SNOW AND AVALANCHES FORECASTING OVER THE ANDES MOUNTAlNS
Jose A. Vergara*
Departamento de Geoflsica, Universidad de Chile
Rene leon,
Caminos y Nieve, ANDINA-CODElCO, CHILE
ABSTRACT: During the last three years (1998-2000), the University of Chile and COOElCO-ANOINA
has directed significant research towards the implementation and evaluation of an operational regional
weather forecast over the Andes Mountains (Vergara et. AI, 1999). The implementation is designed to
provide traditional synoptic forecasts at high-resolution grid spacing and local weather forecasts for
providing warning of avalanches during winter time.
The current analysis uses data based on daily snow precipitation in lagunitas (2670 m above
msl), located in the central part of Chile of the Andes Mountains, as well as daily model forecasts. For
this station the period study covers the last years. The 1997 and 2000 winter was one of the wettest
winters during the last 30 years.
This paper explores the performance of snow and avalanches forecasts during several cases of
strong weather conditions over the Andes Mountains (Figure 1) dUring EI NiJio year of 1997 and
compares the model results with snow observations. The results show that the 24 hour forecasts explain
about 80% of the variance in snow depth data. The study demonstrates the operational capability of the
model to forecasts total snow depth during the storm. Finally, the model is shown to be an effective tool
for support of avalanches forecasts over the Andes Mountains due to the strong relationship between
snowfall and avalanches.
KEYWORDS: Snow, avalanches, meteorology, Andes Mountains, Chile
1. INTRODUCTION
The Andes mountains, the longest chain
of mountains in the south hemisphere, shows a
strong and complex variations in the snow and
meteorological conditions as one move from east
to west, because the Andes is the narrows and
in the world. The annual
longest mountains
mean snowfaH in the Andes central part Chile is
close to 10 meters, with a strong interannual
variability from 2.2 meters on 1a Nina year of 1998
to 21.44 meters on the el Nino years of 1972
(Figures 2 and 3), and the snow may faU in the
any time on the winter from April to September
with a maximum in June/July (Figures 3 and 4).
During the winters the means minimum
temperatures is close to -10C and the maximum
mean temperature is close 12C (Figure 5). The
high temperatures on the mountains usually take
place immediately following the storms associate
to anticyclone subsidence, and the avalanches
formation on the study area (Figure 1) is
proportional to the number of hours of closed
roads due to avalanches (Figure 6), them the
maximum avalanches formation are maximum in
June-July (Figure 7). But the ratio of the number
of avalanches to the total snowfall of the storm
increase during the second part of the winter and
spring time because the temperature increase
dramaticaHy and them melting (Figure 7).
Because the strong relationships from the
since of the storm and the number of avalanches
(Figures 6 and 8). The weather observation a
forecasts are the primary tools for predicting the
total snow precipitation
on the storm,
the
avalanches potential and hours of roads are
close due to the avalanches.
Corresponding author address: Jose Vergara,
Departamento de Geofisica Universidad de Chile,
Santiago-Chile; email: [email protected].
566
Figure 1~ Snow depth stake at Lagunitas stations (2670 m above ms/); From September of 1998 at the
end of the winter (right), one of the driest winters of the last one hundred years and September of 2000
(left) one of wettest last thirty years.
567
Fjgure 2: Piuquenes to Haulage Roads though the avalanches area, during of year 1998 winter (top),
and 2000 (bottom).
568
2. RESULTS
75O-I---+---+--+-+-HI&-ti-..-t.IIIH-+--t--t
LlO -I--+-~-+--+-+----'l"'~-fQ'I"
~
! -I--+-+--+-+-+*i-O't
1SO
Jj,
31
51
91
121
1S1
1S1
2"
241
271
J01
331
361
Days
Figure 3: Time evolution of snowpack depth
during the EJ Nino year 1997 (circle) and La Nina
year 1998 (squared) and 2000 year (triangle) at
Lagunitas station central part of Chile (Figure 2).
This paper explores the performance of
the NCEP Regional Spectral Model (Vergara, et
.al, 1998) during winters of 1997 (EI Nifio) and
1998 (La Nina). The winter of 1997 one of the
wettest on record is important in terms of
precipitation. The 1998 winter was one of driest
winters of the last one hundred years (Figure 2)
and no important rainfall events occurred. During
both years the RSM was run operationally and
several cases of extreme weather events occurred
during these years (in particular during the 1997
winter).
1000
800
700
!600
j
500
400
300
100
I
I
I
I
I
j
I
JAN
1/
,/
Y
-15
\..
1""-~
. . .
'"
A:rI4AI
I
,I
I
I
-,.......
1'-
I
I I
~~
I I
,-
I
~
1
I
1
1
'
"
IJUL
.. ~ .. L
1M
J
ISEP
. !o ...
•
..
IJ'm
I
"
I
LJ..,
MAR
I
I
-Hl
. I . .. 1\1\.....r-....
-
,
15
-5
1\
J
II
!
2D
I ""'\
\
I
I
I
25
I ..
I I
I
In order to evaluate the model
performance, model outputs for the two-years are
correlated with observed precipitation data. The
RSM precipitation forecasts over the Andes
Mountains in the central part of Chile explain
about 70% of variance in the observed daily rain
data, with explained variance of 78% for the 24
hour forecasts and 66% for the 48 hour forecasts
with a mean error close to 25 mm (Vergara,
2000). Snowfall forecasts at Lagunitas local
station explain about 81 % (24 hour forecasts,
Figure 9a) and 78% (48 hour forecasts, Figure 9b)
of the snowdepth, with a mean error close to
20cm (24 hours) and 3Ocm( 48 hours). The RSM
provides
accurate forecasts of the heaviest
rainfall and snowfall events over Chile and the
Andes Mountains. These forecasts of heavy
events are of particular interest since they can
provide better avalanches forecasts and warning
due to the strong relationship between snowfall
and avalanches (Figures 6 and 8). The high
resolutions weather forecast, is able to provide
better forecasts for local weather conditions
because of its ability to resolve small scale
weather features and the complex and strong
.
slope over the Andes topography.
MAY
",,-
JUL
SEP
1-
NOV
Figure 5: Annual cycle of maximum (continues
line) and mean (dots line) temperature in the
Lagunitas station.
Month
~i9ure 4: Annual cycle of maximum (continues
hne) and mean (dots line) snow precipitation in
the Lagunitas station.
569
1200 -,-----,------,--,---...,.----,----;
'" 1000
:;
o
:; 800
'"
Q
240
+---'---"'-q'--"--~r---+--~J---1
:s
+---+-------1e-------+--._------b~-----j
192 +-----+----+--~'_t_---l
o
~ 168 +---+---+----.~_j__---j
Q)
£oo+--_j__-----1'-a---h~-+-_1
g' 144 +---+---+~-_j__---j
III
£
'0
~
"S
::I:
200
~
120 -j------+---'f,-----+-----I
'"
Q)
96 +-----+--.,..-+----+-----1
,
400 +---+-------1--:1..---+---+------j
~
o
t-----I
fn~
~
~
-,-----,------,---~--__,
216 +-----+----+---
+--_j__~y,-::--+---,---+----I
ii
72 +-----f-:;;;:-F--+----_j__---j
£
48
~
C'"
+----,,~--+----_j__---j
24 -tiii;;-::iI....---t----r---t------j
500
1000
1500
2000
2500
OF-=---+----+---+-----l
o
Total Annual Snow (em)
100
200
300
400
Total Snow (em)
Figure 6: The annual number of hours of closed
roads due to avalanches as a function of total
annual snow at Lagunitas station (2670 msnm).
Figure 8: The number of hours of closed roads
due to avalanches as a function of total snow
during 1997 at Lagunitas station {2670 msnm).
3. CONCLUSIONS
I
I
I
If "''\,
I
J
MAY
I
\
. . I\.
".......
/ ~ I"'"
MAR
I
~
J
JAN
I
JUL
SEP
I
I
NOV
Figure 7: Annual cycle
of the number of
maximun (continues line) and mean (dots line)
hours of closed roads due to 1997 at Lagunitas
station.
Performance of the operational forecast
system over the Andes Mountains in the central
part of Chile was presented. The objective of the
system is to provide better snowfall forecasts for
synoptic and mesoscale phenomena. The model
demonstrated an operational capability to run
successfully over South America (in particular
Chile) and successfully simulated strong rainfall
events, especially during the El Nino event of
1997/1998. The model proved to be an effective
tool for the support of mountains weather snowfall
and rain forecasts over the chilean mountains.
4. ACKNOWLEDGEMENTS
This study was supported by grants FONDECYT
1970507-1997, COOELCO ANOINA-99.
570
..,-
2
Vergara, J., 2000: Snow and Rain forecasting
over the Andes Mountains,
AWRA"s Spring Specialty
Conference "Water Resources In
Extreme Environments", 243-248.
Lagunitas
200
S 150
no
w 100
(c
m
)
50
0
o
100
50
150
200
FORECAST 24 Hours (mm)
r
Lagunitas
200
SN 150
•
0
W
(c 100
r
m)
50 ----,-ifL---+----f-----i
0
o
50
100
150
200
FORECAST 48 Hours (mm)
Figure ga, b: Regression of 24 (top) and 48
(bottom) h<:>ur snow forecasts on control station
observations. The solid line represents the linear
regression.
5. References
Vergara. J., Juang, H.-M and S-Y Hong, 1998: An
of
the
regional
evaluation
forecasting
for
Chile-South
America, AMS 12th Conference on
Numerical Weather Prediction,
151-152.
Vergara, J A. Ellena and R. Leon, 1999: La
meteorologia en el mantenimiento
de caminos de alta montana, IV
Congreso Previal Chile, 623-630.
.......
571