A SatelliteBased Objective Analysis Scheme for Nowcasting Applications

1
A Satellite-Based Objective Analysis Scheme for
Nowcasting Applications Robert M. Aune Advanced
Satellite Products Team NOAA/NESDIS/ORA/ARAD and R
alph Petersen Cooperative Institute for
Meteorological Satellite Studies University of
Wisconsin, Madison Analysis of Record
Workshop USWRP June 29 - 30, 2004
2
Project Goals Develop an objective analysis
system for nowcasting that leverages the temporal
and spatial attributes of the GOES
sounder Desired attributes 1) Observation
based, i. e. minimal dependence on forecast
models 2) 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 3)
Computationally efficient to allow fast
dissemination 4) Be capable of updating forecast
guidance in the near term
3
Observation-based Analysis Versus Numerical
Prediction System Numerical Prediction
Systems - Do not guarantee accurate
analyses of atmospheric parameters -
Initial model fields frequently contain
unrepresentative horizontal and vertical
gradients with reduced maxima and minima. -
Uses optimal statistical approaches to extract
information from observations by fitting the
model variables to the observations in a
least-squared sense across three space dimensions
over a discrete time intervals (computationally
intensive) - Absolute accuracy is
sacrificed to achieve the best overall spatial
fit that satisfies the model physics, dynamics
and numerics - Information previously
introduced (maxima and minima) is lost due to
imperfections in the analysis and/or the
prediction model.
4
Observation-based Analysis Versus Numerical
Prediction System Observation-based Analysis
- horizontal and vertical gradients, and
local maxima and minima are preserved -
preserve absolute accuracy -
computationally efficient - requires
observations at high spatial and temporal
resolutions, and effective quality control
5
We are currently testing a system that uses a
LaGrangian approach to optimize the retention of
information provided by the GOES sounder. Tests
with real data are currently being conducted
using full resolution (12 km) derived layer
moisture products from the GOES-12 sounder over
an area in the Central U.S. These tests focus on
the ability of the proposed system to retain and
capture details important to the development of
convective instability. The advantages of using
a LaGrangian to updating short-range model
guidance will be demonstrated for conditions that
are dominated by differential advection.
6
GOES Sounder provides information at 19 Spectral
Bands (14.7 to 3.7 um and visible) hourly at 10
km spacing
7
Benefits of single field-of-view (FOV) processing
5x5 FOV 50km
3x3 FOV 30km
Single FOV 10km
8
Processing of GOES sounder data has increased by
a factor of 15.
9
Satellites are becoming more mesoscale
3 by 3 FOVs for GOES
1 by 1 FOVs for GOES
5 by 5 FOVs for MODIS
10
High spatial coverage of observations is
crucial. Observational coverage provided by the
GOES sounder is enhanced by projecting
observations forward in time along trajectories
using - kinematic trajectories using
analyzed wind fields - pressure gradient
trajectories using heights du/dt
- g dz/dx f v dv/dt - g dz/dy
f u
11
700 hPa mixing ratios from GOES-8 valid 12 UTC
projected forward in time and compared with
corresponding GOES retrievals.
initial
3 hours
Fits remain good through 3 hours. Biases
increase after that, especially at low levels
where diurnal effects are large.
6 hours
12
3x3 GOES sounder observations projected forward
3 hours at three levels track differently due
to wind shear.
13
Statistics comparing trajectory observations to
actual equivalent potential temperature
(K) retrievals from GOES-8.
14
Gridding of observations is performed using a
recursive filter (Hayden and Purser, 1995)
- empirical linear interpolation - does
not use knowledge of model or observational
error - extremely fast Form
Ai aAi-1 (1 a)Ai, 0ltalt1 Forward
filter bias is removed by using adjoint
filter Ai (1 a)(Ai aAi-1
. . . . . aj Ai-j ...) Applies locally
varying scaling and QC Can be applied to limited
area
15
GOES-12 nowcasting test domain 10km
16
An Example of the Trajectory Analysis
Approach The following sequence is an example of
how the trajectory observations can be used to
augment an analysis of precipitable water (PW)
retrieved from the GOES-12 sounder. The analysis
is performed hourly using observations at T-3,
T-2, and T-1 starting at 09 UTC and ending at 15
UTC. The resulting analysis at 15 UTC is then
projected forward 3 hours to produce a nowcast of
PW.
17
Initial analysis of GOES-12 level 1 (sfc-.9s) PW
valid 09UTC 04Nov03 Upper left is corresponding
GOES sounder image. Observation fit is shown at
right
18
Analysis of GOES-12 level 1 (sfc-.9s) PW valid
12UTC 04Nov03 after four analysis updates Upper
left is corresponding GOES sounder image.
Observation fit is shown at right
19
Analysis of GOES-12 level 1 (sfc-.9s) PW valid
15UTC 04Nov03 after seven analysis updates Upper
left is corresponding GOES sounder image.
Observation fit is shown at right
20
3-hour nowcast of GOES-12 level 1 (sfc-.9s) PW
valid 18UTC 04Nov03 after seven analysis
updates Upper left is corresponding GOES sounder
image. Observation fit is shown at right
21
Initial analysis of GOES-12 level 2 (.9s-.7s) PW
valid 09UTC 04Nov03 Upper left is corresponding
GOES sounder image. Observation fit is shown at
right
22
Analysis of GOES-12 level 2 (.9s-.7s) PW valid
12UTC 04Nov03 after four analysis updates Upper
left is corresponding GOES sounder image.
Observation fit is shown at right
23
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
24
3-hour nowcast of GOES-12 level 2 (.9s-.7s) PW
valid 18UTC 04Nov03 after seven analysis
updates Upper left is corresponding GOES sounder
image. Observation fit is shown at right
25
Initial analysis of GOES-12 level 3 (.7s-.3s) PW
valid 09UTC 04Nov03 Upper left is corresponding
GOES sounder image. Observation fit is shown at
right
26
Analysis of GOES-12 level 3 (.7s-.3s) PW valid
12UTC 04Nov03 after four analysis updates Upper
left is corresponding GOES sounder image.
Observation fit is shown at right
27
Analysis of GOES-12 level 3 (.7s-.3s) PW valid
15UTC 04Nov03 after seven analysis updates Upper
left is corresponding GOES sounder image.
Observation fit is shown at right
28
3-hour nowcast of GOES-12 level 3 (.7s-.3s) PW
valid 18UTC 04Nov03 after seven analysis
updates Upper left is corresponding GOES sounder
image. Observation fit is shown at right
29
GIFTS
New Technology for Atmospheric Temperature,
Moisture, Winds
4-d Digital Camera
Horizontal Large area format Focal Plane
detector Arrays
Vertical Fourier Transform
Spectrometer

Time Geostationary Satellite
30
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.
31
Further Reading Browning, K. A., editor, 1982
Nowcasting. Academic Press, 256pp.   Daley, R.,
1991 Atmospheric data analysis. Cambridge
Atmospheric and Space Science Series. Cambridge
University Press, 457pp.   Hayden, C. and J.
Purser, 1995 Recursive filter objective analysis
of meteorological fields Applications to NESDIS
operational processing. J. Appl. Meteor., 34,
3-15.   Menzel, W. P., F. Holt, T. Schmit, R.
Aune, A. Schreiner, G. Wade, and D. Gray, 1998
Application of GOES-8/9 soundings to weather
forecasting and nowcasting. Bull. Amer. Meteor.
Soc., 79, 2059-2077.  
32
1) What can be learned from the literature and
applications of existing methodologies as far as
benefits and limitations of a particular
approach that you may be advocating for an
analysis of record? 2) What are the critical
issues that must be faced in order to
successfully develop a quality analysis of record
at spatial scales of 2.5-5 km every hour? 3) Are
there some aspects of an analysis of record
effort that are more straightforward to
accomplish than others, i.e., specific
variables (temperature vs. precipitation),
real-time analyses vs. retrospective analyses?
33
4) To what extent will the analysis of record be
constrained by limitations of the existing and
future observational data base vs. that available
in the past? What observational data sets do you
view to be most critical? 5) To what extent will
the analysis of record be constrained by
limitations of an underlying model? Sensitivity
to boundary layer parameterizations, soil
moisture, clouds, etc.? 6) What are appropriate
measures of the skill of an analysis of record
on these spatial and temporal scales? 7) What are
the resource implications of a particular
method? Computational costs? RD costs? Ball park
numbers on costs are more than adequate. Keep in
mind that no funds have been committed yet
to support the development of an analysis of
record.
34

• Analysis of Record
• Issues
• Funding to support 30 minute latency
• Proprietary datasets
• Do we impose a wind mass balance
• How do we validate the AOR?
• How can satellite data contribute?
• How can satellite data benefit from the AOR?