<s>
Numerical	B-General_Concept
weather	I-General_Concept
prediction	I-General_Concept
(	O
NWP	O
)	O
uses	O
mathematical	O
models	O
of	O
the	O
atmosphere	O
and	O
oceans	O
to	O
predict	O
the	O
weather	O
based	O
on	O
current	O
weather	O
conditions	O
.	O
</s>
<s>
Though	O
first	O
attempted	O
in	O
the	O
1920s	O
,	O
it	O
was	O
not	O
until	O
the	O
advent	O
of	O
computer	B-Application
simulation	I-Application
in	O
the	O
1950s	O
that	O
numerical	B-General_Concept
weather	I-General_Concept
predictions	I-General_Concept
produced	O
realistic	O
results	O
.	O
</s>
<s>
The	O
improvements	O
made	O
to	O
regional	O
models	O
have	O
allowed	O
for	O
significant	O
improvements	O
in	O
tropical	B-Application
cyclone	I-Application
track	I-Application
and	O
air	O
quality	O
forecasts	O
;	O
however	O
,	O
atmospheric	O
models	O
perform	O
poorly	O
at	O
handling	O
processes	O
that	O
occur	O
in	O
a	O
relatively	O
constricted	O
area	O
,	O
such	O
as	O
wildfires	O
.	O
</s>
<s>
Manipulating	O
the	O
vast	O
datasets	O
and	O
performing	O
the	O
complex	O
calculations	O
necessary	O
to	O
modern	O
numerical	B-General_Concept
weather	I-General_Concept
prediction	I-General_Concept
requires	O
some	O
of	O
the	O
most	O
powerful	O
supercomputers	B-Architecture
in	O
the	O
world	O
.	O
</s>
<s>
Even	O
with	O
the	O
increasing	O
power	O
of	O
supercomputers	B-Architecture
,	O
the	O
forecast	O
skill	O
of	O
numerical	B-General_Concept
weather	I-General_Concept
models	I-General_Concept
extends	O
to	O
only	O
about	O
six	O
days	O
.	O
</s>
<s>
Factors	O
affecting	O
the	O
accuracy	O
of	O
numerical	O
predictions	O
include	O
the	O
density	O
and	O
quality	O
of	O
observations	O
used	O
as	O
input	O
to	O
the	O
forecasts	O
,	O
along	O
with	O
deficiencies	O
in	O
the	O
numerical	B-Application
models	I-Application
themselves	O
.	O
</s>
<s>
In	O
addition	O
,	O
the	O
partial	O
differential	O
equations	O
used	O
in	O
the	O
model	O
need	O
to	O
be	O
supplemented	O
with	O
parameterizations	B-Application
for	O
solar	O
radiation	O
,	O
moist	O
processes	O
(	O
clouds	O
and	O
precipitation	O
)	O
,	O
heat	O
exchange	O
,	O
soil	O
,	O
vegetation	O
,	O
surface	O
water	O
,	O
and	O
the	O
effects	O
of	O
terrain	O
.	O
</s>
<s>
In	O
an	O
effort	O
to	O
quantify	O
the	O
large	O
amount	O
of	O
inherent	O
uncertainty	O
remaining	O
in	O
numerical	O
predictions	O
,	O
ensemble	B-Application
forecasts	I-Application
have	O
been	O
used	O
since	O
the	O
1990s	O
to	O
help	O
gauge	O
the	O
confidence	O
in	O
the	O
forecast	O
,	O
and	O
to	O
obtain	O
useful	O
results	O
farther	O
into	O
the	O
future	O
than	O
otherwise	O
possible	O
.	O
</s>
<s>
The	O
history	B-Application
of	I-Application
numerical	I-Application
weather	I-Application
prediction	I-Application
began	O
in	O
the	O
1920s	O
through	O
the	O
efforts	O
of	O
Lewis	O
Fry	O
Richardson	O
,	O
who	O
used	O
procedures	O
originally	O
developed	O
by	O
Vilhelm	O
Bjerknes	O
to	O
produce	O
by	O
hand	O
a	O
six-hour	O
forecast	O
for	O
the	O
state	O
of	O
the	O
atmosphere	O
over	O
two	O
points	O
in	O
central	O
Europe	O
,	O
taking	O
at	O
least	O
six	O
weeks	O
to	O
do	O
so	O
.	O
</s>
<s>
It	O
was	O
not	O
until	O
the	O
advent	O
of	O
the	O
computer	O
and	O
computer	B-Application
simulations	I-Application
that	O
computation	O
time	O
was	O
reduced	O
to	O
less	O
than	O
the	O
forecast	O
period	O
itself	O
.	O
</s>
<s>
The	O
ENIAC	B-Device
was	O
used	O
to	O
create	O
the	O
first	O
weather	O
forecasts	O
via	O
computer	O
in	O
1950	O
,	O
based	O
on	O
a	O
highly	O
simplified	O
approximation	O
to	O
the	O
atmospheric	O
governing	O
equations	O
.	O
</s>
<s>
Operational	O
numerical	B-General_Concept
weather	I-General_Concept
prediction	I-General_Concept
in	O
the	O
United	O
States	O
began	O
in	O
1955	O
under	O
the	O
Joint	O
Numerical	B-General_Concept
Weather	I-General_Concept
Prediction	I-General_Concept
Unit	O
(	O
JNWPU	O
)	O
,	O
a	O
joint	O
project	O
by	O
the	O
U.S.	O
Air	O
Force	O
,	O
Navy	O
and	O
Weather	O
Bureau	O
.	O
</s>
<s>
In	O
1956	O
,	O
Norman	O
Phillips	O
developed	O
a	O
mathematical	O
model	O
which	O
could	O
realistically	O
depict	O
monthly	O
and	O
seasonal	O
patterns	O
in	O
the	O
troposphere	O
;	O
this	O
became	O
the	O
first	O
successful	O
climate	B-Application
model	I-Application
.	O
</s>
<s>
Following	O
Phillips	O
 '	O
work	O
,	O
several	O
groups	O
began	O
working	O
to	O
create	O
general	B-Application
circulation	I-Application
models	I-Application
.	O
</s>
<s>
The	O
first	O
general	O
circulation	O
climate	B-Application
model	I-Application
that	O
combined	O
both	O
oceanic	O
and	O
atmospheric	O
processes	O
was	O
developed	O
in	O
the	O
late	O
1960s	O
at	O
the	O
NOAA	O
Geophysical	O
Fluid	O
Dynamics	O
Laboratory	O
.	O
</s>
<s>
These	O
newer	O
models	O
include	O
more	O
physical	O
processes	O
in	O
the	O
simplifications	O
of	O
the	O
equations	O
of	O
motion	O
in	O
numerical	B-Application
simulations	I-Application
of	O
the	O
atmosphere	O
.	O
</s>
<s>
As	O
such	O
,	O
a	O
statistical	O
relationship	O
between	O
the	O
output	O
of	O
a	O
numerical	B-General_Concept
weather	I-General_Concept
model	I-General_Concept
and	O
the	O
ensuing	O
conditions	O
at	O
the	O
ground	O
was	O
developed	O
in	O
the	O
1970s	O
and	O
1980s	O
,	O
known	O
as	O
model	O
output	O
statistics	O
(	O
MOS	O
)	O
.	O
</s>
<s>
Starting	O
in	O
the	O
1990s	O
,	O
model	O
ensemble	B-Application
forecasts	I-Application
have	O
been	O
used	O
to	O
help	O
define	O
the	O
forecast	O
uncertainty	O
and	O
to	O
extend	O
the	O
window	O
in	O
which	O
numerical	B-General_Concept
weather	I-General_Concept
forecasting	I-General_Concept
is	O
viable	O
farther	O
into	O
the	O
future	O
than	O
otherwise	O
possible	O
.	O
</s>
<s>
As	O
such	O
,	O
the	O
idea	O
of	O
numerical	B-General_Concept
weather	I-General_Concept
prediction	I-General_Concept
is	O
to	O
sample	O
the	O
state	O
of	O
the	O
fluid	O
at	O
a	O
given	O
time	O
and	O
use	O
the	O
equations	O
of	O
fluid	O
dynamics	O
and	O
thermodynamics	O
to	O
estimate	O
the	O
state	O
of	O
the	O
fluid	O
at	O
some	O
time	O
in	O
the	O
future	O
.	O
</s>
<s>
Stations	O
either	O
report	O
hourly	O
in	O
METAR	O
reports	O
,	O
or	O
every	O
six	O
hours	O
in	O
SYNOP	B-Application
reports	O
.	O
</s>
<s>
A	O
variety	O
of	O
methods	O
are	O
used	O
to	O
gather	O
observational	O
data	O
for	O
use	O
in	O
numerical	B-Application
models	I-Application
.	O
</s>
<s>
Different	O
models	O
use	O
different	O
solution	O
methods	O
:	O
some	O
global	O
models	O
and	O
almost	O
all	O
regional	O
models	O
use	O
finite	B-Algorithm
difference	I-Algorithm
methods	I-Algorithm
for	O
all	O
three	O
spatial	O
dimensions	O
,	O
while	O
other	O
global	O
models	O
and	O
a	O
few	O
regional	O
models	O
use	O
spectral	B-Algorithm
methods	I-Algorithm
for	O
the	O
horizontal	O
dimensions	O
and	O
finite-difference	B-Algorithm
methods	I-Algorithm
in	O
the	O
vertical	O
.	O
</s>
<s>
The	O
length	O
of	O
the	O
time	O
step	O
chosen	O
within	O
the	O
model	O
is	O
related	O
to	O
the	O
distance	O
between	O
the	O
points	O
on	O
the	O
computational	O
grid	O
,	O
and	O
is	O
chosen	O
to	O
maintain	O
numerical	B-Algorithm
stability	I-Algorithm
.	O
</s>
<s>
The	O
UKMET	O
Unified	B-Application
Model	I-Application
is	O
run	O
six	O
days	O
into	O
the	O
future	O
,	O
while	O
the	O
European	O
Centre	O
for	O
Medium-Range	O
Weather	O
Forecasts	O
 '	O
Integrated	B-Application
Forecast	I-Application
System	I-Application
and	O
Environment	O
Canada	O
's	O
Global	B-Application
Environmental	I-Application
Multiscale	I-Application
Model	I-Application
both	O
run	O
out	O
to	O
ten	O
days	O
into	O
the	O
future	O
,	O
and	O
the	O
Global	O
Forecast	O
System	O
model	O
run	O
by	O
the	O
Environmental	B-Application
Modeling	I-Application
Center	I-Application
is	O
run	O
sixteen	O
days	O
into	O
the	O
future	O
.	O
</s>
<s>
Some	O
meteorological	O
processes	O
are	O
too	O
small-scale	O
or	O
too	O
complex	O
to	O
be	O
explicitly	O
included	O
in	O
numerical	B-General_Concept
weather	I-General_Concept
prediction	I-General_Concept
models	O
.	O
</s>
<s>
For	O
example	O
,	O
the	O
gridboxes	O
in	O
weather	O
and	O
climate	B-Application
models	I-Application
have	O
sides	O
that	O
are	O
between	O
and	O
in	O
length	O
.	O
</s>
<s>
Weather	B-General_Concept
models	I-General_Concept
that	O
have	O
gridboxes	O
with	O
sizes	O
between	O
can	O
explicitly	O
represent	O
convective	O
clouds	O
,	O
although	O
they	O
need	O
to	O
parameterize	O
cloud	O
microphysics	O
which	O
occur	O
at	O
a	O
smaller	O
scale	O
.	O
</s>
<s>
The	O
cloud	B-Application
fraction	I-Application
can	O
be	O
related	O
to	O
this	O
critical	O
value	O
of	O
relative	O
humidity	O
.	O
</s>
<s>
Within	O
air	O
quality	O
models	O
,	O
parameterizations	B-Application
take	O
into	O
account	O
atmospheric	O
emissions	O
from	O
multiple	O
relatively	O
tiny	O
sources	O
(	O
e.g.	O
</s>
<s>
The	O
horizontal	O
domain	B-Algorithm
of	I-Algorithm
a	I-Algorithm
model	I-Algorithm
is	O
either	O
global	O
,	O
covering	O
the	O
entire	O
Earth	O
,	O
or	O
regional	O
,	O
covering	O
only	O
part	O
of	O
the	O
Earth	O
.	O
</s>
<s>
High-resolution	O
models	O
—	O
also	O
called	O
mesoscale	O
models	O
—	O
such	O
as	O
the	O
Weather	B-Application
Research	I-Application
and	I-Application
Forecasting	I-Application
model	I-Application
tend	O
to	O
use	O
normalized	O
pressure	O
coordinates	O
referred	O
to	O
as	O
sigma	O
coordinates	O
.	O
</s>
<s>
Statistical	O
models	O
were	O
created	O
based	O
upon	O
the	O
three-dimensional	O
fields	O
produced	O
by	O
numerical	B-General_Concept
weather	I-General_Concept
models	I-General_Concept
,	O
surface	O
observations	O
and	O
the	O
climatological	O
conditions	O
for	O
specific	O
locations	O
.	O
</s>
<s>
Model	O
output	O
statistics	O
differ	O
from	O
the	O
perfect	O
prog	O
technique	O
,	O
which	O
assumes	O
that	O
the	O
output	O
of	O
numerical	B-General_Concept
weather	I-General_Concept
prediction	I-General_Concept
guidance	O
is	O
perfect	O
.	O
</s>
<s>
Extremely	O
small	O
errors	O
in	O
temperature	O
,	O
winds	O
,	O
or	O
other	O
initial	O
inputs	O
given	O
to	O
numerical	B-Application
models	I-Application
will	O
amplify	O
and	O
double	O
every	O
five	O
days	O
,	O
making	O
it	O
impossible	O
for	O
long-range	O
forecasts	O
—	O
those	O
made	O
more	O
than	O
two	O
weeks	O
in	O
advance	O
—	O
to	O
predict	O
the	O
state	O
of	O
the	O
atmosphere	O
with	O
any	O
degree	O
of	O
forecast	O
skill	O
.	O
</s>
<s>
While	O
a	O
set	O
of	O
equations	O
,	O
known	O
as	O
the	O
Liouville	O
equations	O
,	O
exists	O
to	O
determine	O
the	O
initial	O
uncertainty	O
in	O
the	O
model	O
initialization	O
,	O
the	O
equations	O
are	O
too	O
complex	O
to	O
run	O
in	O
real-time	O
,	O
even	O
with	O
the	O
use	O
of	O
supercomputers	B-Architecture
.	O
</s>
<s>
Edward	O
Epstein	O
recognized	O
in	O
1969	O
that	O
the	O
atmosphere	O
could	O
not	O
be	O
completely	O
described	O
with	O
a	O
single	O
forecast	O
run	O
due	O
to	O
inherent	O
uncertainty	O
,	O
and	O
proposed	O
using	O
an	O
ensemble	O
of	O
stochastic	O
Monte	B-Algorithm
Carlo	I-Algorithm
simulations	I-Algorithm
to	O
produce	O
means	O
and	O
variances	O
for	O
the	O
state	O
of	O
the	O
atmosphere	O
.	O
</s>
<s>
Since	O
the	O
1990s	O
,	O
ensemble	B-Application
forecasts	I-Application
have	O
been	O
used	O
operationally	O
(	O
as	O
routine	O
forecasts	O
)	O
to	O
account	O
for	O
the	O
stochastic	O
nature	O
of	O
weather	O
processes	O
that	O
is	O
,	O
to	O
resolve	O
their	O
inherent	O
uncertainty	O
.	O
</s>
<s>
This	O
method	O
involves	O
analyzing	O
multiple	O
forecasts	O
created	O
with	O
an	O
individual	O
forecast	O
model	O
by	O
using	O
different	O
physical	O
parametrizations	B-Application
or	O
varying	O
initial	O
conditions	O
.	O
</s>
<s>
Starting	O
in	O
1992	O
with	O
ensemble	B-Application
forecasts	I-Application
prepared	O
by	O
the	O
European	O
Centre	O
for	O
Medium-Range	O
Weather	O
Forecasts	O
(	O
ECMWF	O
)	O
and	O
the	O
National	O
Centers	O
for	O
Environmental	O
Prediction	O
,	O
model	O
ensemble	B-Application
forecasts	I-Application
have	O
been	O
used	O
to	O
help	O
define	O
the	O
forecast	O
uncertainty	O
and	O
to	O
extend	O
the	O
window	O
in	O
which	O
numerical	B-General_Concept
weather	I-General_Concept
forecasting	I-General_Concept
is	O
viable	O
farther	O
into	O
the	O
future	O
than	O
otherwise	O
possible	O
.	O
</s>
<s>
The	O
ECMWF	O
model	O
,	O
the	O
Ensemble	O
Prediction	O
System	O
,	O
uses	O
singular	O
vectors	O
to	O
simulate	O
the	O
initial	O
probability	O
density	O
,	O
while	O
the	O
NCEP	O
ensemble	O
,	O
the	O
Global	O
Ensemble	B-Application
Forecasting	I-Application
System	O
,	O
uses	O
a	O
technique	O
known	O
as	O
vector	O
breeding	O
.	O
</s>
<s>
The	O
UK	O
Met	O
Office	O
runs	O
global	O
and	O
regional	O
ensemble	B-Application
forecasts	I-Application
where	O
perturbations	O
to	O
initial	O
conditions	O
are	O
used	O
by	O
24	O
ensemble	O
members	O
in	O
the	O
Met	O
Office	O
Global	O
and	O
Regional	O
Ensemble	O
Prediction	O
System	O
(	O
MOGREPS	O
)	O
to	O
produce	O
24	O
different	O
forecasts	O
.	O
</s>
<s>
In	O
a	O
single	O
model-based	O
approach	O
,	O
the	O
ensemble	B-Application
forecast	I-Application
is	O
usually	O
evaluated	O
in	O
terms	O
of	O
an	O
average	O
of	O
the	O
individual	O
forecasts	O
concerning	O
one	O
forecast	O
variable	O
,	O
as	O
well	O
as	O
the	O
degree	O
of	O
agreement	O
between	O
various	O
forecasts	O
within	O
the	O
ensemble	O
system	O
,	O
as	O
represented	O
by	O
their	O
overall	O
spread	O
.	O
</s>
<s>
Ensemble	O
spread	O
is	O
diagnosed	O
through	O
tools	O
such	O
as	O
spaghetti	B-Application
diagrams	I-Application
,	O
which	O
show	O
the	O
dispersion	O
of	O
one	O
quantity	O
on	O
prognostic	O
charts	O
for	O
specific	O
time	O
steps	O
in	O
the	O
future	O
.	O
</s>
<s>
Another	O
tool	O
where	O
ensemble	O
spread	O
is	O
used	O
is	O
a	O
meteogram	B-Application
,	O
which	O
shows	O
the	O
dispersion	O
in	O
the	O
forecast	O
of	O
one	O
quantity	O
for	O
one	O
specific	O
location	O
.	O
</s>
<s>
In	O
the	O
same	O
way	O
that	O
many	O
forecasts	O
from	O
a	O
single	O
model	O
can	O
be	O
used	O
to	O
form	O
an	O
ensemble	O
,	O
multiple	O
models	O
may	O
also	O
be	O
combined	O
to	O
produce	O
an	O
ensemble	B-Application
forecast	I-Application
.	O
</s>
<s>
This	O
approach	O
is	O
called	O
multi-model	O
ensemble	B-Application
forecasting	I-Application
,	O
and	O
it	O
has	O
been	O
shown	O
to	O
improve	O
forecasts	O
when	O
compared	O
to	O
a	O
single	O
model-based	O
approach	O
.	O
</s>
<s>
Urban	O
air	O
quality	O
models	O
require	O
a	O
very	O
fine	O
computational	O
mesh	O
,	O
requiring	O
the	O
use	O
of	O
high-resolution	O
mesoscale	O
weather	B-General_Concept
models	I-General_Concept
;	O
in	O
spite	O
of	O
this	O
,	O
the	O
quality	O
of	O
numerical	O
weather	O
guidance	O
is	O
the	O
main	O
uncertainty	O
in	O
air	O
quality	O
forecasts	O
.	O
</s>
<s>
A	O
General	B-Application
Circulation	I-Application
Model	I-Application
(	O
GCM	O
)	O
is	O
a	O
mathematical	O
model	O
that	O
can	O
be	O
used	O
in	O
computer	B-Application
simulations	I-Application
of	O
the	O
global	O
circulation	O
of	O
a	O
planetary	O
atmosphere	O
or	O
ocean	O
.	O
</s>
<s>
An	O
atmospheric	O
general	B-Application
circulation	I-Application
model	I-Application
(	O
AGCM	B-Application
)	O
is	O
essentially	O
the	O
same	O
as	O
a	O
global	O
numerical	B-General_Concept
weather	I-General_Concept
prediction	I-General_Concept
model	O
,	O
and	O
some	O
(	O
such	O
as	O
the	O
one	O
used	O
in	O
the	O
UK	O
Unified	B-Application
Model	I-Application
)	O
can	O
be	O
configured	O
for	O
both	O
short-term	O
weather	O
forecasts	O
and	O
longer-term	O
climate	O
predictions	O
.	O
</s>
<s>
Along	O
with	O
sea	O
ice	O
and	O
land-surface	O
components	O
,	O
AGCMs	B-Application
and	O
oceanic	O
GCMs	O
(	O
OGCM	O
)	O
are	O
key	O
components	O
of	O
global	B-Application
climate	I-Application
models	I-Application
,	O
and	O
are	O
widely	O
applied	O
for	O
understanding	O
the	O
climate	O
and	O
projecting	O
climate	O
change	O
.	O
</s>
<s>
For	O
aspects	O
of	O
climate	O
change	O
,	O
a	O
range	O
of	O
man-made	O
chemical	O
emission	O
scenarios	O
can	O
be	O
fed	O
into	O
the	O
climate	B-Application
models	I-Application
to	O
see	O
how	O
an	O
enhanced	O
greenhouse	O
effect	O
would	O
modify	O
the	O
Earth	O
's	O
climate	O
.	O
</s>
<s>
Since	O
surface	O
winds	O
are	O
the	O
primary	O
forcing	O
mechanism	O
in	O
the	O
spectral	O
wave	O
transport	O
equation	O
,	O
ocean	O
wave	O
models	O
use	O
information	O
produced	O
by	O
numerical	B-General_Concept
weather	I-General_Concept
prediction	I-General_Concept
models	O
as	O
inputs	O
to	O
determine	O
how	O
much	O
energy	O
is	O
transferred	O
from	O
the	O
atmosphere	O
into	O
the	O
layer	O
at	O
the	O
surface	O
of	O
the	O
ocean	O
.	O
</s>
<s>
Along	O
with	O
dissipation	O
of	O
energy	O
through	O
whitecaps	O
and	O
resonance	B-Application
between	O
waves	O
,	O
surface	O
winds	O
from	O
numerical	B-General_Concept
weather	I-General_Concept
models	I-General_Concept
allow	O
for	O
more	O
accurate	O
predictions	O
of	O
the	O
state	O
of	O
the	O
sea	O
surface	O
.	O
</s>
<s>
Tropical	O
cyclone	O
forecasting	O
also	O
relies	O
on	O
data	O
provided	O
by	O
numerical	B-General_Concept
weather	I-General_Concept
models	I-General_Concept
.	O
</s>
<s>
Three	O
main	O
classes	O
of	O
tropical	B-Application
cyclone	I-Application
guidance	I-Application
models	I-Application
exist	O
:	O
Statistical	O
models	O
are	O
based	O
on	O
an	O
analysis	O
of	O
storm	O
behavior	O
using	O
climatology	O
,	O
and	O
correlate	O
a	O
storm	O
's	O
position	O
and	O
date	O
to	O
produce	O
a	O
forecast	O
that	O
is	O
not	O
based	O
on	O
the	O
physics	O
of	O
the	O
atmosphere	O
at	O
the	O
time	O
.	O
</s>
<s>
Dynamical	O
models	O
are	O
numerical	B-Application
models	I-Application
that	O
solve	O
the	O
governing	O
equations	O
of	O
fluid	O
flow	O
in	O
the	O
atmosphere	O
;	O
they	O
are	O
based	O
on	O
the	O
same	O
principles	O
as	O
other	O
limited-area	O
numerical	B-General_Concept
weather	I-General_Concept
prediction	I-General_Concept
models	O
but	O
may	O
include	O
special	O
computational	O
techniques	O
such	O
as	O
refined	O
spatial	O
domains	O
that	O
move	O
along	O
with	O
the	O
cyclone	O
.	O
</s>
<s>
In	O
1978	O
,	O
the	O
first	O
hurricane-tracking	B-Application
model	I-Application
based	O
on	O
atmospheric	O
dynamics	O
—	O
the	O
movable	O
fine-mesh	O
(	O
MFM	O
)	O
model	O
—	O
began	O
operating	O
.	O
</s>
<s>
Within	O
the	O
field	O
of	O
tropical	B-Application
cyclone	I-Application
track	I-Application
forecasting	I-Application
,	O
despite	O
the	O
ever-improving	O
dynamical	O
model	O
guidance	O
which	O
occurred	O
with	O
increased	O
computational	O
power	O
,	O
it	O
was	O
not	O
until	O
the	O
1980s	O
when	O
numerical	B-General_Concept
weather	I-General_Concept
prediction	I-General_Concept
showed	O
skill	O
,	O
and	O
until	O
the	O
1990s	O
when	O
it	O
consistently	O
outperformed	O
statistical	O
or	O
simple	O
dynamical	O
models	O
.	O
</s>
<s>
Predictions	O
of	O
the	O
intensity	O
of	O
a	O
tropical	O
cyclone	O
based	O
on	O
numerical	B-General_Concept
weather	I-General_Concept
prediction	I-General_Concept
continue	O
to	O
be	O
a	O
challenge	O
,	O
since	O
statistical	O
methods	O
continue	O
to	O
show	O
higher	O
skill	O
over	O
dynamical	O
guidance	O
.	O
</s>
<s>
When	O
there	O
is	O
a	O
low	O
amount	O
of	O
moisture	O
in	O
a	O
cellulose	O
fiber	O
,	O
volatilization	O
of	O
the	O
fuel	O
occurs	O
;	O
this	O
process	O
will	O
generate	O
intermediate	O
gaseous	O
products	O
that	O
will	O
ultimately	O
be	O
the	O
source	O
of	O
combustion	B-Application
.	O
</s>
<s>
The	O
chemical	O
kinetics	O
of	O
both	O
reactions	O
indicate	O
that	O
there	O
is	O
a	O
point	O
at	O
which	O
the	O
level	O
of	O
moisture	O
is	O
low	O
enough	O
—	O
and/or	O
heating	O
rates	O
high	O
enough	O
—	O
for	O
combustion	B-Application
processes	O
to	O
become	O
self-sufficient	O
.	O
</s>
<s>
More	O
complex	O
models	O
join	O
numerical	B-General_Concept
weather	I-General_Concept
models	I-General_Concept
or	O
computational	O
fluid	O
dynamics	O
models	O
with	O
a	O
wildfire	O
component	O
which	O
allow	O
the	O
feedback	O
effects	O
between	O
the	O
fire	O
and	O
the	O
atmosphere	O
to	O
be	O
estimated	O
.	O
</s>
<s>
In	O
fact	O
,	O
a	O
full	O
three-dimensional	O
treatment	O
of	O
combustion	B-Application
via	O
direct	O
numerical	B-Application
simulation	I-Application
at	O
scales	O
relevant	O
for	O
atmospheric	O
modeling	O
is	O
not	O
currently	O
practical	O
because	O
of	O
the	O
excessive	O
computational	O
cost	O
such	O
a	O
simulation	O
would	O
require	O
.	O
</s>
<s>
Numerical	B-General_Concept
weather	I-General_Concept
models	I-General_Concept
have	O
limited	O
forecast	O
skill	O
at	O
spatial	O
resolutions	O
under	O
,	O
forcing	O
complex	O
wildfire	O
models	O
to	O
parameterize	O
the	O
fire	O
in	O
order	O
to	O
calculate	O
how	O
the	O
winds	O
will	O
be	O
modified	O
locally	O
by	O
the	O
wildfire	O
,	O
and	O
to	O
use	O
those	O
modified	O
winds	O
to	O
determine	O
the	O
rate	O
at	O
which	O
the	O
fire	O
will	O
spread	O
locally	O
.	O
</s>
