8oI.lG-USP, Publ.Esp., 8:75-88,1991.

A QUANTITATIVE MODEL FOR DESCRIBING SNOWUNES IN THE CENTRAL

ANDES MOUNTAINS

ABSTRACT

The modem snowline elevation of the central Andes can be descrlbed numerlcally from

available data in terms of regional varlations in climate anel topography. Multiple regression analysls of

modem climatic anel snowline data for locations in the Andes between 10 0 anel 30 o south latitude produces

an equation which relates snowline elevatlon to rnean annual temperature, precipltaton, anel the elevatlon of

the annual O o C isotherm. The coefficient of correlation (R = 0.83) for this model Is considered to be

significant because most of the rernaining error can be attrlbuted to the varlabillty of the reported snowline

observations which have not yet been evaluated for errors. An Inversion of the mode! should ultlmately

a1low, for the flrst time, a quantltative reconstruction of Andean paleoc1imates from Pleistocene snowline

data. The ultimate goal is to determine the differences between glacial anel modem climates, not only for the

Andes but for other major mountaln beIts of the worlcl.

INTRODUCTlON

The evaluation of present anel past snowline elevations is important for climatic

interpretations because they reflect the net interaction between weather systems and topography.

Geomorphic evidence of former valley and cirque glaciers below the modem snowline in the central Andes

demonstrates that significant climate change has occurred durlng the late Quaternary. The term snowllne

as applied in thispaper refers to the lowest Iimit of perrnanent snow on mountain slopes averaged

, Departament of GeologicaJ Scienees-Snee HalI, ComeU University, Ithaca, New Vorle 14853 USA.

regionallyanel over many years (I.e., cJimatic snowline). Snowllne e1evatlons ànd6ràdient~:.'àtong with the

size and distribution of glaciers, are determined by cJimate (temperature, precipitatlon, anel wlnd direction)

and topography (mountain elevation and morphology). The investigation of the relationship between this

variables is comp/icated between 23 o anel 32 o south latitude in a significant zone of regionally abrupt

changes in ali of the above factors (Fig. 1). This situation provides a unique opportunity to study the

interrelationship of spatially and temporally changing climatic and topographic variables on Iandscape

evolution, anel is the primary reason why this portion of the central Andes was selectecl for investigation.

The observed vertical change in snowline e1evation through time permits inferences about temporal

variations in climate as long as there is a demonstrable relationship between snowline e1evation and climatic

parameters.

THE ANDEAN SNOWUNE MODEL

Data compilation analysis

I have compiled available data on snowline positions and climatic parameters trom

numerous published anel unpublished sources, including values of surface temperature, precipitation and

evaporation. A1though the database contains records for the entire Andes, this study concentrates on the

central Andes between 10 0 anel 30 0 south latitude. The locations of the 718 stations with snowline andjor

climate data used in this study are shown in Figure 2.

To determine whether or not a general correspondence exists between the regional

change in cJlmatic parameters anel snowline gradients, a series of topo-climatic profiles, each 1,000 km long

and 100 km wide were constructed at 1 o latitude intervals across the Andes, aJthough only 3 representative

profiles are included here (Figs. 3-5). The topograpic profiles represent the averaged topography for a 100

km wide swath a10ng the profile trend, and are based upon a 5 x 5 km gricl. A1though the e1evation and

distribution of the highest peaks control climatic anel snowline pattems on a local scale, averaged

topography was selected for cJarification of regional trends.

Temperature is very sensitive to changes in surface e/evation (Fig. 6) and is controlled by

the regional temperature lapse rate. The temperature curves in Figures 3-5 iIIustrate this relationship, with

higher portions of the averaged topography having lower mean temperatures.

Topography a1so controls the zones of highest precipitation by acting as a barrier as well

as a lifting mechanism for moisture-bearing winds. Within the latitudes of the study area, these winds are

predominantly from an easterly direction, so that the lifting condensation levei is reached by molst air rising

over the Andes producing a classic rain shadow effect to the west. As a result, annual m&an precipitation is

76

-..J

-..J

MA

XlM

UM

AN

NUAL

PR

ECIP

ITA

TIO

N

(dm

) AN

D

POTE

NTI

AL

EVA

POTR

AN

SPlR

ATI

ON

(d

m)

AVER

AGE

ELEV

ATI

ON

(k

m)

IO-N

O

· 10

-

'l;R

AN

SITI

PN

I ZO

N!':

I I

~

~

ANNU

AL

ISO

THER

M

SNOW

LINE

20-

30

· 40

-50

· ro

-7

0·

LATI

TUD

E

~ A

CTIV

E V

OLC

AN

ISM

6 NO

CU

RREN

T V

OLC

AN

IC

AC

TIV

ITY

6 4 C

RES

TLIN

E 2

ELEV

ATI

ON

(k

m)

Fig

ure

1 -

Pre

cipi

tatio

n, p

oten

tial

evap

otra

nspi

ratio

n, t

em

pe

ratu

re,

and

snow

line

gra

die

nts

alo

ng t

he n

orth

-sou

th t

op

og

rap

hic

cre

stlin

e o

f th

e A

ndes

Mou

ntai

ns.

Not

e

the

rapi

d in

crea

se i

n m

ean

annu

al p

reci

pita

tion

and

coo

ler

tem

pera

ture

s so

uth

of

30

o S

lat

itude

, an

d th

e co

rres

pond

ing

depr

essl

on o

f sn

owlin

e.

Oim

ati

c d

ata

fro

m

HE

NN

ING

& H

EN

NIN

G (

1981

), N

OG

AM

I (1

976)

and

SC

HW

ER

DT

FE

GE

R (

1976

).

10 O

80 O 60 O

r---~~~~~----------,-------~r---'------------.IO O

30 O

3

. ; . •

• •

• •

•

•

L---------~------~~~ .. --------~---------J30 o

80 O 60 o

16. III """

Flgure 2 - Location of 718 central Andean c/imate and snowline observation stations used in this study. Coordinates are in degrees

south latitude and west longitude. Axes of the profiles in Flgures 3-5 (north-south) are also shown.

~ ,---,----,---,------.---,---,-------,--,-----,25 ü ü 50 z· 40 TEMPERATURE 15 ~ O ::> ~ 30 5 ~ ~~ ffi - ·5 9:: fil 10 PRECIPITATION ~ oc ~ a.. O ·15

6

5 MODERN SL

~ ::.::: 4 Z PLEISTOCENE SL O 3 f= « AVERAGED > 2 w TOPOGRAPHY ...J w

O 1000

Flgure 3 - To~imatic profile acr088 the Andes. Profile width is 100 km. and length is in kilometers from 18 o S. 73.5 °w along an

azimuth 45 o east of geographic north.

78

~ 50 ü

z' 40 O ~ 30 l-ã: 20 (3 w a: n..

10

O

6

5 ~ ::.:::: 4 z O 3 ~ [ij 2 ...J W

PRECIPITATION

MODERN SL ~~~~---&~~~

~----.!l~_6 .

~ PLEISTOCENE SL

25 ü

15 g! :::>

5 ~ a: w

·5 n.. ~ w l-

·15

Figure 4 · To~imatic profile across the Andes. Profile width 100 km, and length is in kilometers from 22°8, 72°W a10ng an

azimuth 55 ° east of geographic north.

~ 50 ü

z 40 O ~ 30 I-ã: 20

@ 10 a: n.. O

6

5 ~ ::.:::: 4 z O 3 i= « > 2 w ...J W

O

TEMPERATURE

PRECIPIT ATION

25 ü

15 g! :::>

5 ~ a: w

·5 ~ w I­

b----~~~======~~~--------~-15

1000

Figure 5 - To~imatic profile across the Andes. Profile width is 100 km, and length is in kilometers from 24.5°8, 72°W a10ng an

azimuth 97 ° east of geographic north.

79

greater on the eastem side of the Andean Cordillera in the study area (Figs. 3-5). A comparison of scatter

plots in Figures 6 and 7 show that precipitation is less sensitive than temperature to variations in elevation.

The combined temperaturejprecipitation effect on snowline position is interesting, however, because higher

peaks tend to have higher modem snowlines (Fig. 8). Most snowlines are within 1 km below nearby

mountaln peaks, regardless of snowllne elevation or clirnatic regime. This relationship suggests a dynamic

feedback mechanism between cllmate and topography.

In general, snowline gradients are consistent with climatic trends. Snowllnes are lower to

the east where annual precipitation is greater. An apparent contradiction exists between the eastward drop

in snowline elevatlon and a general rlse In temperature to the east, where surfaces are warmer because they

are lower (Flgs. 3-5). The loglcal explanation is that higher precipitation ettectively maintalns lower snowline

positions despite s1ightly higher surface temperatures.

MODELDEVELOPMENT

Annual mean temperature and precipitation are taken to be the prlmary climatic factors

controlling regional snowllne elevation, and are the most reliable and spatially comprehensive

measurements available in the central Andes. Secondary controlling factors, such as evaporation, are a1so

important but cannot be adequately quantified because of lack of data, anel probably influence the regional

snowline elevation only minimally. In order to determine the empirlcal relationship between the modem

swowline elevation and the prirnary cllmatic parameters, the annual mean temperature and precipitation

were plotted separately against the corresponding snowline elevation.

Figure 9 iIIustrates the simple linear relatlonship between mean annual temperature and

modem snowline elevation in the central Andes. Lower snowlines generally correspond with higher surface

temperatures. Again, this apparent contradiction is explained by normal temperature lapse rates in the

lower troposphere, where lower surfaces have higher temperatures.

lsotherms are often used to quantify and display the spatial variatlon in regional

temperature. Because an annual mean temperature near or below freezing is required to maintain perennial

snow, and because the freezing levei varies by over 6 km in elevation throughout the entire Andes, the

incorporation of the annual O o C isotherm elevation parameter into the model is justified in order to provide a

reference levei from which to calculate the snowline elevation. The relationship between the annual O o C

isotherm and modem snowline elevation is shown in Figure 10. Notice that in this portion of the central

Andes (10 0 to 30 0 south), there is Iittle variation in the elevation of the isotherm. This situation ilTl~ .ies that

the annual O o C isotherm elevation does not directly account for the variation in modem snowline elevations

in the central Andes. The isotherm elevation variable has, however, been incorporated into the present

80

25

() ~

w a: ~ 15 • • • •• I- • < a: • w Q. - • • • ::I • w I. I-

... 5 •• ._-- • • < , . • ~ • • • • • Z • • z • • • ._- -< • .. -.-. ~ • '1 .... rI •

-5 • -. ::I •

-15+---~--~~r-----~--~----------~------~--r---------~--~------~ 2000 3000 4000 5000 6000 7000 aooo

1oI0UNTAIN PEAK ELEVATlON (m)

Figure 6 - Comparison of mountain peak elevation and mean annual temperature in the central Andes (10-30 o S latitude).

laoo

1600

Ê 1400 !.

Z o 1200 • ~ < • • I-Q:

1000 • • • i3 w • • a: • • • Q.

800 - • • • • • ... < • .- J.. • • ~ ,

· Z • ri • • z 600 • • o< • • Z • • • • < 400 .... w • • ::I • • • • • • •

200 • • • r' .... ·)· · •••• • • • •• • \.~·h • ,. I • • o 40.00 5000 6000 7000 40 00

MOUNTAIN PEAK ELEVATlON (m)

Figure 7 - Comparision of mountain peak elevatlon and mean annual precipitation in the central Andes (10-30 o S latitude).

81

101 Z :::J ~ O Z 111

Z CC 101

8 ::I

a~~----------------------------------------------------------'

5000

• -• • -.. -I.

• • • • • -~. rl •••

M'e.- .. ·1 , .. , •• !t • ... ~ ..... .... • 11 • • • I ... \I • • - •

• •

3~4-----~-----r----~-----'----~~----r-----~----~----------~ 3000 ~ooo 6000 8000 7000 8000

REPRESENTATlVE MOUNTAIN PEAK ElEVATlON (m)

Rgure 8 - Compari80n 01 mountain peak elevation and modern snowlime elevation in the central Andes (10-30 o S latitude).

~~------------------------------------------------------------------~

:[ 1000

z o ;: o( • • ~ -I l1OOO 101

101 • Z ::::;

~ 5000 z 111

Z cc 101

8 ::I .000

~+---------~----------~--------~----------~--------~--------~ -, o o 10 20

MUH AHHUAL TEMPERATlJRE ("C)

Rgure 9 - Compari80n of mean annual temperature and modem snowline elevation in the central Andes (10-30 o S latitude).

82

model as a mathematical reference, and for intended future modeling north and south of the central Andes

where regional variations in the annual O o C isotherm elevation are significant.

Modem snowline altitude was a1so plotted against mean annual precipitation, and a

second-order polynomial regression was determined to be the "best fil" (Fig. 11). In general, a large increase

in precipitation is required to lower snowline by only a few hundred meters. As can be seen in Figures 3-5,

modem snowlines are as much as 2 km above the averaged topography in the central Andes depending

upon mean annual precipitation. For example, the profiles in Figures 3-5 show that from north to south, and

from west to east, an increase in precipitation produces a corresponding drop in modem snowline

elevation in relation to the avaraged topography.

It is acknowledged that evaporation modifies the importance of annual precipitation in

determining snowline elevation, but it appers that an increase in the rate of evaporation, physically linked to

a corresponding rise in temperature, is offset by a similar increase in precipitation. As mentioned above,

measured values for surface evaporation in the central Andes are sparse, and have not been incorporated

into the model. Available evaporation data are based empirically on temperature data, and indicate a strong

correlation between measured precipitation and calculated effective precipitation (precipitation minus

evaporation). Despite the strong correlation illustrated in Figure 12, annual precipitation was selected as the

more appropriate parameter for modeling bacause of the higher degree of confidence in measured rather

than calculated values.

After the relationships between the mean annual surface temperature, precipitation, and

O o C isotherm elevation were determined separately, a multiple regression analysis of these variables taken

together against modem snowline elevation was the final step in constructing the snowline model. The

resulting equation calculated using the StatWorks ™ 1.1 software packge for the Apple Macintosh is:

MSL ELEV = 0.001 (PRECIP)2_2(PRECIP)-32(TEMP) + 0.2(0 o C ELEV) + 5379

where MSL ELEV is the modem snowline elevation in meters, PRECIP is the mean annual precipitation in

millimeters, TEMP is the mean annual temperature in o C, and O o C ELEV is the annual O o C isotherm

elevation in meters. Observed modem snowline elevations were then plotted against snowline elevations

calculated from the above equation in order to determine the integrity (Le., R=O.83) of the model (Fig. 13).

CONCLUSIONS ANO IMPLlCATIONS

Snowlines in the central Andes can be modeled numerically in terms of an empirical

equation relating snowline elevation to surface mean annual temperature, precipitation, and the O o C

83

&~~------------------------------------~----~-------------------

! 7000

Z I O • i= • •• o( • -> • • • • • • 'l w 6000 • • •••• ··rl ..... .... • • W

~ • • J" •• w • -:.l z • • :J • rv·' ~ I} •• .. O SOOO ,--Z ••• 111

Z a: w 8

.~ :li

~;---~--~--~--~----~--~--~------~------------r---~------~ .000 .200 .600 HOO 5000

ELEVAnON OF THE ANNUAL O'C ISOTHERN (m)

Rgure 10 - Comparison of the annual O o C isotherm and modern snowline elevation in the central Andes (1D-30 o S latitude).

8~~-------------------------------------------------------------------------,

! 7000

w I z • ::l • ~ o Z 6000 11'1

~ • I UJ • • o • o • •• • • :li • • • • - • 5000 • o • • • • UJ

> a: w 11'1 !li o .000

~+----------r---------'----------r-~------~--~~---r------~-'----~ o 200 400 600 &00 1000 1200

NEAN ANNUAL PRECIPITATION (mm)

Rgure 11 - Comparison of mean annual precipitation and modern snowline elevation in the central Andes (1D-30 o S latitude).

84

1400

Ê 1200

! Z O 1000 i= oC ... ii: ~ 800 a: Co

w > 600 i= (.) LU "-"-LU

400 ..J oC :J Z Z

200 oC

O O 200 400 .00 800 1000 1200 1400

"UH ANNUAL PREClPlT.l.TION !Jnm}

Figure 12 - Comparison of mean annual precipitation and eftective annual precipitation (precipitation - evaporation) in the central Andes (10-30· S latitude). The ooefticient of correlation is 0.98.

7000,-----------------________________________________________ ~

:[ In LU Z :J 6000 ~ o z • • •• In • • • • z • • a: LU

:5 • ::I

53 • -' 5000 LU o o ::I

4000~~--------~----------,-----------~----------r_----------~------__ ~ 4000 5000 6000 7000

OBSERVED MOOERN SNOWUNES (m)

Figure 13 - Comparison of observed and modeled modern snowline elevations in the central Andes (10-30#S latitude). The ooefticient of correlation is 0.83.

85

isotherm elevation. The coefficient of correlation for this model (R = 0.83) is considered to be significant

because discrepancies between snowline elevations reported by different authors at different times are

assumed to be the source of most of the uncertainty in the model. Ali reported values for snowline, not yet

evaluated for error, were included in the dataset used to construct this model. Analysis of available remotely

sensecl imagery and evaluation of the criteria usecl to define snowline position by the various authors are

required before select observations can be justifiably removecl from the database. Now that an empirical

relationship has been developed for paleoclimatic modeling purposes, the task of reevaluating reported

snowline observations, and the collection of new ones has become more worthwhile.

The modem latltudinal and longitudinal variation of climatic zones in the Andes allOWS

one to model the results of changing one or more of the factors responsible for glaciation. The model can

be appliecl to proxy dl~tic data (i.e., cirque floor elevations) of former snowline positions to quantitatively

estimate Pleistocene temperature and precipitation values. The late Pleistocene glacial-age snowline

elevations are about 800 m lower, on the average, in altitude and generally parallel to modem ones (Figs. 3-

5 and 14). This suggests a strong similarity in wind and moisture regimes throughout much of the late

Ouaternary in the central Andes. An inversion of the derivecl snowline model should ultimately allow a

quantitative reconstruclon of Andean paleoclimates from late Pleistocene snowline data. Independently

derivecl estimates for temperature from pollen data, and precipitation trom closed basin hydrologic budget

modeling will help test the validity of this model. The ultimate goaJ is to regionally map climatic conditions

for the Andes at various times during the late Ouaternary. Further work in the Andes, especially in the

important climatic transition zone between 23 o and 32 o south latitude where there are only sparse snowline

and climatic data, is required.

Extensive summits and plateaus between 21 o and 26 o south latitude are above the

altitude of the annual O o C isotherm today, cold enough for permanent snowfields and active rock glaciers

but toe dry to support major valley glaciers as they did at least twice in the past. The position of the present

summer O o C isotherm below the modem and Pleistocene snowlines in this area implies that no decrease in

temperature would be required to initiate glaciation, but simply an increase in the amount of pecipitation.

This evidence, however, does not preclude a probable lower temperature for the central Andes during

Pleistocene glaciations.

Figure 15 shows the areas above late Pleistocene glacial-age snowlines as determined

by the author from topographic information integrated with snowline observations compiled by NOGAMI

(1976) and by STRECKER (1976). This map iIIustrates the relative extent of land surfaces subjected one or

more times to intense glacial and periglacial modification during the late Ouaternary, and serves as a useful

guide for selecting sitas for future work designed to further increase our knowledge of late Ouaternary

climates in the Andean Cordillera.

86

7~~----------------------------------------------------------~

g li 6000

;::

"" • C > .., •• • ..J .., • • .., • z :;

5000 • • .. '" ~ O z ...... o> .., .. ~~ Z .., U . ~ . • O •• • t-o>

~~ • I -• ill ..J • ... • •

~+-----~-------r------~----~------~-----' __ ----------~ 3000 ~ooo 5000 6000 7000

MODERN SNOWUNE ELEVATION (111)

Figure 14 - Compari80n of modern and late Pleistocene glacial-age snowline elevations in the central Andes (10-30 o S latitude).

61 l

"I r---~~~Y-----------~------~~---'----------~I : I

~ SI I

'z

Kllo.eLet"s 111 UI

li. L---~-----L------~~~--------~----____ -Jl' I SI • 60 •

Figure 15 - Mapped extent of areas above late Pleistocene glacial..age snowline in the central Andes based on available data

(NOGAMI, 1976 and STRECKER, 1987). The observed distribution of perennially snow-covered areas reflects the latltudinal variatíon

in late Plelstocene climate and topography. Coordinates are in degrees south latitude and west longitude.

87

ACKNOWLEDGEMENTS

Research for this paper was supportecl by NASA-JPL SIR-B continuation contract

number 956926, Task Order RE-185 to ComeU University. I thank Professor Bryan lsacks for constructing

the topo-climatic profiles anel his assistance in refining the scope of this paper. Special thanks are also due

Professor Arthur Bloom for his guidance during the development of the snowline model, anel Chalmes

Oapperton for ecIiting the manuscript

REFERENCES

HENNING, I. & HENNING, D. (1981) Potential evapotranspiration in Mountain geoecosystems of different

altitudes anel latitudes. Mountain Research anel Development, 1(3-4):267-274.

NOGAMI, M. (1976) Altitude of the modem anel Pleistocene snowline in the Aneles. Geographical Reporta

ofTokyo Metropolitan University, 11:71-86.

SCHWERDTFEGER, W. (1976) Clirnatea of Central and South America Amsterdam, Elsevier. 532p.

(Wor1d Survey of Oimatology, 12).

STRECKER, M.R. (1987) Late Cenozoic lanelscape development the Santa Marina Valley, northwest

Argentina. Ithaca, 261 p. (Ph.D. Thesis - Comell University).

88