1.
IntroductioN
Scientists
at NOAA's Forecast Systems Laboratory are attempting to address the
issue of
local-scale modeling using a new version of our Local Analysis and
Prediction
System (LAPS, Albers et al. 1996) to directly initialize the cloud and
precipitation fields of a local forecast model (hot start). LAPS
has been
integrated into the Advanced Weather Information Processing System
(AWIPS) as
part of the National Weather Service (NWS) modernization. Research to
expand
LAPS capabilities is one avenue toward providing advanced technologies
and new
innovations to the operational forecaster. One of the greatest
deficiencies
of numerical weather prediction models is their lack of skill in
predicting
clouds and precipitation in the early portions (0-6 h) of their
forecast
period.
A project
has been underway for the past year to develop a national,
high-resolution,
real-time analysis of water in all phases (McGinley
et al. 2000) using LAPS. As part of this project, two
significant
improvements were implemented which ultimately made it possible to
perform a
diabatic initialization of a mesoscale model with analyzed hydrometeor
species. This
paper describes work in progress, and the next step toward advancing
hot-start
model initialization. A demonstration
of the local modeling capabilities with this system will be shown for a
severe
weather event in eastern Colorado on 20 July 2000.
1.1
LAPS
Background
During the 1980s FSL
conducted forecast exercises to test
its workstation prototypes. Forecasters
were burdened with the impossible task of reviewing all the incoming
data made
possible through new technologies, while producing timely forecasts. It
became
obvious that local data needed to be objectively analyzed in
conjunction with
nationally disseminated data. Conceived
as a resolution to this challenge, LAPS was designed to analyze all
local data
in real time on an affordable computer workstation, using its own
output fields
to initialize local-scale forecast models.
A more detailed review of LAPS is available in McGinley et al.
(1991).
Under development since
1990, LAPS combines nationally
disseminated data with local data for real-time objective analyses of
all data
available to the local weather forecast office. LAPS
analyses are of suitable quality to initialize a local-scale
forecast model. With a wide
variety of private and government users around the world, it has
demonstrated a
robust capability to combine nearly all available sources of
meteorological
information into a single, coherent three-dimensional view of the
atmosphere
for real-time "nowcasting" and short-range prediction. Throughout its
history, the LAPS analyses have been coupled with a variety of
mesoscale
forecast models, including RAMS, MM5, Eta, and ARPS. Its versatility to
ingest
a multitude of data types combined with its portability and
computational
efficiency have made it ideal for such applications. Mesoscale forecast models
can address specific problems of a
small forecast domain with greater detail than can be achieved with
nationally
disseminated model guidance (Snook et al. 1998). Typically
this has been identifying mountain waves, quantitative
precipitation forecasts (QPFs), and locating the sea breeze interaction
and
other boundaries (Szoke et al. 2000).
Forecasts have been used to identify regions of instability in
the
preconvective environment. Since 10-km
grid resolution is generally too coarse to predict convection and storm
interactions, the models generally have not been found to be that
useful for
forecasting actual storms, though some rather dramatically successful
forecasts
have been documented, Szoke and Marroquin 2000.
The LAPS system is
routinely tested with new data sources
and innovative improvements, using more than "conventional" data, which
have potential for national dissemination.
2.
LAPS
enhancements for water in all
phases
2.1 Cloud Scheme
As part of our recent
upgrades, two major changes have
occurred. First, the LAPS cloud
analysis (Albers et al. 1996) was modified to improve the diagnosis of
hydrometeor concentrations in all phases. The cloud analysis begins by
combining a variety of data sources including a model first guess
of cloud liquid and ice (or relative humidity), observed satellite
brightness
temperatures, surface and aircraft cloud observations, radar
reflectivity
(low-level or volume), and visible satellite imagery.
Other LAPS state variables are used for supporting information,
such as the temperature, humidity, and height analyses. At this point,
a
three-dimensional distribution of cloud fraction is calculated. This cloud field is then processed into
concentrations of the various water species using temperature,
humidity,
stability, and radar reflectivity. The
analyzed water species include cloud liquid and ice, as well as
precipitating
rain, snow, and ice. The precipitating
species are calculated with an improved one-dimensional model that
considers
these hydrometeors (mainly indicated by radar) as they fall through the
known
temperature and humidity profiles. A
surface
precipitation type field is computed using surface precipitation
observations
as additional information. Along with this step various other
parameters are
derived, including heights of the bases and tops and the
three-dimensional
cloud type. From the type and depth of each cloud layer, an appropriate
vertical motion profile is determined for each vertical column
containing
clouds (Schultz and Albers 2001).
2.2 Balance
Scheme
The second
major improvement has been to the LAPS dynamical balance package (Shaw
et al.
2001). The balance package is run using the initial analysis of the
atmospheric
state variables after the cloud analysis has completed. The purpose of
the
balance package is to ensure that the final mass and momentum fields
are
consistent with the cloud and precipitation fields. This is the crucial
component of the analysis, since attempts by others to directly
initialize
clouds have met with limited success (Cram et al. 1994, Raymond 1995).
The
usual result is a rapid dissipation of the clouds within the first few
time
steps of model integration due to lack of such a balance. The scheme
employs
several dynamical constraints as well as a diabatic term within a
three-dimensional, variational formulation to adjust the wind,
temperature, and
height fields based on the background vertical velocity field and the
diagnosed
cloud motions from the cloud analysis. During the minimization of the
variational cost function in this step, the time tendencies of the u
and
v wind components are also minimized, which results in a very
stable
initial condition such that the numerical forecast model is not
"shocked" at the initial time and the cloud field is maintained.
2.3 Model
testing
For several
months beginning in the fall of 2000, we have been using the improved
LAPS
analyses to diabatically initialize the NCAR/PSU MM5 forecast model for
a
domain centered over Colorado and Wyoming with a grid-spacing of 10 km.
These
simulations are run four times per day in realtime on FSL's massively
parallel
High Performance Computing System (HPCS). Products from these runs are
posted
on our web site (http://laps.fsl.noaa.gov).
The gridded fields are also provided to the Denver-Boulder National
Weather
Service (NWS) Forecast Office (WFO) for display on AWIPS, and have
proven to be
a valuable source of information during the preparation of operational
public
forecasts for their area of responsibility (Shaw and Thaler 2001).
3
CASE STUDY
The case study employed
version 3
of the NCAR/PSU MM5 model. A grid
spacing of 10 km was used for a 125 by 105 grid point horizontal domain
covering all of Colorado and Wyoming, eastern Utah, western Kansas and
Nebraska, and the northern fringes of Arizona and New Mexico. The vertical grid consists of 41 levels,
with the highest resolution contained within the boundary layer. The Schultz (1995) explicit microphysics and
the Kain-Fritsch convective parameterization were employed. The Rapid Radiative Transfer Model (RRTM)
scheme was used as the longwave radiation package, and the Blackadar
scheme was
used for the PBL parameterizations.
We have used this
initialization technique as part of a case
study of a severe weather event that occurred in eastern Colorado on 20
July
2000. Figures 1-6 show a comparison between observed low-level
reflectivity and
simulated reflectivity from two different MM5 forecasts valid at the
initial
time (1800 UTC) and 4 h into the simulation (2200 UTC). One of the
simulations
("MM5HOT") was initialized using the technique described above and
the other ("MM5ETA") was initialized using the 00 h forecast from the
1800 UTC cycle of the NCEP Eta model. The 1800 UTC cycle of the Eta
provided
lateral boundary conditions for both simulations. Since the MM5HOT
simulation
had the hydrometeor fields provided by LAPS, the diagnosed reflectivity
pattern
matches the observed reflectivity almost identically. Furthermore, by
2200 UTC
(Fig. 2), the major area of convection initially in southern Nebraska
at 1800
UTC was generally maintained and moved southward into eastern Colorado
and
western Kansas, where numerous reports of hail and tornadoes occurred
during
the next few hours. Fig. 2 shows additional convection not produced by
the
model (Fig. 5), however, the model did simulate convection in the
regions that
did receive severe weather. Additionally, new convection developed in
the
MM5HOT simulation in eastern Wyoming, consistent with the observed
reflectivity. In contrast, the MM5ETA simulation had very limited
reflectivity
fields by this point. The MM5ETA method of initialization, which does
not
include any hydrometeor fields (no “hot-start”) or the addition of any
new
observation data (no LAPS), is the typical configuration used by many
real-time
users of mesoscale models.
Fig.
1. LAPS analyzed reflectivity
(verification) 1800 UTC (initial time)
Fig. 2.
LAPS reflectivity at 2200 UTC, 4 h into the forecast.
Fig.
3. MM5eta 4 h into the forecast just
beginning to indicate convection in
central KS.
Fig.
4. MM5HOT at initial time showing
initialized presence of convection; compare to Fig. 1.
Fig. 5
MM5HOT at 4 h into the forecast.
H H
Fig. 6.
Location of severe reports in the Colorado eastern plains and Kansas
2100-2300
UTC. (H=hail, T=tornado, F=flood)
4. Summary
and recommendations
This work
has promising implications for short-range NWP forecasts. We
acknowledge this
is only one case and do not claim the spinup problem has been solved by
this
hot start method. However, it is a
promising start to the types of methods that may be very useful in
advancing
the use of mesocale modeling for short-range QPF and severe weather
forecasting. Testing with our real-time
runs continues,
and we anticipate interesting results during the 2001 convective
weather
season. Thus far, the experimental runs are having a positive impact on
the
operations at the Denver-Boulder WFO.
Also, such a system will be ported to three NWS WFOs located in
the
southern region later this year, as well as at the two USAF spacelift
facilities at Patrick AFB, Florida, and Vandenberg AFB, California.
Future work
includes continued enhancements to the balance package and cloud
analysis,
improved verification methods, coupling to other forecast models, and
the use
of higher resolution grids.
5. REFERENCES
Albers,
S., J. McGinley, D. Birkenheuer, and J. Smart 1996: The Local Analysis
and
Prediction System (LAPS): Analyses of clouds, precipitation, and
temperature. Wea.
and Forecast, 11, 273-287.
Cram,
J., and S. Albers, 1994: The use of meso-beta scale analyzed cloud
cover to
initialize a numerical model. Sixth Conf. on Mesoscale Processes,
Portland, OR, Amer. Meteor. Soc., 130-133.
McGinley,
J. A., S. Albers, and P. Stamus, 1991: Validation of a composite convective
index as defined by a real-time
local analysis system. Wea.
Forecasting, 6, 337-356.
_______,
_______, D. Birkenheuer, B. Shaw, P.
Schultz, 2000: The LAPS water in all
phases analysis: the approach and
impacts on numerical prediction. 5th International
Symposium on
Tropospheric Profiling, Adelaide, Australia, Amer. Meteor. Soc.,
133-135.
Raymond,
W.H., W.S. Olson, and G. Callan,
1995: Diabatic forcing and
initialization with assimilation of cloud water and rainwater in a
forecast
model, Mon. Wea. Rev., 46, 8-68.
Schultz, P. and S. Albers, 2001: The use of
three-dimensional analyses of cloud attributes for diabatic
initialization of
mesoscale models. 14th Conf. on Numerical Weather Prediction,
Ft.
Lauderdale, FL, Amer. Meteor. Soc. (this volume)
_______, 1995: An explicit cloud physics
parameterization
for operational numerical weather prediction.
Mon. Wea. Rev., 123, 3331-3343.
Shaw, B.L.
and E. Thaler, 2001: Operational
evaluation of the LAPS-MM5 "hot start" local forecast model. 18th
Conference on Weather Analysis and Forecasting, Fort
Lauderdale, FL Amer. Meteor. Soc. (this
volume).
_______,
J.A. McGinley, and P.Schultz, 2001:
Explicit inialization of clouds and precipitation in mesoscale forecast
models.
14th Conference in Numerical Weather Prediction, Fort
Lauderdale, FL Amer. Meteor. Soc. (this volume). .
Snook J. S.,
P. A. Stamus, J. Edwards, Z.
Christidis, and J. A. McGinley, 1998: Local-domain mesoscale analysis
and
forecast model support for the 1996 Summer Olympic Games.
Wea. Forecasting, 13,
138-150.
Szoke, Ed,,
D. Barjenbruch, and E. Thaler, 2000:
An assessment of the utility of a local model for operational mountain
snowfall
predictions. 9th Conf. On Mountain Meteorology.
Amer. Meteor.
Soc., 322-326.
_______, and
A. Marroquin, 2000: Some Remarkable
Supercell Simulations from a
Quasi-operational Local-Scale Model – Skill or Shear Luck? 20th Conference on Severe Local Storms,
Orlando FL Amer. Meteor. Soc., 555-558.