
GOES-9/10/12 sounders, GOES-12 imager 4-layer
thickness - GOES-9/10/12 sounders Cloud-track
and water vapor winds - GOES-9/10/12 Polar Cl
oud-top pressure and effective cloud amount -
MODIS (Aqua and Terra) Other Gridded hourly
precip, Stage II, from NCEP SST and sea ice
coverage from NESDIS IMS
90
CRAS Forecast Validation
http//cimss.ssec.wisc.edu/model/daily/satellite/s
atellite.html

• SATELLITE VALIDATION
• CRAS forecasts 11um and 6.5um satellite imagery.
  This imagery is being validated with actual GOES
  imagery.

CRAS 24hr and 12hr forecast 11um images
validated against GOES Imager
CRAS 36hr forecast 11um satellite image at 40km
resolution
CRAS 36hr forecast 11um satellite image at 40km
resolution
CRAS 36hr forecast 6.7um satellite image at 40km
resolution

• SURFACE/UPPER AIR VALIDATION
• CRAS surface and upper air forecasts are
  validated against observation.

http//cimss.ssec.wisc.edu/model/daily/surface/sur
face_validation.html
CRAS 12Z surface wind initialization validated
against observation
CRAS 12-hr forecast surface Td validated against
observation
91

• GOES-12 CLOUD/PW DATA IN THE 20-km CRAS

• GOES-12 imagery is mapped onto the CRAS map
  projection and used to validate the CRAS forecast
  imagery. A comparison of a 24-hr, 18-hr, 12-hr
  and 6-hr forecast 11um image with actual GOES
  imagery is shown above. A quantitative
  validation approach is under development.

92
Using GOES-10 6.7um to validate the 40km East-Pac
CRAS
93
61km CRAS Outperforms 22km Eta
The figures below show 48-hour forecast of
24-hour precipitation accumulation ( gt 6mm )
valid 12 UTC September 26, 2002, from the CRAS
real-time forecast and the operational NCEP Eta
forecast. The CRAS uses 3-layer precipitable
water and cloud top pressure retrievals from the
GOES sounders to initialize water vapor and
clouds. The precipitation for this case was
generated by Tropical Storm Isidore as it came
ashore.
CRAS
NCEP Eta
48 hour forecast of 24 hour accumulated
precipitation ( gt 6mm ) from the CIMSS Regional
Assimilation System (CRAS) valid 12UTC, September
26, 2002. Validating rain gauges are shown in
green. Threat score .2035 Threat bias 1.45.
48 hour forecast of 24 hour accumulated
precipitation ( gt 6mm ) from the NCEP Eta
forecast model valid 12UTC, September 26, 2002.
Validating rain gauges are shown in green.
Threat score .1654 Threat bias .90.
94
Real-time CRAS at CIMSS All forecast initialized
at 00/12 UTC Location Res BCs Hours Input_________
____________ CONUS 61 km GFS 60 GOES PW3,
winds, clds, sfc, precip Central (nest) 20
km CRAS 36 GOES PW3, winds, clds, sfc,
precip Antarctica 48 km GFS 48 MODIS clds, TPW,
winds NE Pacific 40 km GFS 60 GOES PW3, winds,
clds, sfc Production Machines Dual 2.0 GHz Intel
Xeon, 1Gbyte RAM, Linux, Intel FORTRAN Dual 2.4
GHz Intel Xeon, 2Gbyte RAM, Linux, Intel
FORTRAN Website Production sgi Octane (Dual 300
MHz R12K) running xsau graphics package (xlib)
http//cimss.ssec.wisc.edu/model/daily/daily.ht
ml
95
Products available from the four CRAS real-time
production runs at http//cimss.ssec.wisc.edu/mod
el/daily/daily.html
61 km CONUS
40 km NE PAC
40 km NE PAC
CRAS forecast imagery
CRAS precipitation forecast
48 km Antarctica
CRAS 12-hr forecast radar
20 km Cent US
Observed radar
96
CRAS Forecast Radar
12-hour forecast rain-rate from the real-time 20
km CRAS (above) valid 12UTC 16Apr03. The
validating composite radar is shown at right.
NEXRAD Composite Radar (Unisys)
97
33 hr CRAS forecast 11um product
GOES-10 11um valid 21UTC 20Oct04
33 hr CRAS forecast 6.7um product
GOES-10 6.7um valid 21UTC 20Oct04
98
09 hr CRAS forecast 11um product
GOES-10 11um valid 21UTC 20Oct04
09 hr CRAS forecast 6.7um product
GOES-10 6.7um valid 21UTC 20Oct04
99
(No Transcript)
100
3D Cloud Initialization in the CIMSS Regional
Assimilation System (CRAS) using the GOES Sounders
GOES
Cloud-top pressure and effective cloud amount are
retrieved every hour from the GOES sounders and
are used to clear model cloud profiles and to
construct multi-layer cloud profiles in the CRAS.
101
36 hour loop (hourly) of forecast 11um images
from the realtime 20km CRAS commencing 12UTC,
August 10, 2004
102
Can Numerical Prediction Models Forecast
Visibility?
Requirements for Visibility Prediction Powerful
computers to allow higher resolution (horizontal
and vertical) while maintaining timely
delivery New observing systems with higher
spatial and temporal coverage that observe
low-level moisture, aerosols, winds, and
pollutants Sophisticated model physics to
predict cloud formation and dissipation Mass
conserving model dynamics to accurately predict
the transport of cloud and moisture Improved
data assimilation methods
Limitations Accurate and timely observations of
clouds and water vapor are required to predict
the onset and dissipation of precipitation and
fog Forecast models generally dont conserve
mass and gradient structures Need
high-resolution climatologies of surface
parameters to specify the lower boundary
The CRAS is used at CIMSS to exploit the spatial
and temporal advantages of the GOES-10 sounder to
initialize moisture and clouds in the Eastern
Pacific. A 24-hr CRAS forecast of low level RH
is shown here with areas of low visibility
depicted in red.
103
Hourly water vapor and cloud observations from
the GOES-12 sounder are used to initialize 36-hr
CRAS forecasts for central U.S. Time series
plots are generated for instrumented sites used
by the Road Weather Information System (RWIS)
maintained by WisDOT.
30 hour surface relative humidity forecast from
the CIMSS Regional Assimilation System (CRAS)
valid 18UTC, Dec 6, 2003
36 hour forecast time series of temperature, dew
point, wind, precipitation and cloud cover from
the CIMSS Regional Assimilation System (CRAS)
initialized at 12UTC Dec 5, 2003.
Surface stations reporting fog at 15UTC, Dec 6,
2003
Road Weather Information System (RWIS) tower
locations maintained by WisDOT
104
69hr Forecast Track of Hurricane Frances from
the61 km CRAS valid 09UTC September 5, 2004
Actual track
The 61 km real-time CRAS uses 3-layer
precipitable water and cloud-top pressure from
the GOES sounders to initialize clouds and water
vapor. Five inserts are made during a 12 hour
forecast spin up. Initial winds and temperatures
come from the NCEP GFS.
Actual position at 09UTC 05Sep04
69-hr CRAS forecast position of Hurricane Frances
C indicates CRAS 6-hourly positions based on 850
hPa vorticity
Initial location
105
Initial conditions for the 20km CRAS forecast of
Hurricane Frances valid 12UTC September 3, 2004,
containing water vapor and clouds from the
GOES-12 sounder.
Surface Wind Speed (kts)
Boundary Layer RH ()
11um Brightness Temperature (K)
Total precipitable water (mm)
106
36hr forecast of Hurricane Frances from the 20 km
CRAS valid 00UTC September 5, 2004
Surface Wind Speed (kts)
Rain Rate (mm/hr)
Hourly Precip Amount (1/100 in)
11um Brightness Temperature (K)

107
The future lies in high-spectral resolution!
3 May 1999 -- Oklahoma/Kansas tornado outbreak
All three solutions show rapid atmospheric
destabilization (decreasing LI) between 14 and 20
UTC. GIFTS better depicts the absolute values
and tendencies compared to GOES. The total
precipitable water (TPW) increases through the
period. Both current and future sounding
measurements capture the correct trends.
UW-Madison/CIMSS
108
Questions to address
A satellite that you use has just failed. How
will the service or product that you regularly
provide be impacted ?
Given that forecasts sometimes miss the mark,
what poor forecasts have you seen where moisture
data were critical ?
How would I like to be able to access satellite
data in trying to positively impact shorter term
gridded forecasts ?
109
VISIT Training from COMET
Modules are readily accessible and many are on
satellite applications.
http//www.cira.colostate.edu/ramm/visit/ts.html
110
Using VISITview for training
Use real-time collaborations or build your own
modules.
http//www.ssec.wisc.edu/visit/dpi.html
111
And turning to
Woody Wang
112
The COMET MetEd training program
http//www.meted.ucar.edu
113
Questions to address
A satellite that you use has just failed. How
will the service or product that you regularly
provide be impacted ?
Given that forecasts sometimes miss the mark,
what poor forecasts have you seen where moisture
data were critical ?
How would I like to be able to access satellite
data in trying to positively impact shorter term
gridded forecasts ?
What satellite products are there that I mighty
actually use, if I understood them better, or had
more confidence in them ?
114
Questions to address
A satellite that you use has just failed. How
will the service or product that you regularly
provide be impacted ?
Given that forecasts sometimes miss the mark,
what poor forecasts have you seen where moisture
data were critical ?
How would I like to be able to access satellite
data in trying to positively impact shorter term
gridded forecasts ?
What satellite products are there that I mighty
actually use, if I understood them better, or had
more confidence in them ?
Anything else to consider to take better
advantage of satellite data ?
115
Thanks to contributors
Dr. Sherwood (Woody) Wang (COMET) NPOESS
training section Liam Gumley et al (CIMSS)
MODIS Scott Bachmeier, Scott Lindstrom (CIMSS)
satellite apps and training Dr. Jun Li (CIMSS)
GOES retrieval science Jim Nelson, Tony Schreiner
(CIMSS) GOES retrieval processing and apps Dr.
Hank Revercomb (UW-SSEC) high spectral apps Dr.
Robert Rabin (NSSL, CIMSS) GOES winds and
satellite apps Chris Schmidt et al (CIMSS) GOES
fires and ozone Tim Schmit (NESDIS) overarching
direction and support Dave Stettner et al (CIMSS)
tropical cyclone apps Matt Zaffino (KGW TV
Portland) local imagery To all those who share
their progress and post their efforts on the web
and to all the engaged participants in the
audience !
116
Sample of comments during the workshop
Satellites help -to make a more efficient
use of forecaster time -to support/quality
control lone observations -to improve
warning lead times, from visible loops
-to convey information to the average person
-(more) if there are
operational web sites -to issue aviation
forecasts TAFS (ceiling/visibility)
-(more) if there are not eclipse outages (more
than just an annoyance) -when there is
radar blocking over land (e.g. in AK)
-by providing fog products
-to improve the quality of NWP forecasts

(from TJS)