The Effects of Automation on Expertise

possehastyMechanics

Nov 5, 2013 (3 years and 10 months ago)

64 views

Dual
-
Polarization Operations
Assessment

Status Update on Development Process


Progress on RDA work


Progress on RPG work

Where we are with Beta and Deployment

RDA Work


Data Quality Team


Many, many, many problems identified,
researched and fixed


Training, engineering and research experts
who meet regularly






Beta test will be key…


Currently Scheduled for
Dec 2010


March
2011

What Dual
-
Pol

Does
Not

Change


Basic WSR
-
88D characteristics


Frequency,
beamwidth
, data resolution


VCP sampling


Elevations, update time, number of pulses per
radial


All Pre
-
Dual
-
Pol

products and algorithms


You

control how and when
you

integrate
Dual
-
Pol

information into
your

methodology

Dual
-
Pol

and CMD


CMD will
not

be operational with initial
deployment


Manual clutter management returns!


Bypass Maps for normal clutter


All Bins for AP clutter



CMD will return with the first full build
after Dual
-
Pol


Build 13…
scheduled

Jan/Feb 2012


Development work underway…


Dual
-
Pol

and Sensitivity


Sensitivity: ability to detect weak returns


Measured as ___
dBZ

@ ___ range for returned
signal = noise


Affected by transmitted power and system
noise


If returned signal is above the noise, life is good!

Dual
-
Pol

Sensitivity Loss


Dual
-
Pol

radars transmit and receive both
horizontally (H) and vertically (V)


For same range, same target, WSR
-
88D Dual
-
Pol

has
less

returned power than before

Dual
-
Pol

and Sensitivity


Dual
-
Pol

sensitivity loss means more gates
that are “Off” than before


Returned signal
not

processed into base data

Detectable

Not Detectable

Dual
-
Pol

Sensitivity Loss


A bit more loss from hardware that does
the splitting


Cabling, built
-
in test equipment, etc.







Per site expected sensitivity loss due to
Dual
-
Pol

conversion 3.5


4 dB



Impact on operations?!!?

Dual
-
Pol

Sensitivity Loss

Impact on Operations?


SME panel assembled to evaluate

impact of 4 dB loss


ROC engineering, NWS SSDs, NSSL, and WDTB



Spectrum of weather events evaluated


4 dB loss simulated (Don’t try this at
home)

Dual
-
Pol

Sensitivity Loss

Impact on Operations


Events evaluated


Dryline



KS (KDDC)


Freezing drizzle


KS (KICT)


Marginal lake effect snow


NY (KBUF)


Tornadic

supercell

along boundary


CO (KFTG)


Outflow boundaries from convection


OK (KTLX)


Short range smoke plumes


OK (KTLX)


Multiple outflow boundaries/popcorn showers


FL (KEVX)


California wildfires


CA (KSOX)


Hurricane Ike landfall


TX (KHGX)


Nor’easter snowstorm (
snowmaggedon
)


VA (KLWX)



Go Paul!!

Now For Calibration of Z


Calibration is about the
magnitude

of Z values

KCRI

0.5˚

KOUN

0.5˚

Calibration of Z


Given appropriate Z calibration and same
scatterers

in a volume


40
dBZ

is 40
dBZ
, 30
dBZ

is 30
dBZ
, etc.


Assuming Rayleigh scattering…




Work is ongoing


Z calibration


ZDR calibration

Sensitivity and Calibration

Summary


Sensitivity: more Off gates with Dual
-
Pol


Per site expected sensitivity loss due to Dual
-
Pol

conversion 3.5


4 dB


Less return from very weak signal areas



Z Calibration: magnitude that is assigned



Dual
Pol

calibration much more complex


Be kind to your techs!


8 hour check no longer transparent

Now For Attenuation


Impact on base moments does not change


Impact on Dual
-
Pol

products is more
significant

KOUN: Dual
-
Pol


KCRI: Single
Pol

Attenuation and Dual
Pol

Products


What has happened to ZDR?


What are the consequences?

Attenuation and Dual
Pol

Products


What has happened to CC?


What are the consequences?

RPG Work


Dual
-
Pol

algorithms: lots of room to grow*


Knowledge base (research) central Oklahoma


Base data from RDA
only recently
mature
enough for RPG testing


Greatest emphasis now is QPE

MLDA

HCA

*Been there, done that: WSR
-
88D original development

Quantitative Precipitation
Estimation Algorithm


Benefits of Dual
-
Pol

applied to rainfall
estimation


Dual
-
Pol

better at identifying ground clutter,
biological
scatterers
, large hail, etc.


Prevent these returns from being converted to
rainfall


Legacy Precipitation Processing System
and products do
not

go away

QPE Basics


QPE is an RPG algorithm


QPE & PPS have similarities & differences


QPE supported by other Dual
-
Pol

algorithms


Hydrometeor Classification Algorithm (HCA)


Melting Layer Detection Algorithm (MLDA)


QPE Similarities With Legacy PPS


Accumulations start and stop based
on coverage and intensity


Actual

start and stop times
will

differ

QPE Similarities With Legacy PPS


Exclusion zones


Prevent returns (e.g. moving clutter) from

being converted to rainfall


Zones defined separately for PPS and QPE


Recommend same zones initially

QPE Similarities With Legacy PPS


Storm total accumulations
reset at RPG

QPE Similarities With Legacy PPS


Concept of Hybrid Scan same


Find lowest unblocked elevation of valid
precipitable

return


Common accumulation periods for PPS
and QPE products


One hour, storm total, user defined

QPE Differences From Legacy PPS


HCA and MLDA input used to build

QPE Hybrid Scan


Prevent non
-
meteorological returns (e.g.
clutter, biological) from conversion to rain rate


Determine best rain rate computation to use

How QPE Builds Hybrid Scan


Each bin gets
Hydroclass

value from HCA


Biological

Clutter

Rain

Heavy Rain

Ice Crystals

Wet Snow

Big Drops

Graupel

Hail
-
Rain

Unknown

Dry Snow

How QPE Builds Hybrid Scan


GC (clutter), UK (
unkown
)


Go up next tilt


BI (biological), no echoes


Don’t search higher and set rain rate to 0.0


Remaining
Hydroclass

values are
precipitable


Lots

of details on this one…some other time!



QPE Differences From Legacy PPS

QPE

Hybrid Scan
Hydroclass

R(Z)

R(Z,Z
DR
)

R(K
DP
)

PPS

R(Z)

Hybrid Scan
Reflectivity


QPE rain rate product

every volume scan

Why is QPE Work Going On NOW?


Base data from RDA only recently (May!)
mature enough for RPG testing


But wait! There’s more! LNAs switched

out in July





RDA work requires frequent shutdowns


QPE and PPS both require


High quality base data


Appropriate clutter filtering


RDA stable and calibrated

QPE is an
Algorithm


Base Data Rules!

QPE Work Currently Underway


Attenuation and QPE


Easy to identify in base data

QPE Work Currently Underway


QPE can show zero rain rates


Attenuation correction for DP variables being
worked

QPE Work Currently Underway


R(KDP) performance looks good so far


R(Z,ZDR) performance


Significant underestimation of high efficiency
(warm rain) rainfall


Underestimation similar to R(Z)

QPE

Hybrid Scan
Hydroclass

R(Z)

R(Z,Z
DR
)

R(K
DP
)

QPE Work Currently Underway


R(Z,ZDR) performance


Warm rain event, flash flooding in OKC


We were lucky to discover this (warm rain in OK and
KOUN was running)

Gage: 10.21

QPE: 4.40

PPS: 4.81

QPE Work Currently Underway


Current version of R(Z,ZDR) based on
large group of central OK rain events


“Individual cases may vary”


Solution requirement: no operator

parameter changes


Thank You


You are an important part of this process