CASA’sIm
pact on Radar-Based W
ind Assessm
ents
Don J. Rude*, Ellen J. Bass
*and Brenda Philips**
*University of Virginia, **University of Massachusetts
This work is supported primarily by the Engineering Research Centers Program of the National Science Foundation under NSF award number
0313747. Any opinions, findings, conclusions, or recommendationsexpressed in this m
aterial are those of the authors and do not necessarily
reflect those of the National Science Foundation.
DISCUSSIO
N
Sin
ce w
ind
sp
eed
pla
ys
a cr
itic
al
role
in t
he d
efin
itio
n o
f se
ver
e
thu
nd
erst
orm
s, t
he
pu
rpo
se o
f th
is s
tud
y w
as
to m
easu
re t
he
imp
act
of
CA
SA
rad
ar
dat
a o
n w
ind
ass
essm
ent
per
form
ance
.
The
resu
lts
rep
ort
ed h
ere
in c
lear
ly i
nd
icat
e th
at f
or
win
d s
pee
d
asse
ssm
ents
CA
SA
rad
ar d
ata
can s
imu
ltan
eousl
y
•in
crea
se w
ind
sp
eed
est
imat
es
•re
duce
ass
essm
ent
erro
r
•in
crea
se c
on
fid
ence
Th
is o
utc
om
e is
par
ticu
larl
y p
rom
isin
g g
iven t
hat
•d
ata
are
fro
m e
xp
erim
en
tal
rad
ars
•p
arti
cip
an
ts w
ere
giv
en m
inim
al
train
ing
•th
e ra
dar
dis
pla
y a
nd
co
mp
ute
r-in
terf
ace
wer
e u
nfa
mil
iar
•m
an
y p
arti
cip
an
ts s
aid
CA
SA
dat
a co
nfi
rms
their
menta
l m
od
el
The
incre
ase
in w
ind
sp
eed
ass
essm
ents
su
pp
ort
s re
sear
ch b
y B
row
n
and
Wo
od
(B
row
n e
t al.
, 2
00
5)
wh
ich p
red
icts
that
incre
ased
sp
atia
l
sam
pli
ng r
esu
lts
in h
igher
rad
ial
velo
cit
y r
ead
ings.
The
shif
t in
war
nin
g
dec
isio
ns
when g
iven C
AS
A d
ata
is e
spec
iall
y n
ota
ble
bec
ause
on
ly
one
exp
erim
en
tal
task
was
co
ver
ed b
y a
n a
ctual
war
nin
g a
cco
rdin
gto
NW
S a
rch
ives
.
As
syst
em
s su
ch a
s C
AS
A,
NE
XR
AD
Sup
erR
es,
and
MP
AR
(Hein
selm
an,
Pri
eg
nit
z, M
anro
ss,
Sm
ith,
& A
dam
s, 2
00
8)
com
e o
n-
line,
the
fore
cast
ing c
om
mu
nit
y m
ay n
eed
to
rev
isit
po
licie
s and
thre
sho
lds
for
issu
ing w
arn
ing
s.
Future W
ork
The
curr
ent
stud
y c
ou
ld b
e en
hance
d b
y t
he s
yst
em
ati
c co
ntr
ol
of
rad
ar
bea
m a
ttri
bute
s. U
pd
ate
rate
, b
eam
heig
ht,
win
d-t
o-b
eam
in
ters
ecti
on
ang
le,
and
sam
pli
ng f
ideli
ty e
ach i
nfl
uence
per
form
ance
. F
utu
re w
ork
cou
ld q
uanti
fy t
he
imp
act
of
thes
e at
trib
ute
s in
div
iduall
y.
Ad
dit
ional
mea
sure
s o
f p
erfo
rmance
co
uld
be
use
d i
nclu
din
g t
he
size
of
war
nin
gs,
their
du
rati
on,
and
lea
d t
ime.
Fin
all
y,
a st
and
ard
ized
and
rig
oro
us
train
ing t
oo
l fo
r C
AS
A d
ata
wo
uld
help
rem
ove
an
y n
ovelt
y e
ffec
ts
fro
m f
utu
re s
tud
ies.
METHODOLOGY
Introduction
Th
e C
ente
r fo
r C
oll
abo
rati
ve
Ad
apti
ve
Sen
sin
g o
f th
e A
tmosp
her
e (C
AS
A)
is
crea
tin
g a
new
par
adig
m f
or
radar
sys
tem
s b
ased
on
lo
w-c
ost
net
wo
rks
of
X-
Ban
d r
adar
s th
at c
oll
ect
hig
h s
pat
ial
and
tem
po
ral
reso
luti
on d
ata
fro
m t
he
low
er
tro
posp
her
e. C
AS
A o
per
ates
a 4
no
de
radar
tes
t bed
in
so
uth
-wes
t O
kla
ho
ma
to
rese
arch
the
cap
abil
itie
s o
f th
is n
ew t
echn
olo
gy
and
to
und
erst
and
ho
w C
AS
A
dat
a w
ill
imp
act
dec
isio
n m
akin
g b
y u
sers
su
ch a
s N
atio
nal
Wea
ther
Ser
vic
e
fore
cast
ers.
We
hyp
oth
esiz
e th
at C
AS
A t
echn
olo
gy w
ill
bet
ter
info
rm
fore
cast
ers,
im
pro
ve
thei
r per
form
ance
, an
d u
ltim
atel
y im
pro
ve
sever
e w
eath
er
war
nin
gs.
Th
is s
tud
y w
as d
esig
ned
to
mea
sure
the
impac
t o
f C
AS
A d
ata
on
win
d
spee
d a
sses
smen
ts, a
crit
ical
com
po
nen
t o
f th
e se
ver
e th
und
erst
orm
war
nin
g
pro
cess
.
Experim
ental Task
Fo
reca
ster
s p
red
icte
d w
ind
s sp
eed a
t th
e gro
un
d, 2
to
5 m
inu
tes
into
the
futu
re a
t
spec
ific
lo
cati
ons
usi
ng o
nly
NE
XR
AD
dat
a o
r C
AS
A a
nd
NE
XR
AD
dat
a
com
bin
ed.
Cas
e st
ud
ies
wer
e cr
eate
d f
rom
dat
a co
llec
ted
in
th
e O
kla
hom
a te
st
bed
. G
roun
d s
enso
r re
adin
gs
fro
m t
he
Okla
ho
ma
Mes
onet
(McP
her
son
et
al.,
200
7)
pro
vid
ed g
roun
d t
ruth
. T
his
gro
un
d t
ruth
was
use
d t
o c
alcu
late
ab
solu
te
win
d s
pee
d a
sses
smen
t er
ror.
Six
sce
nar
ios,
eac
h a
pp
roxim
atel
y 1
2 m
inu
te l
on
g,
wer
e ch
ose
n f
rom
May
7,
200
8, w
her
e a
cold
-co
re s
yste
m p
rod
uce
d s
mal
l bo
win
g c
ells
and
su
per
cell
s.
Par
tici
pan
ts m
ade
pre
dic
tion
s b
ased
on
th
ese
dat
a. T
he
scen
ario
sh
ad s
imil
ar b
ut
dis
cret
e st
orm
cel
ls p
rodu
cin
g s
trai
gh
t li
ne
win
ds
in t
he
20
-50
kt
ran
ge
at
loca
tio
ns
wit
h a
deq
uat
e C
AS
A a
nd
NE
XR
AD
cover
age.
Participants
Six
teen
Nat
ional
Wea
ther
Ser
vic
e fo
reca
ster
s (1
2 m
ale,
4 f
emal
e)w
ith
oper
atio
nal
exp
erie
nce
ran
gin
g f
rom
5 t
o 2
5 y
ears
(M
= 1
4.4
yea
rs, SD
= 5
.9)
par
tici
pat
ed i
n t
he
exp
erim
ent
at t
he
33
rd A
nnu
al m
eeti
ng o
f th
eN
atio
nal
Wea
ther
Ass
oci
atio
n i
n L
ou
isvil
le,
KY
, O
ctob
er 1
1-1
6, 2
00
8.
Independent Variables
•Weather Scenario.
Wea
ther
sce
nar
io r
efer
s to
the
6 s
elec
ted
atm
osp
her
ic
situ
atio
ns
in t
ime
ord
er, 1
st t
o 6
th.
•Data Source.
Par
tici
pan
ts e
val
uat
ed w
eath
er s
cenar
ios
wit
h o
nly
NE
XR
AD
dat
a o
r b
oth
CA
SA
an
d N
EX
RA
D d
ata.
Dat
a so
urc
e in
dic
ates
if
rad
ar d
ata
wer
e
supp
lied
fro
m “
NE
XR
AD
On
ly”
or
“CA
SA
& N
EX
RA
D”.
Example W
eather Scenario (#5)
The
two
im
ages
in t
he
top
ro
w a
re f
rom
NE
XR
AD
’sK
FD
R r
adar
at
21
:58
and
22
:02
UT
C.
Bea
m h
eig
ht
6.4
kft
@ 6
5 n
mi
range.
The
images
in t
he
bo
tto
m r
ow
are
fro
m C
AS
A’s
KS
AO
rad
ar a
t
21
:58
, 2
1:5
9,
22
:00
, 2
2:0
1,
and
22
:02
UT
C.
Bea
m h
eig
ht
0.4
kft
@
2 n
mi
ran
ge.
At
22
:05
UT
C s
enso
rs a
t th
e C
HIC
Mes
onet
stat
ion
rep
ort
ed a
47
kt
max w
ind
sp
eed
.
Wind Speed Assessment
The
mea
n w
ind
sp
eed
ass
essm
ents
wer
e st
atis
tica
lly h
igh
er w
hen
bo
th C
AS
A a
nd
NE
XR
AD
dat
a w
ere
avail
ab
le a
s o
pp
ose
d t
o
when o
nly
NE
XR
AD
dat
a w
ere
avail
able
(F
(1,2
8)=
15
.26
4,
p =
0.0
01
).
•M
ean w
ith C
AS
A &
NE
XR
AD
: 4
0.1
kt
•M
ean w
ith N
EX
RA
D O
nly
: 3
3.9
kt
Absolute W
ind Speed Assessment Error
The
mea
n o
f th
e ab
solu
te v
alu
es o
f th
e w
ind
sp
eed
ass
essm
ent
erro
rs w
ere
stat
isti
call
y s
mall
er
when b
oth
CA
SA
and
NE
XR
AD
dat
a w
ere
avail
able
as
op
po
sed
to
when
on
ly N
EX
RA
D d
ata
wer
e av
ail
able
(F
(1,2
8)=
7.3
09
, p
= 0
.01
2).
•M
ean w
ith C
AS
A &
NE
XR
AD
: 7
.9 k
t
•M
ean w
ith
NE
XR
AD
On
ly:
11
.5 k
t
Confidence Ratings
05
10
15
20
25
1-Not
confident
23
45
67-Very
Confident
CASA & NEXRAD
NEXRAD Only
Assessment Confidence
Fo
reca
ster
co
nfi
dence
ass
essm
en
ts w
ere
rank
ed
stat
isti
call
y h
igher
when b
oth
CA
SA
and
NE
XR
AD
dat
a w
ere
avail
able
as o
pp
ose
d t
o w
hen
on
ly N
EX
RA
D d
ata
wer
e av
ail
able
(χ2
= 1
6.0
30
,
df=
1,
p <
0.0
00
5).
•M
od
e w
ith C
AS
A &
NE
XR
AD
: 5
•M
od
e w
ith N
EX
RA
D O
nly
: 4
RESULTS
Dependent Variables
•Wind Speed Assessment.
Win
d s
pee
d a
sses
smen
t is
th
e gro
un
d l
evel
win
d s
pee
d f
ore
cast
ed f
or
the
targ
et l
oca
tion
by
the
par
tici
pan
tto
th
e
nea
rest
1 k
no
t.
•Absolute Wind Speed Assessment Error.
The
abso
lute
val
ue
of
the
dif
fere
nce
bet
wee
n t
he
win
d s
pee
d a
sses
smen
t an
d t
he
auto
mat
ed g
roun
d
sen
sor
read
ing.
•Assessment Confidence
. A
fter
pro
vid
ing t
hei
r w
ind
spee
d a
sses
smen
t
par
tici
pan
ts w
ere
asked
“h
ow
co
nfi
den
t ar
e yo
u i
n t
his
est
imat
e”on
a
scal
e fr
om
1-“
No
t C
on
fid
ent”
to 7
-“V
ery
Co
nfi
den
t”.
•Warning Decision.
Aft
er p
rovid
ing t
hei
r co
nfi
den
ce r
atin
g p
arti
cip
ants
wer
e as
ked
“d
o t
hes
e ra
dar
bas
ed w
ind
s in
dic
ate
a w
arn
ing i
s nee
ded
”
and
a “
Yes
” o
r “N
o”
resp
on
se w
as r
eco
rded
.
REFERENCES
Bro
wn,
R.
A.,
Fli
ckin
ger
, B
. A
., F
orr
en,
E.,
Schult
z,
D.
M.,
Sir
mans,
D.,
Spen
cer,
P.
L.,
et
al.
(2
00
5).
Im
pro
ved D
etec
tio
n o
f S
ever
e S
torm
s U
sin
g E
xper
imenta
l
Fin
e-R
esolu
tio
n W
SR
-88
D M
easu
rem
ents
. Weather and Forecasting
, 20
(1),
3-1
4.
Hei
nse
lman,
P.
L.,
Pri
eg
nit
z,
D.
L.,
Manro
ss,
K.
L., S
mit
h,
T.
M.,
& A
dam
s, R
. W
.
(20
08).
Rap
id S
am
pli
ng o
f S
ev
ere
Sto
rms
by t
he
Nati
onal
Wea
ther
Radar
Tes
tbed
Phase
d A
rray R
adar.
Weather and Forecasting
, 23
(5),
80
8-8
24
.
McP
her
son,
R.
A.,
Fie
bri
ch,
C.
A.,
Cra
wfo
rd,
K.
C.,
Ell
iott
, R
. L
., K
ilb
y,
J. R
.,
Gri
msl
ey,
D.
L.,
et
al.
(2
00
7).
Sta
tew
ide
Mo
nit
ori
ng o
f th
e M
esosc
ale
Envir
on
ment:
A T
ech
nic
al
Up
date
on t
he
Okla
ho
ma M
eso
net
. Journal of
Atmospheric and Oceanic Technology,
24
(3),
30
1-3
21.
Warning Decision
The
pro
po
rtio
n o
f d
ecis
ions
“Y
es”
to w
arn w
as s
tati
stic
all
y
hig
her
when b
oth
CA
SA
and
NE
XR
AD
dat
a w
ere
avai
lab
le a
s
op
po
sed
to
when o
nly
NE
XR
AD
dat
a w
ere a
vail
able
(χ2
=
7.2
51
, d
f=1
, p
= 0
.00
7).
•W
ith C
AS
A &
NE
XR
AD
: 3
0 o
f 4
7
•W
ith N
EX
RA
D O
nly
: 6
of
48
Sce
nari
o 5
wit
h N
EX
RA
D O
nly
had
all
“N
o”
resp
onse
s (8
)
wher
eas
CA
SA
& N
EX
RA
D h
ad a
ll “
Yes
” re
spo
nse
s (7
) and
one
mis
sin
g r
esp
onse
.
ACKNOWLEDGMENTS
We
thank
the
16
fo
reca
ster
s w
ho
vo
lun
teer
ed t
o p
arti
cip
ate
in
the
stud
y d
escr
ibed
her
ein
. A
lso
, w
e th
ank
Pat
rick
Mar
sh a
nd
Ro
n P
rzyb
yli
nsk
ifo
r th
eir
fo
reca
stin
g i
nsi
ghts
, an
d t
he
NW
A
for
ho
stin
g u
s at
their
an
nual
mee
tin
g.