joseph p. gerrity, jr. development division · importance of resolution and physical process...
TRANSCRIPT
U.S. DEPARTMENT OF COMMERCENATIONAL OCEANIC AND ATMOSPHERIC ADMINISTRATION
NATIONAL WEATHER SERVICENATIONAL METEOROLOGICAL CENTER
OFFICE NOTE 247
The Significance of Vertical Resolution for Predicting thePresident's Day Snowstorm of 1979
Joseph P. Gerrity, Jr.Development Division
NOVEMBER 1981
This is an unreviewed manuscript, primarilyintended for informal exchange of informationamong NMC staff members.
0
Introduction
In February 1979, an east-coast cyclone developed which caused a
very large snowfall to occur in the mid-Atlantic states. This impressive
storm, called the President's-Day storm, has been documented by Bosart
(1981), and has been the basis for a number of numerical experiments
with NMC prediction models. The storm was employed by Newell (1981) in
his study of the impact of sensible and latent heat transfer from the
sea upon the performance of an experimental ten-layer version of the LFM
model, and by Deaven in his evaluation of a new moist convection parameterization
system. Recently we completed a series of numerical experiments using a
variety of versions of the LFM model in order to isolate the relative
importance of resolution and physical process parameterizations in the
prediction of this storm. It is the objective of this paper to document
the results obtained and to propose a line of development which seems to
flow from these results.
At the time (18-20 Feb 1979) of occurrence of the President's Day
storm, the NMC was using the LFM II model described by Deaven and Newell
(1981), except that the precipitation parameterization system was still
that described by Gerrity (1976). The performance of the operational
forecast was critiqued by Bosart who rightly observed that the forecast
was insufficiently accurate to provide clear guidance that a major snowfall
was to be anticipated.
Bosart suggested that this relatively unsatisfactory performance could
be attributed to certain deficiencies in the then-operational model; viz,
1. inadequate vertical resolution
2. omission of significant level data in analysis
3. improper boundary layer physics
4. inadequate simulation of bulk effects of convective-scale processes.
The experiments reported here address to varying degrees the deficiencies
1, 3 and 4. Significant-level data are not presently available at NMC
in time for use in the objective analysis code. Future plans do call
for NMC to run a high-resolution forecast model with a later starting
time; this will permit the acquisition and use of significant level data
in the objective analysis codes.
2. The Storm
Figure 1 taken from Bosart (1981), shows the synoptic evolution of
the east coast cyclone during the period 1200 GMT 18 Feb. 79 to 1200 GMT
19 Feb. 79. In figure 2, also taken from Bosart's paper, the six-hourly
precipitation amounts are shown. The lower-right figure covers the six
hour period 1200-1800 GMT 19 Feb. 79, during which the snowfall intensity
was especially large in the Washington DC area.
In figure 3 we show the analyzed sea-surface temperature field used
in the operational NMC forecasts. The figure includes the difference
between the analysis and the RAND climatological norm for February.
Over a long strip, parallel to the east coast, the analysis shows sea
surface temperatures to be about 3°C colder than normal. Positive depar-
tures from climatology appear only off the coasts of New England and Florida.
This depiction is significantly at variance with the analysis shown by
Bosart in his figure 20. That depiction shows sea surface temperature
of 24°C off the Carolina coast which exceeds the RAND climatological
norm by two or more degrees.
Id
JULY 1981 LANCE F. BOSART 1543
w- W LI W- t- r r
Fka. 1. Surfce, 850, 500 and 300 mb maps for 1200, 0000 and 1200 GMr 18-19 February 1979. Conventional plotting and analysisscheme. Winds in m s-' [pennant - 25 m s-1, fll (half) barb - 5 (2.5) m s-r], temperature in *C. Heights, surface pressures andisotherms indicated by solid and dashed lines, respectively. Solid station circles above the surface indicate a temperature-dew pointtemerature swead , 5eC. Aircraft observations are entered on the 300 mb charts.
MONTHLY WEATHER REVIEW VOLUME 109
I; .Sa
I~~~~~~~~~~-- ~' '"<-~%'.~ ~..'~, t'/~. ~~...;.~~~~~~~~~
. oil - ...
S -~~~- '
S D S . f I a . ,, % .
'*'. . , , ,. ,,;,, -- ,____ h ia. I. (Continued)*'.."' ' . ;; * : . - _ ;:
a.
or
w i
%f
1544
441 ~ ~ ~ ~ ~ ~ ~ a
~:~ t ....I,, ,
.-",~, -;.
.. .'.I4
. I .
. '-
w .i WE~~~~~~~~~~tw'g*+-!~~~~~~0 "s
:>XU: ;~~~~~~~~~10 7
. _ _ _ _ _ . _oOf-w _ m c
~ -- i ~ re L .jR I
iI
I"
I
- w:,r ) yl,
or
...
.
m
Ir
LANCE F. BOSART
.:( . :: D:~' Li_**_or
L ~ ,-, ,.' '..~, ~~~,.~~~~~~~~~of,,
~~~-~. ''-~t,,b ,~10
'" ..,2 L _ __ - F7. h . i(.IIX~~~ , ... I.-':~~~~~~~.
,i
JULY 19811545
-F~~~~~~~5
q. . N " ( t' 'b ? -
.~~~~~~~~~~~~~~~~~~~S
I-,,
1"1'. t ': ,5 1- ': ' ' .: ' '
za " " ' " _ "
_ _ or
ct t',,
I
Ir
Bo WV _
.- : R I.
aI
I
Im
r
!fl
Pra-
ONw Ir
-., ~ , ': . .'X, o
a * * * * a . aidM A1111h
4a'e
Sh, Pc4fitk ItrI) u00oo0
or# s1-1-71
rr40
..- 'a4..
us.,a.
' ~~~~~~~~~~~~~Ji
.2%:
·'. i: .4,
*t aj Pi s.: :.a:
I.. -
-e
.i .
I I'
I0 :' '
:' ;t
' ,,, ,;. I-,,, I· :.'- .J;: s *r
',. .: rn-
. .? .,
.2 ,
* 1
. ' .N
.I \ o W 61el w . ia . ;.
-~~~~~~~~-N~~~~~~~~~~~~ -w~~~~~~~~~
I :,- . (wa OOOQtT t \ .so - 7 -.:.' .
tOk PmmimgI-)- .: , '1n
.?..I
1 9 % r-, I r I
, , ! *. -,
r . *.,
4 ' A' rm .~
' ,'E a -'* ' :
,: , . . ' ' m.,s.,"
, , , , .'.,z''''0,. 'i'..
*5.
* . ' . mu- st_ 's ' r1 v 7
Fmo. 2. Accumulated 6 h precipitation and present weather analysis for indicated times. Analyses in millimeters; raw observations for somestations plotted in hundredths of an inch (I mm = 0.04 inch).
0z-q
r* P<
:C*m
.q-4
hi
hi'C4-
4 'C0,Cx'i
.. .
I
"'I :~i~
·3 ~ .'::. ..
.,!
.
4. .
,. . .
-~~ c
oNATIONALWEATHER ERVICESEA:SURfAC~y TEMPERA+4RE X
U, -' 'ANALYSISv 4_ ,,1 FEB. ,91 9 ' "'
I PI , i-+
In figure 4 is shown an experimental regional analysis of sea surface
temperature produced by NMC's Ocean Services Group for the period 14-21
February 1979. The axis of warmest water runs from 80°W, 30°N to 75°W
34°N. The sea surface temperature in the warm-water streak is about
22.5°C except in its southwestern section where it reaches 24°C. This
objective analysis is similar to that presented by Bosart.
We are forced to conclude that the paramerization of air-sea inter-
action used in the experiments discussed subsequently is not as realistic
as one would wish. By comparison of forecasts based upon 'climatological'
and 'analyzed' sea surface temperature we may be able to clarify the
magnitude of the effect. Ideally, we would like to perform an integration
using the more realistic temperatures shown by both Bosart and the NMC
Ocean Services Group. Such an experiment will be attempted at a later
date.
Bosart presented a detailed description of the evolution of the
coastal cyclone. He shows evidence for the development of a coastal
front along which a cyclonic circulation develops. The cyclone appears
to move over land just west of Cape Hatteras at 0900 GMT 19 Feb 79. At
1200 GMT 19 Feb 79, the storm is located just east of the Delmarva penin-
sula with central pressure less than 1008 mbs.
In subsequent discussion, we will emphasize the 24 hr forecast
position of the storm produced in the several experiments. We have
examined neither the six-hourly storm positions, nor the detailed structure
of the forecasts. If we are able to repeat the experiment with corrected
sea surface temperature fields, a detailed diagnosis of the forecast
fields may be warranted.
Operational Model Capability
In this section we discuss the forecasts produced by the operational
version of the LFM used at the time the storm occurred and the forecast
produced by the presently operational version of the LFM model. The
same initial analysis fields were used in both forecasts. The presently
operational version of the LFM uses a coarser grid mesh (190.5 km @
60°N) than the originally operational version (127 km @ 60°N), but it
employs more accurate numerical approximations. The presently operational
LFM also employs improved formulations of convection and sea-air interactions.
The 12 hourly precipitation and mean sea-level pressure forecasts
are shown in figure 5. The map on the left in each panel is obtained
from the presently operational version of the LFM, while that on the
right is the forecast from the LFM which was operational at the time of
the storm occurrence.
Comparison of these forecasts with each other and with Bosart's
analysis of the storm shows that the presently operational version of
the LFM produces a forecast that is no better than the forecasts made
back in February of 1979.
To clarify the possible impact of the poor sea surface temperature
analysis available to the model in 1979, we have made an experimental
forecast using climatological sea surface temperatures which appear to
have been somewhat more accurate. We also introduced into this experiment
the new convection parameterization. The results through 36 hours are
shown in figure 6.
We notice no improvement in the forecast. The results in this run
appear somewhat more like those obtained with the presently operational
LFM, even though the result came from a model with higher horizontal
PRECIP./700
r Q13 S A.
,, 1'~ \ , - . . ) - :5
CST SFC/1000 500 T~HICK~J SVLID OO N19 FEB 1979 '9~
11 -iJ N' 1 1 - I Z - N
) I.(
--- ,,,i/g.4 va \
/- f S < , - .. t_; ' " ] " : >
,~ +~~~~~
?f~~~~J )\~~~' +-~.,-../ : ; Lo . . .. ..QP 24
.. / .I /'
M..24HR FCST. PRECI../700 ERT VEL VAL 19 F:.--B 1979 L1M9
. ,~1 -....,r~ .....1 ...... '/~~~~~~~~~~~~~~~~,---._ ' /"L:~ / t,'~iN0.~~~~ / -'--- . I-.' II . . . . .
-/ ,3 I.~-4- N, ' / -. .. 'I"1 N ./-~~~~~~~~~~~~~ //- '" /,) . ~ - ' ,O q ~ . . - z-~ .[~~~~~~~~~~~~~~~ ~ ~ ~. / If7 '~ ~ ~ ~~~~~~ ~ 'I_ .Lr~. ..,
' _7 / ' ~- ) '1-U , . Ii'',,o, ~ \ } ,...-'-' ~ ', ,' ~N./ *,. . , ' 'C .t . -- , -'/x- '. io, _ t / *ty.; ,,:- ,@~....I ~ ~ ~ ~ ~ : ..-- -:--".,~ ,
. ~ ~~~ N. \- +~ '. .':-: N'-'--- .. ~ ~...
H .. 2qR FDS T PRECIP./700 VERT VEL VALID 12Z-HION 19' FE'B 1979 LI19...
,I- �:( g- (�-
%:' I s 3
I,
I -
1
'
hf
-. �01
- '2N
A + �..
-' '� ,**
�,n. * / I agzza A-
- � *O���*)2o I � �. ( g-- - - \
* - 1, , '.4- * Tg?
1* �-'
* 36HR FCST PRECIP./700 ERT VEL VALID OOZ TUE 20' FEB
\ \_Ip l_? ,, '4 ......- ~:<·i '~ < "I')/~~- " "" ' -
j
I ,t
_ r_ ...... r\' ,. ' 't--' X~- ' ......-2 .. . . .. t ,' .- ;.
+ J s 0 t . _ I
......... '~ _.-%''~ ;''S-s '' 1a _ J_. ?-
'i- ... ~' ~-'--"C:IN: _ i ,
+
'!\P 36
) -, f\ -,- >,~ ~ ~~~~ " , F
s _ t n, v .1 Qp 36 0 ' L , -- /A - -_ _ - .; \ T 2....
L VID OOZ T - F ' LF-0- . '/ N -.
l l S ! + i / /
+ N ~ ~ ~Vt / \ A6' )P~-RF
L VALID OOZ TUE 20 FEB 1979 LFM09 - .
F.' s
,Fk -� , , F--
,0~~ ~d~ %/' C : Z 0;-:; : .Q ~ ~' 48 -/ -c7
.- Q
'BHR FCST TREUP/7020r VLE I*~k~ \/V.N.48 CS fCP1W TVS AI 2 w n
A<
; :: . ' I
SFC/1000-500
�-- ( �A
V / : 4/ i t' I
,~~~~~MT
/~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~/ I --- , .-, v -:
7t', 'k-) J.._ ,' I/~,"~;' ~-..->,,.. :;/, .-.. _. 'I- ' ', ~ . 1' '. . '~.-_. ,..----¥' ,;c. ...:'- "'
-. '2Z 1 · ' -,
~~~~~~'.--2'- ',- ....
24HR FCST SCIO 50THCNS.VLD1ZMN9 81979 L
FeS "~.x,
tF l~ ~-"-'--- ..-. ' --'·. ..
7- I f
r-c,- 7
'x~~~~~~~~~~~~~~~~~~~~ -~~~~~/'"
,'
I . ,, - , . I.- .c '..'J,,.-
- ' '~ ~ _t&~~~~~9 '
-.~~~~~~~~~~~~~ ,,~ ~ ~ ~ ~ ~ ~ ~~~~2 ,,._, ' :,".':*_--~ _:_,~.~_,/..i",o ~ ~ ~ ~ ~ -oj..?~ '',-' ",'~-L.-~ p .*~ ~o *~jJ ; 'A"
,- ..~.~.~ ~_~~**
36' FS RCP/0 R E V~ 0 U~ E 99 LM
I /' ' ' ~:~ ~ ""~~~,. -: -
Hs e~~~~~~
'Fuz..
36HR FCS SFC/1000-500 I S .. %s36HR~~~~~~~~~~~~~.~--. FC/F/00SOTICNS AI O UE2FB99lto
V-,G- 76i3
resolution. This result suggests that the forecast is not very sensitive-
to the sea-surface temperature, convection formulation or to horizontal
resolution.
Ten Layer Version of the LFM
Newell (1981) presented results obtained with a ten layer version
of the LFM II model. Those experiments were conducted to evaluate the
impact of air-sea interactions on the model forecasts. Very recently,
Collins et al. (1981) presented their results of an intercomparison of
several regional forecast models, including the ten layer version of the
LFM II. Except for some problems with warm season convective precipitation,
the ten layer LFM II performed as well or better than all other models
tested.
This ten layer version of the LFM has tropospheric vertical resolution
of about 120 mbs and stratospheric resolution of about 70 mbs. It also
includes a 50 mb surface boundary layer. The horizontal resolution is
127 km at 60°N latitude on the polar stereographic map. The model
incorporates all the physical processes contained in the presently operational
LFM, including a bulk aerodynamic estimate of evaporation from the sea.
Figure 7 shows the 24 and 36 hr forecasts of 12-hourly precipitation
and mean sea level pressure produced by the ten-layer model. The predicted
position of the storm at 1200 GMT 19 Feb. 79 agrees with Bosart's analysis
of the storm's position at 0900 GMT 19 Feb 79, although the forecast
storm is to the east of Bosart's position and is about 3 mb deeper. The
predicted precipitation (isoplethed at intervals of 1/2 inch) agrees
quite well with Bosart's analysis.
It seems clear that the increased vertical resolution of the model
permitted the achievement of a much more accurate forecast of this storm.
Conclusions
From the results presented here, it seems evident that for certain
types of storms greater vertical resolution is required in NMC's operational
regional forecast models.
We could show no direct evidence that more accurate sea surface
temperature analyses are necessary, but since the impact of air-sea
interactions on model forecasts was conclusively shown by Newell (1981),
it is obvious that reasonably accurate sea surface and atmospheric temperature
analysis are essential for predicting east-coastal cyclogenesis. We
ought to repeat this forecast using more accurate sea surface temperatures.
The details of the coastal frontogenesis and the impact of convection
are difficult to document in forecasts produced with models developed
for operational use. It seems that additional progress in operational
model development will require the construction of versions of the
operational model that lend themselves to detailed diagnosis. To do
this effectively NMC will need to significantly enhance its computer
graphics capability for use in research and development.
At NMC we are planning continued development of high resolution
models for operational use; and in addition we are emphasizing the develop-
ment of improved objective-analysis systems. The implementation of
these new methods is dependent upon NOAA's acquisition of an augmented
computational facility. It is our expectation that a President's Day
storm in 1983, should it occur, will be predicted with a precision far
surpassing the result achieved in 1979.
References
Bosart, L. F., 1981: The President's Day Snowstorm of 18-19 February
1979: A Subsynoptic-Scale Event. Monthly Weather Review 109:7,
pp. 1542-1566.
Collins, W., et al., 1981: Regional Model Intercomparison Unpublished
manuscript (70 pp and figures). Development Division, National
Meteorological Center, NWS, NOAA, Washington, DC.
Gerrity, J. P., 1978: The LFM Model - 1976: A Documentation NOAA Tech.
Memo NWS NMC 60, 68 pp., National Weather Service, NOAA, Dept. of
Commerce, Washington, DC
Newell, J. E., 1981: Experiments on the Parameterization of the Flux of
Sensible Heat and Moisture from the Sea Surface into the LFM Boun-
dary Layer. Office Note 237, National Meteorological Center, NWS,
NOAA, Washington DC
Newell, J. E. and D. G. Deaven, 1981: The LFM-II Model - 1980.
NOAA Tech. Memo. NWS NMC 66, 20 pp., Nat. Wea. Service, NOAA, Dept.
of Commerce, Washington, DC