Prior to the last 100 years, there is a significant lack of hydrologic knowledge for the Amazon Basi

1
Assesment of the Relationship between Andean Ice
Core Precipitation Indicators and Amazon River
Discharge
Natalie Johnson1, Doug Alsdorf2,3, Lonnie
Thompson2,3, Ellen Mosley-Thompson3,4, and John
Melack5 1.Environmental Sciences Graduate
Program, The Ohio State University,
johnson.2024_at_osu.edu 2.School of Earth Sciences,
3.Byrd Polar Research Center, 4.Department of
Geography, 5.Bren School of Environmental Science
and Management, University of California, Santa
Barbara
ABSTRACT
Prior to the last 100 years, there is a
significant lack of hydrologic knowledge for the
Amazon Basin. A 100-year record of discharge
from the city of Manaus, located at the
confluence of the Solimoes and Negro rivers, is
the most complete record for the basin. Inundated
wetlands play a key role in carbon out-gassing to
the atmosphere whereas discharge from the Amazon
River contributes about 20 of the total
freshwater flux delivered to the worlds oceans.
As discharge (Q) and inundation are directly
related to precipitation, we are developing a
method to extend our understanding of Q and
inundation into the 19th century. Using proxy
data preserved in Andean glaciers and recovered
from ice cores, annually resolved histories of
delta O-18 and mass accumulation are available.
Ice accumulation is a proxy for local
precipitation amount whereas delta O-18 is
influenced by continental scale processes (i.e.,
evaporation, convection) as well as by
temperature and hence, by varying climate
regimes. We have correlated the accumulation and
delta O-18 records from Core 1 drilled on the
Quelccaya ice-cap in the southern Andes of Peru
with the Manaus discharge data. As ice core
annual layers correspond to the thermal year (in
Peru, July to June of the following year) and
the discharge records are kept daily (January to
December), we averaged 365 days of Q data seeking
the optimal correlation for each start and end
date. The best statistical relationship between
delta O-18 and Q (r -0.41, p lt0.001) is
attained when Q is averaged from March 4 to March
3 of the following year. We also correlated 23
years of ENSO events, which are linked to both
Amazon River discharge and ice core delta O-18 (r
-0.60, p lt0.001). These linear relationships
are used to create Amazon discharge for the 20th
century and to extrapolate Q into the 19th
century. Previously developed relationships
between Q and mainstem inundated area are then
used to estimate inundated area along the main
Amazon channel for the past 200 years. The ice
core-derived estimate of inundated area for the
past 100 years compares well with the previous
and more straightforward estimates based on
discharge and remote sensing data.
METHODS
ESTIMATED DISCHARGE, INUNDATION, CO2 AMOUNTS
For this study, two methods of correlation
are used, (1) delta O-18 isotope values 6 and
Manaus discharge 7 and (2) Quelccaya ice core
accumulation 8 and Manaus discharge. To achieve
this correlation, we used the Manaus stage record
to develop a rating curve, thus estimating a
yearly discharge. Since the isotopic record from
Quelccaya is based on a thermal year (July to
July) and not a calendar year, the optimal lag
time 9 between the ice core data and the
discharge data was found (March to February) and
the discharge was recalculated to represent that
lag. Each of the two Quelccaya ice core proxies
were then correlated to the discharge. Two
discharge data sets were used all years 10 and
Richey et al.s (1989) ENSO years 11. Ice core
accumulation appears not to be correlated with
the Manaus discharge data since ice accumulation
is a local proxy.
The delta-O18 isotope vs. Manaus discharge
correlation was used to estimate the annual
discharge 12 for the Amazon going back 200
years. This discharge allows us to estimate
annual inundation for the past 100 and 200 years,
as well as to make comparisons over the past 100
years. Using the estimated discharge, inundation
for the area was calculated 13 using Sippel et
al.s (1998) predictive equation in which the
standard error of the estimate is 6400 km2.
6 Annual values of delta O-18 for the Quelccaya
ice-cap, Core 1.
12 Variations of discharge for the Amazon
River. Predicted discharge from this study is in
dark purple. Actual inundation is depicted in the
dark green. A linear predicted inundation
suggests an overall decrease in discharge rates
(r-squared value is for this straight line).
STUDY SITES
7 Annual values of discharge for the Amazon
River at the Manaus gauging station.
In order to understand Amazonian CO2
fluxes, a 94 year old (1903 to 1996) record of
Amazon discharge, taken from the stage of the Rio
Negro at Manuas 2, is being used to recreate
the past. A proxy of comparison was chosen. An
ice core history has been constructed for the
past 1500 years at the Quelccaya ice-cap 3
(13.93 degrees S, 70.83 degrees W) in the
southern Andean Mountains of Peru (Thompson et
al, 2000). Two separate records are available for
the Quelccaya ice-cap Core 1 and Summit core.
For this study, since both core data sets are
similar, Core 1 was chosen.
8 Annual values of ice accumulation for the
Quelccaya ice-cap, Core 1.
13 Variations in inundation for the mainstem
Amazon River floodplain. Predicted inundation
from this study is in dark purple. Actual
inundation is depicted in the dark green. A
linear predicted inundation suggests an overall
decrease in inundated area (r-squared value is
for this straight line).
CONCLUSIONS
9Correlation of Manaus discharge to Quelccaya
delta O-18 (Core 1). The lag is /- one year with
optimal lag noted by the vertical line at March
4.
Based on our preliminary results, we are
able to reach three main conclusions if our proxy
correlations are correct 1. The
estimated discharge in 12 and the estimated
inundation in 13 compare well with the actual
discharge from the past 100 years. Only slight
variations occur, probably as a result of
slightly different averaging. 2. 12
shows an estimated decrease of about 10,000 m3/s
discharge in the past 200 years. Although that
volume is relatively small compared to the whole
Amazon River basin, it would be similar to
loosing half of the Mississippi River flow. The
decrease suggests that there has been a
comparable decrease in precipitation. It is
important to note that this could be an
underestimate because the discharge at the Manaus
gauging station is 30,000 to 50,000 m3/s less
than the discharge at the mouth of the river.
3. 13 shows a decrease in the inundation
of the Amazon River floodplain in the past 200
years. An estimated decrease of 8000 km2 of
wetland area, such as we are predicting, is like
loosing more than one Floridas Everglades. Once
again, it is important to note that this could be
an underestimate as well since the focus of the
correlations was only on the mainstem Amazon
floodplain.
DATA RECORDS
5 Here two parameters, oxygen isotopes
(percentage delta-O18, which is used as a
temperature proxy) and dust concentration, are
shown in two sections of the ice core. The very
distinct annual layers in dust illustrate how
well the cores can be dated. Figure Thompson, et
al. 2006
For this study, two data sets were chosen the
discharge record from the Manaus gauging station
and the ice core record from the Quelccaya
ice-cap in the southern Andes of Peru 5.
Glacial records are important in the study of
global climate change. Glacial ice thickness is
an indicator of local precipitation oxygen 18
isotope values, however, are indicators of
continental-scale temperature fluctuations and
contain records of past climates within their
layers. The ice core record is one data point per
year and the exact date of that year is accurate
to /- one year. Discharge is daily daily record
with a well known temporal accuracy 4.
10 Correlation of Manaus discharge (m3/s) and
Quelccaya ice-cap oxygen 18 isotope values. Both
the discharge and delta-O18 values have been
averaged with a three year mean for optimal
correlation.
11 Correlation of Manaus discharge (m3/s) and
Quelccaya ice-cap delta-O18 isotope values. The
years selected for correlation are known ENSO
years as per Richey et al. 1989.
4 Actual Amazon River discharge from the Manaus
gauging station. Arrows indicate ENSO events.
Figure Richey, et al. 1989.
Funded by the Terrestrial Hydrology Program and
Solid Earth Natural Hazards Program at NASA