Presentation

Report
Weather Impacts Decision Aids (WIDA) Workshop
15 March 2012, Reno, NV
Improving the Characterization of the
Environment for AREPS Electromagnetic
Performance Predictions
Paul Frederickson
Department of Meteorology, Naval Postgraduate School
Monterey, CA
OBJECTIVE
More accurate Advanced Refractive Effects Prediction System (AREPS)
electromagnetic (EM) propagation predictions through improved characterization of the
environment, to be accomplished through the following tasks:
1) Incorporate a new Navy Atmospheric Vertical Surface Layer Model (NAVSLaM) version
with improved performance in stable conditions into AREPS.
2) Incorporate the NPS vertical refractivity profile blending algorithm into AREPS.
3) Replace the old Historical Electromagnetic Propagation Conditions (HEPC) datasets
with vastly improved NPS EM climatology datasets & methods within AREPS.
4) Improved access to and use of COAMPS® data by AREPS.
Requirements and Capabilities
More accurate Battlespace on Demand (BonD) Tier 2 EM system performance
TDAs and guidance products being provided to the Navy and DoD warfighter for:
• Radar detection of periscopes, missiles, low-fliers, small boats, and other targets.
• Radar counter-detection of our own submarines, NSW platforms and other platforms.
• Effectiveness of electronic attack, surveillance, communications, any propagation
operations.
Performers
 Naval Postgraduate School (NPS):
Paul Frederickson, Peter Guest, Tom Murphree
 Naval Research Laboratory, Monterey (NRL-MRY):
John Cook, Michael Frost and Tracy Haack
 SPAWAR Systems Center, Pacific (SSC-Pac):
Amalia Barrios, Gary Lindem and others
Impact of Different Ducting Types on EM Propagation
Different duct types have distinct impacts on EM propagation near the surface which are
highly dependent upon the duct attributes (duct height, trapping layer thickness and
strength, etc.), so we have to get the environment right!
Propagation loss coverage diagram examples are for a 6 GHz radar at typical shipboard height above the surface.
Impact of the Evaporation Duct Height on
Radar Detection Ranges
• Examples shown are for radars of different frequencies at a typical shipboard mast
height and for small and medium-sized surface targets.
• The impact of the evaporation duct height on radar detection ranges can be very strong,
depending upon the radar frequency and height and the target type, demonstrating that
it is critical to correctly take the evaporation duct into account.
Note the different range scales in these two figures.
Incorporating the new Navy Atmospheric Vertical
Surface-Layer Model (NAVSLaM) version with
improved performance in stable conditions into
AREPS
NAVSLaM Performance in Stable Conditions
• NAVSLaM is highly dependent upon the empirically-determined Monin-Obukhov
Similarity Theory dimensionless profile functions in stable conditions (ASTD > 0).
• NAVSLaM is much less sensitive to input parameters and has a much wider region of
applicability over which EDH can be defined in stable conditions when using the two
new candidate functions, especially the Grachev et al (2007) functions on the right.
Evaporation Duct Height (m)
Existing NAVSLaM Function
Candidate NAVSLaM Functions
Greatly reduced
sensitivity & wider
model applicability
High sensitivity to
input data and
limited model
applicability
Air-Sea Temp Difference (°C)
Air-Sea Temp Difference (°C)
Air-Sea Temp Difference (°C)
Validating the New NAVSLaM Model Versions
Validate the new NAVSLaM versions and compare with the old, using concurrent
met & propagation measurements from the Wallops Island, VA, experiment of 2000.
X Band
S&C
Bands
NPS Buoy
(Met data for input
to NAVSLaM)
MPMS Shore
Station
(Receivers)
NSWC-DD EM prop
data in S/C/X bands
(J. Stapleton et al)
R/V Sealion
(Transmitters in S,
C & X bands)
NAVSLaM Model Validation: Strongly Stable Case
• Moderately stable with low humidity: ASTD = 3.2 °C, RH = 62%, WS = 5.3 m/s.
• Current NAVSLaM using Beljaars & Holtslag functions cannot define the EDH.
• NAVSLaM with Cheng & Brutsaert & Grachev produce high & very different EDH’s.
Current
NAVSLaM
cannot define
the EDH
EDH = 63.7 m (209 ft)
EDH = 46.3 m (152 ft)
Modified
refractivity
profiles were
computed by
NAVSLaM using
different Ψ
functions and
then input to
APM for
propagation
predictions.
NAVSLaM Model Validation: Strongly Stable Case
• S-band example
• ASTD = 3.2 °C, RH = 62%, WS = 5.3 m/s.
• Current NAVSLaM cannot define the
EDH, so no propagation loss results
could be produced by APM.
• New NAVSLaM with Grachev et al
functions produces best agreement
with propagation measurements.
• Further NAVSLaM validation is
continuing.
• Results so far indicate the new
NAVSLaM version should replace the
old in OAML as soon as practicable.
Will begin this process later this FY.
Propagation Observations
NAVSLaM (Cheng & Brutsaert)
NAVSLaM (Grachev et al)
Incorporating the NPS Vertical Refractivity Profile
Blending Algorithm into AREPS
Blending Upper-Air & Surface-Layer Refractivity Profiles
• NAVSLaM is used to most accurately characterize the evaporation duct using COAMPS®
or climatological near-surface data as input.
• The NAVSLaM surface-layer refractivity profile must be realistically blended with the
COAMPS® or climatology upper-air profiles for accurate EM propagation predictions.
NAVSLaM
Blended
Height
COAMPS®
or climo
Height
+
=
Note that
the evap
duct ‘fills in’
the skip zone
Example propagation loss coverage diagrams are for a 6 GHz radar at a typical shipboard height above the surface.
Existing AREPS Profile Blending Versus NPS Blending
• Currently when using COAMPS® data AREPS either ignores the evaporation duct or
appends a Paulus-Jeske evaporation duct profile below the lowest model level, which
creates both an unrealistic ‘kink’ in the refractivity profile and an artificial evap duct.
EDH = 55 m
EDH = 33 m
(Note ‘kink’)
NPS Vertical Refractivity Profile Blending Algorithm
Blends NAVSLaM surface-layer modified
refractivity (M) profile smoothly onto the
bottom of a COAMPS® or climatological
upper-air refractivity profile.
Height Above Surface
COAMPS®
NAVSLaM
Blended
NPS Blending Method
• Algorithm smoothly blends profiles
together and does not introduce
erroneous refractivity features (‘kinks’).
• Blending interval heights chosen so as
COAMPS M
slope
not to distort either the NAVSLaM or
COAMPS® significant refractivity
features, as much as possible.
• Below blending interval, blended profile
Blending
Interval
NAVSLaM
M slope
shape is equal to NAVSLaM profile.
• Within blending interval, blended
profile smoothly transitions from slope
of NAVSLaM profile at bottom of
interval, and slope of COAMPS® profile
at top of interval.
• Above blending interval, blended profile
is equal to COAMPS® profile.
Modified Refractivity (M)
NPS Vertical Refractivity Profile Blending Algorithm
Common Ocean Unstable Case
• Choose point farther offshore of Virginia
with unstable conditions.
Coastal-Influenced Stable Case
• Choose point near shore with offshore
advection of warm air and stable
conditions.
NPS Vertical Refractivity Profile Blending Algorithm
Common Ocean Unstable Case
• Near-standard COAMPS® profile above sfc.
• Blended between 35-50 m, well above EDH
(18 m), so no distortion of EDH possible.
• Very smooth transition between profiles.
Coastal-Influenced Stable Case
• Blended profile departs dramatically from
COAMPS® if same blending interval is used.
• COAMPS® has full 3-D physics for such
situations, so preserve COAMPS® profile.
COAMPS
NAVSLaM
NPS Blended
Blending Interval
EDH
If same blending
Interval is used
NPS Vertical Refractivity Profile Blending Algorithm
Common Ocean Unstable Case
• Near-standard COAMPS® profile above sfc.
• Blended between 35-50 m, well above EDH
(18 m), so no distortion of EDH possible.
• Very smooth transition between profiles.
Coastal-Influenced Stable Case
• Use blending interval that smoothes nearsurface portion of profile and also
preserves the COAMPS® deep surfacebased trapping layer.
COAMPS
NAVSLaM
NPS Blended
Blending Interval
EDH
Blending Interval
Interactive Blending Capability for AREPS?
• NPS refractivity profile blending algorithm will be incorporated into AREPS to
automatically blend upper-air and surface-layer refractivity profiles.
• NPS has developed an interactive GUI which can be incorporated into AREPS to allow the
user to examine and change, if necessary, the vertical blending intervals to best fit the
specific upper-air and surface-layer refractivity conditions.
}
}
Selected Blending Interval
The user can easily
change the vertical
blending levels to best
fit the specific situation
Can AG’s be trained
to do this accurately
and quickly?
Incorporating the New NPS EM Climatology Data
Sets into AREPS to Replace HEPC
Evaporation Duct Climatology
Upper-Air Duct Climatology
Comparison of HEPC & New NPS Evap Duct Climatology
Existing HEPC Climatology Now in AREPS
New NPS Climatology Going into AREPS
• Based on ICOADS (mainly volunteer observing
ships) data for 1970-84.
• Based on NCEP CFSR Reanalysis data for 19792010, results in much better statistics.
• Computed by outdated Paulus-Jeske ED model
which over-predicts EDH in most situations.
• Computed by modern NAVSLaM ED model
with much improved performance.
• Very coarse 10° x 10° Marsden Square averages
for spatial resolution and day/night only
temporal resolution.
• ~0.3125° (~35 km) grid spatial resolution and
hourly temporal resolution, important for
resolving coastal regions and diurnal variations.
HEPC does not
provide map views.
This figure was
produced by NPS
for comparison
purposes only.
EDH = 17.9 m
EDH = 13.8 m
Comparison of Impact of HEPC & NPS Evap Duct Climos
• Scenario is a typical shipboard X-band search radar looking for a small near-surface
target in the East China Sea in October at 12Z.
• New NPS ED climatology results in a predicted radar detection range of 14.4 nmi.
• Old HEPC ED climatology results in a predicted radar detection range of 6.3 nmi.
8.1 nmi
New NPS climatology
results in a 8.1 nmi
greater detection range
than HEPC for this Xband example.
Evaporation Duct Climatology Propagation Predictions
• User selects grid point, month and time of day of interest to obtain an EDH histogram and
‘climatological’ refractivity profiles for 5th, 10th, 25th, 50th, 75th, 90th or 95th EDH percentiles.
• AREPS has only a ‘mean’ refractivity profile, with an EDH that does not match HEPC.
• Climatological refractivity profiles are input to AREPS for EM performance predictions.
User clicks on grid
point to create
histogram and
AREPS ‘ENV’ input
file for EM system
performance
predictions
AREPS
ENV File
Radar Probability
of Detection
Median EDH = 13.8 m
EDH = 13.8 m
Refractivity
Profile
EDH Histogram
22
NPS Upper-Air Ducting Climatology
• The NPS Upper-Air Ducting Climatology was computed from the Integrated Global
Radiosonde Archive (IGRA) data base using data from 1971 to 2010.
• IGRA radiosonde stations with sufficient number of soundings with valid refractivity
data are shown for the northern Indian Ocean and western North Pacific Ocean.
Compare the NPS and HEPC
upper-air ducting
climatologies for the Kuwait
Airport to demonstrate
their differences in the
following slides.
Atmospheric Duct Types and EM Propagation Impacts
• The old HEPC upper-air ducting climatology now in AREPS does not distinguish
surface & surface-based ducts and combines them together into the same statistics.
• The NPS upper-air ducting climatology treats all these distinct duct types separately.
AREPS
combines
these,
NPS treats
separately
Propagation loss coverage diagram examples are for a 6 GHz radar at typical shipboard height above the surface.
Comparison of NPS & HEPC Upper-Air Ducting Climatologies
• HEPC data from 1966-69 & 1973-74; NPS data from 1971-2010: much better statistics.
Kuwait example: HEPC 7-58 raobs versus NPS 650-842 raobs per month/launch time.
• HEPC greatly overestimates occurrence of surface-based ducts due to combining duct types.
Kuwait example for April 12Z: HEPC 38% occurrence versus NPS 0.9% occurrence.
• HEPC greatly underestimates surface-based duct top heights and thicknesses.
Kuwait example for October: HEPC 36 m versus NPS 273-280 m.
• HEPC combines 00Z/12Z soundings in statistics; NPS gives separate statistics for 00Z & 12Z.
• HEPC only provides medians; NPS provides 5, 10, 25, 50, 75, 90 & 95th percentiles.
Old HEPC Climatology
New NPS Climatology
Comparison of HEPC & NPS Upper-Air Climo’s
• Scenario is a shipboard X-band search radar looking for a small surface target off the
coast of Kuwait in October at 12Z.
• New NPS UA climatology results in a predicted radar detection range of 8.7 nmi.
• Old HEPC UA climatology results in a predicted radar detection range of 52.1 nmi.
New NPS climatology
results in a 43.4 nmi
lower detection range.
Improved Access to and Use of COAMPS®
Predicted Data by AREPS
COAMPS-OS® Interface from NEP-Oc
(NRL-MRY improving this part)
HPAC/AREPS
AREPS Environment Creator Window
for Ingesting COAMPS® Data
(SSC-Pac improving this part)
Interpolation of COAMPS® Refractivity Profiles
• Refractivity profiles are originally computed for the COAMPS® model grid,
which is defined according to a regular spacing on the particular map
projection being used for the model run.
Interpolation of COAMPS® Refractivity Profiles
• In the HPAC data files refractivity profiles are interpolated from the
COAMPS® model grid to a regular lat-lon grid.
Interpolation of COAMPS® Refractivity Profiles
• Refractivity profiles are interpolated again within AREPS to form a rangedependent refractivity environment along a specified path. Refractivity
profiles therefore currently undergo two interpolations before being used
for EM propagation calculations in AREPS.
Impact of Interpolating Refractivity Profiles
• In this idealized example to illustrate the impacts of interpolating
refractivity profiles, a trapping layer is increasing with height between two
adjacent COAMPS® model grid points.
Trapping layer
increasing in
height between
grid points
Impact of Interpolating M Profiles
• Currently, HPAC files are constructed by interpolating vertical refractivity
profiles between grid points by COAMPS® height levels.
• This can have the undesirable effect of ‘smoothing’ trapping layers, which
can very adversely affect EM propagation predictions.
The current interpolation
scheme results in
unrealistic propagation
predictions
Top & bottom of
trapping layer
are ‘smoothed’
by interpolation
Impact of Interpolating M Profiles
• A more realistic method of interpolating refractivity profiles would be to
interpolate by duct attributes (i.e. duct height, duct thickness, etc.).
• This results in a much more realistic refractivity profile and EM propagation
prediction, as shown in green below.
New interpolation
results in much
more realistic
propagation loss
predictions
Trapping layer not
smoothed with
interpolation by
duct attributes
Solution for Providing COAMPS® Data to AREPS
• New solution is to perform no interpolation of refractivity profiles when
constructing range-dependent refractivity environment from gridded COAMPS®
data.
• COAMPS® profile data will be
provided from FNMOC to AREPS
users in the native COAMPS grid
(i.e. no interpolation to a regular
lat-lon grid as done now).
NRL-MRY (Cook, Frost, Haack).
• When specifying a rangedependent refractivity
environment, AREPS will use the M
profile from the closest raw
COAMPS® model grid-point within
a specified distance threshold and
apply it to the nearest point along
the propagation path.
SSC-Pac (Barrios)
COAMPS® native grid points
Propagation path being modeled by AREPS
Refractivity Structure Matching Algorithm (RSMA)
• The Advanced Propagation Model (APM) within AREPS requires refractivity profile
information at each Parabolic Equation (PE) range step (approximately 270-900 m).
• Currently, refractivity profiles are linearly interpolated by height level indices in APM (see
left panel below), which distorts and ‘smooth’s out important refractivity features.
• NPS, SSC-Pac and NRL-MRY are proposing to jointly develop a new Refractivity Structure
Matching Algorithm (RSMA) in FY13 which will linearly interpolate refractivity profiles by
their trapping-layer features, as shown in red on right.
No RSMA: Trapping layers are distorted
and ‘smoothed out’ by interpolation.
RSMA: Trapping layers vary linearly with
range and sharp edges are preserved.
Summary
1) The new Navy Atmospheric Vertical Surface Layer Model (NAVSLaM) version
with much improved performance in stable conditions is being incorporated
into AREPS now. New model version will be submitted to OAML this FY.
2) Beta-version of NPS vertical refractivity profile blending algorithm code is
being incorporated into AREPS while the algorithm undergoes further
refinement, validation and testing.
3) New and vastly improved NPS upper-air and evaporation duct climatological
datasets and methods are being incorporated into AREPS. These datasets
will also be submitted to OAML this FY.
4) New methods of providing COAMPS® data to AREPS and for using data within
AREPS are being developed by NRL-Monterey and SSC-Pac to correct
interpolation problems and enable use of higher resolution data.
5) NPS, SSC-Pac and NRL-MRY have jointly proposed to develop a new
Refractivity Structure Matching Algorithm (RSMA) to correct the last
refractivity profile interpolation problem within APM.
Questions?
Summary
1)
A new version of the Navy Atmospheric Vertical Surface Layer Model (NAVSLaM) has
been developed with much improved performance in stable conditions. The new
model code has been provided to SSC-Pacific for incorporation into AREPS, while the
algorithm undergoes further validation. The process of submitting the new model
version into OAML will begin this FY.
2)
A beta-version of the NPS vertical refractivity profile blending algorithm code has
been provided to SSC-Pacific for incorporation into AREPS, while the algorithm
undergoes further validation and testing. The process of submitting this algorithm
into OAML will begin this FY or next depending upon progress.
3)
New NPS upper-air and evaporation duct climatological datasets and tools are being
produced, and sample datasets have been provided to SSC-Pacific for incorporation
into AREPS. Final datasets and tools will be provided to SSC-Pacific by early May. The
process of submitting these datasets into OAML will begin this FY.
4)
A new method, data format and interface for providing COAMPS® data to AREPS is
being developed by NRL-Monterey and new methods of using COAMPS® data in
AREPS are being developed that will correct current interpolation problem and
enable use of higher resolution data than before. NPS, SSC-Pac and NRL-MRY have
jointly proposed to develop a new Refractivity Structure Matching Algorithm to
correct the last refractivity profile interpolation problem.
Evap Duct Characterization from COAMPS® Data
• Raw COAMPS® profile including sea surface temp has unrealistic ‘kink’
• COAMPS® profile ignoring sea surface temp has no evaporation duct
• Smooth profile appended below lowest COAMPS® model level
• NAVSLaM profile computed from lowest level COAMPS data and sea surface
temperature - must be smoothly blended onto COAMPS upper-air profile.
NAVSLaM profile from
COAMPS data is most
realistic representation
of the evaporation
duct.
Builder Refractivity Profile Blending Algorithm
• Builder blending algorithm arbitrarily blends the NPS surface-layer model m profile
onto bottom of NWP model profile between the two lowest vertical NWP model
levels (in this case between 10 and 30 m) for every case.
• This arbitrary method has the impact of distorting the evaporation duct height and
surface-layer model refractivity profile with very adverse consequences on the
resulting propagation calculations.
Blended between
10 and 30 m
EDH = 17 m
EDH = 13 m
Calculated propagation loss for
Builder method is under-predicted
due to evaporation duct being overpredicted.
AREPS Refractivity Profile Appending Algorithm
• AREPS simply appends a Paulus-Jeske model m profile onto the bottom of the NWP
model profile at the lowest NWP model level.
• This has the impact of creating a ‘kink’ in the m profile and will in many cases ‘cap’
the evaporation duct height at the lowest NWP model level height, both of which
have adverse impacts on propagation calculations.
PJ model profile
appended at
lowest NWP
model level
Calculated propagation loss for
AREPS method is under-predicted in
this case probably due to sharper
trapping layer.
Vertical Refractivity Profile Blending Algorithm Validation
COAMPS Grid Points:
– 3 km
– 9 km
– 27 km
• Use RED 2001 experiment data to
validate and compare refractivity
profile blending methods.
• Near-surface propagation loss
measured over a 25.8 km (13.9 nmi)
path by SSC-Pacific.
• Used COAMPS data from 3 km grid
point nearest the mid-point of the
propagation path.
• Ran NAVSLaM model with lowest
COAMPS level data.
• Used different blending algorithms on
refractivity profiles.
• Ran APM with all resulting modified
refractivity profiles to estimate
propagation loss.
• Compare predicted prop loss versus
observed prop loss for all cases.
• All cases unstable and dominated by
the evaporation duct.
Refractivity Profile Blending Algorithm Validation
• NAVSLaM only & NPS blending method are virtually identical & perform well.
• Builder method over-estimates EDH and under-estimates propagation loss.
• AREPS and ‘smooth’ append methods under-estimate EDH & over-estimate
propagation loss, though AREPS performs fairly well for this specific case.
• This has the impact of creating a ‘kink’ in the m profile and will in many cases ‘cap’
the evaporation duct height at the lowest NWP model level height, both of which
have adverse impacts on propagation calculations.
Vertical Refractivity Profile Blending Algorithm Validation
Validation with RED Experiment Propagation Data
• New NPS blending method performs best.
• Builder method blends across the EDH and distorts the EDH, resulting in prop
loss being greatly under-predicted.
• AREPS method over-predicts prop loss, though purely by chance it performs
fairly well in this specific case.
• COAMPS® with a smooth surface-layer profile appended greatly over-predicts
prop loss.
Blending Method
New NPS method
Builder method
Current AREPS method
COAMPS + smooth sfc layer
Bias
-1.1 dB
-4.2 dB
+1.5 dB
+6.8 dB
RMS error
3.5 dB
5.6 dB
3.8 dB
7.4 dB
Refractivity Profile Blending Algorithm Validation
RED Experiment
• NPS method
performs best.
• Builder method
under-predicts
prop loss.
• AREPS method
over-predicts
prop loss, though
purely by chance
it performs fairly
well in this
specific case.
• Smooth append
method greatly
over-predicts
prop loss.
Method
NPS
Builder
AREPS
Smooth
Bias
-1.1 dB
-4.2 dB
+1.5 dB
+6.8 dB
RMS error
3.5 dB
5.6 dB
3.8 dB
7.4 dB
Vertical Refractivity Profile Blending Algorithm Validation
• Blended versus NAVSLaM only prop loss scatter plots for different scenarios with
radar at typical ship mast height (20 m) and target at 2 m above surface.
30 km
40 km
6 GHz
9 GHz
20 km
Adverse impact of using AREPS or Builder
methods becomes more apparent at higher
frequencies and longer ranges.
3 GHz
RED Experiment
Impacts less apparent
50 km
60 km
Comparison of NPS & HEPC Upper-Air Ducting Climatologies
Number of Sounding Observations for Kuwait
• HEPC data from 1966-69 & 1973-74, number of observations per month: 7 to 58.
• NPS data from 1971-present, number of observations per month: 650 to 842.
• Much better statistics possible with NPS data set with many more observations.
• Improved representation of the current global climate with more recent data.
Old HEPC Climatology
New NPS Climatology
Comparison of NPS & HEPC Upper-Air Ducting Climatologies
Frequency of Occurrence of Surface-Based Ducts for Kuwait
• HEPC incorrectly classifies surface ducts together with surface-based ducts.
• Occurrence of surface-based ducts greatly overstated in HEPC in most cases.
• For April at 12Z: HEPS has 38% occurrence of SBD’s in HEPC, versus 0.9% in NPS.
• Very unrealistic distribution of frequency of occurrence in HEPC.
Old HEPC Climatology
New NPS Climatology
Comparison of NPS & HEPC Upper-Air Ducting Climatologies
Duct Top Height of Surface-Based Ducts for Kuwait
• HEPC greatly understates surface-based duct top heights and thicknesses.
• For October: HEPC duct top height is 36 m; NPS duct top heights are 273-280 m.
• HEPC combines 00Z/12Z soundings; NPS gives separate results for 00Z and 12Z.
• HEPC only shows medians; NPS shows (5, 10, 25, 50, 75, 90 & 95th percentiles).
Old HEPC Climatology
New NPS Climatology
NPS Upper-Air Ducting Climatology Map-Views
Prototype of Upper-Air EM Ducting Climatology GUI
Summary information
for selected station
Station information
shown by color-coded
station dots (number of
valid soundings in this
example).
Click on
station of
interest
}
Show climatological
ducting figures for
selected station and
duct type/attribute.
Create Refractivity
Profile env file for input
to AREPS
Requirement for Using NWP model Data
The U.S. Navy will increasingly rely upon NWP model data, such as from
COAMPS®, to characterize the refractivity environment for EM system
performance predictions due to the discontinuation of operational
radiosondes and the lack of adequate shipboard surface observations to
model the evaporation duct.
NWP models also have several potential advantages over single point
observations:
• NWP models can provide information on spatial variations in refractivity
conditions along a propagation path.
• NWP models can provide forecasts of refractivity conditions, which is
important for operational planning.
• NWP models can potentially provide useful refractivity information for
areas in which there are no actual observations available, including denied
areas where observations are not possible.
Vertical Resolution Requirements for COAMPS
Prediction of Mesoscale Refractivity
Tracy Haack, Naval Research Laboratory, Monterey
Num. Avg. z
Levels (1km)
85
23 m
70
59
40
76
30
96
Line
Color
black
blue
green
red
Duct/No Duct Event Statistics
500
450
Height (m)
Wallops 2000
Experiment
Num. Levels 30 40 70 85
Event Freq(%):
62 62 62 62
Percent Correct(%):
50 47 53 60
Mean Squ Error(%):
49 53 47 40
Hit Rate(%): 37 30 34 49
Miss Rate(%):
62 70 66 51
False Alm Rate(%):
27 25 18 22
Correct Null Rate(%):
72 75 82 78
Discrim Score(%):
9 5 16 26
400
350
300
250
OBS
MEAN
(-320)
200
Series13
Series20
Series19
Series15
Series14
Series18
Series17
Series12
Series9
Series8
Series10
OBSMEAN
CMPMEAN
BIAS
RMSE
RCOR
Series7
150
100
50
0
-10
0
10
20
30
BIAS RMSE
RCOR
COAMPS MEAN (-320)
40
50
60
Increased vertical resolution has a strong impact
on the frequency of ducting and improves overall
Modified Ref (M-units)
ducting event statistics.
The most significant changes to the vertical structure of the boundary layer occurred for average vertical
grid spacing of ~60 m or better as evidenced by mean profile of modified refractivity.
70
Horizontal Resolution Requirements for COAMPS
Sub Normal Super Trap
Prediction of Mesoscale Refractivity
Grid 1
Tracy Haack, Naval Research Laboratory, Monterey
Duct Event Contingency Table
Horiz. Grid Spacing 36 12 4 1.33 km
Correct (%)
61 70 74 67
Mean Squ Error (%) 39 30 26 33
Hit (%)
43 69 70 63
Miss (%)
59 31 30 37
Correct Null (%)
94 72 81 74
Obs-Model
False Alarm (%)
6 28 19 26
DiscriminationDucting
Score: 37
40 51 37
Strength
Grid 2 (9km)
Grid 2
30
Grid1
Grid2
Grid3
Grid4
HELO
Duct Strength (M-units)
25
Grid 3
Grid 4 (1.33 km)
A
20
15
Grid 4
10
5
B
0
0
20
28Apr
40
29Apr
60
80
30Apr
100
120
140
1May 2May3May
Duct Strength Time series
160
180
4May
Horizontal Distribution
A
185 km
Higher horizontal resolution produced stronger MABL inversions increasing duct strength and occurrence.
The 4-km grid out-performed all the other grids, including Grid 4 (1.33km).
B
Improved Interface between AREPS and COAMPS
John Cook, Naval Research Laboratory, Monterey
• Joint project between NRL
Monterey, NPS, and SSC-Pacific
to improve the interface
between AREPS and COAMPS®
• Optimize model output
parameters and levels
provided for AREPS
• Eliminate interpolation for subgrids
• Support new evaporation duct
and profile blending algorithms
• Efficient use of
communications bandwidth
using GRIB
• Effective for COAMPS-OS®,
METCAST, and CAGIPS
Current Use of COAMPS® Data with AREPS
• COAMPS® gridded data files can
currently be downloaded from
FNMOC COAMPS-OS® application
for use as environmental input to
AREPS.
• Hazard Prediction and
Assessment Capability (HPAC)
data file format was developed
for hazardous material release
modeling, not EM prediction.
• HPAC used as a quick stop-gap
solution, though not ideal for
propagation modeling.
• These HPAC data files are
available for multiple regions
around the globe where
COAMPS® is being run by
request.
Summary
SSC-Pacific (Amalia Barrios), NPS (Paul Frederickson) and NRL-Monterey (John Cook
and Tracy Haack) are collaborating in FY12 to modernize and improve AREPS and its
environmental input data sources and correct many known deficiencies. These
improvements include:
1)
The NPS vertical refractivity profile blending algorithm, which has been shown to
perform much better than the current AREPS and Builder methods, will be
incorporated into AREPS for use with COAMPS® and climatology data.
2)
A new NAVSLaM model version with improved performance in stable conditions will
be incorporated into AREPS.
3)
The new NPS upper-air and evaporation duct climatological datasets, displays and
analysis methods will be incorporated into AREPS to replace the badly out-of-date
HEPC datasets.
4)
A new method and interface for providing COAMPS® data to AREPS is being
developed by NRL-Monterey that will correct known problems in interpolating
refractivity profiles and have other advantages.
5)
New AREPS methods of using COAMPS® data will also correct current interpolation
problem and enable use of higher resolution data than before.
NAVSLaM Model Validation: Strongly Stable Case
The existing NAVSLaM version
cannot define the EDH and
therefore cannot be used by APM
for propagation calculations.
Proposed Improvements to NAVSLaM
• Since the Navy Atmospheric Vertical Surface-Layer Model (NAVSLaM) was
proposed for inclusion to OAML, significant new research has appeared on
surface parameterizations and bulk surface-layer scaling.
• NAVSLaM followed most of the TOGA-COARE bulk model surface
parameterizations and a new TOGA-COARE version has since been released.
We propose to incorporate the latest TOGA-COARE parameterizations into
NAVSLaM.
• New dimensionless profile functions have been proposed for stable
conditions (air warmer than the sea), for which NAVSLaM is highly sensitive
to its input parameters.
• NPS has recently developed new refractivity equations for visible to midwave infrared wavelengths. We propose to incorporate these functions into
NAVSLaM so that it can also be used in EO models.
• These improvements will require the new model version to be submitted to
OAML, but should not require a full CIMREP process.
Data Set FY11 Accomplishments & FY12-13 Plans
• Evaporation duct and upper-air ducting datasets transitioned in FY11.
• Evaporation duct and upper-air ducting datasets to be transitioned in FY 12 or 13.
• Low-res evap duct dataset may be developed for the entire globe in FY13.
• Datasets will be incorporated into the online Advanced Climate Analysis &
Forecasting (ACAF) application on FNMOC website and also into AREPS (FY12).
FY11
FY12-13
* FY12-13 regions under discussion and exact boundaries are still to be determined.
Comparison of Existing and NPS EDH Climatologies
The difference in predicted radar detection ranges between the new NPS EM climatology
and the existing Navy HEPC climatology can be very large, and is a strong function of
radar frequency & antenna height. Example for AUTEC range.
NPS Climatology (EDH = 12.5 m)
HEPC Climatology (EDH = 15.9 m)
Standard Atmosphere
60
NPS & PJ Modeled Propagation Loss vs Measurements
• Roughness & Evaporation Duct (RED) Experiment conducted off Oahu, 2001.
• S-band radar transmitter at 5 meters above mean sea level, receiver at 6 m above
mean sea level, all over-ocean propagation path of 28 km.
• NPS & PJ models produced m profiles from buoy data, which are input to APM.
• All data were obtained in unstable conditions (air cooler than sea surface).
• NPS model agrees much better with prop measurements than PJ model.
• PJ model over-predicts EDH and as a result under-predicts S-band prop loss.
61
Merging Evaporation Duct & Upper-Air Ducting
Climatologies
• Upper-air profiles should only be merged with climatological evaporation duct profiles for low
elevation coastal and island radiosonde stations representative of ocean conditions.
• A nearby ocean grid point should be selected for merging the evaporation duct profile with the
upper-air dataset and both profiles should be valid for close to the same time of day.

similar documents