280
D. Cluis and C. Laberge
water discharge anomalies directly related to abnormal rain the discharges (relative magnitude and timing of occur-
amounts, timing, or distributions are even less well-defined. rence for “normal” and El Niño years), this would be of
Nevertheless, some authors have succeeded in showing definite practical interest for agriculture (selection of the
significant relationships between ENSO and streamflow next crop, for example) and operational hydrology (man-
variations in different parts of the world. For example, agement of dam and reservoir levels).
Simpson et al. (1993) show a significantly high proportion
of low discharge in two Australian rivers (Murray and Dar-
Data Description
ling) for years with warm sea surface temperature (SST)
for the eastern tropical Pacific and inversely a significantly Discharge Data
high proportion of high discharge for cold SST. Moss et al.
The discharge data were directly selected from the
(
1994) establish that the probability of nonexceedance of GRDC data bank using the GRDC Catalogue Tool soft-
3
60m /s for the streamflow of a New Zealand river (Clutha) ware (Version 2.1 for Windows 95-NT). This software
3
can be seen to increase by a factor of more than five for allows the user to query for data according to specific
La Niña years in comparison to neutral years. Piechota et successive selection criteria, such as daily or monthly data
al. (1997), Dracup and Kahya (1994), Kahya and Dracup series, according to WMO region (six continental entities),
(
1993) Piechota and Dracup (1996) study relationships sub-region (regional entities or watersheds), river name,
between El Niño and La Niña events and river streamflow GRDC station number, country code, range of operational
in the United States. Eltahir (1996) in North Africa (Nile years, or size of river basin.
river) suggests that 25 percent of the natural variability in Once the query file for stations is completed, the GRDC
the annual flow of the Nile is associated with El Niño oscil- database system extracts the required selected data and
lations. Finally, it was found that the link between ENSO, provides them to the user as an ASCII file. In this case,
rainfall and streamflow is statistically significant in most stations with monthly records from WMO Regions 5
regions of Australia (Chiew et al., 1998), but not strong (Oceania-Pacific) and 2 (Asia) were extracted from the
enough to consistently and accurately predict rainfall and GRDC database. A working data set of 78 stations was
streamflow.
When relationships are found between streamflow and lines:
ENSO signals, several authors have used composite
streamflow analysis in order to describe the temporal pat- • Length of operation: The selected stations present a
obtained by using the following criteria as selection guide-
terns of these relationships. This analysis is performed using
a compounded index constructed with standardized dis-
charge series for discriminated categorical years (Niño,
Niña, and normal). Guetter and Georgakakos (1996) per-
record of a minimum of 25 recent years of continuous
operation, with generally less than 5 percent missing
data (only eight stations with more than 5 percent but
less than 10 percent missing data were selected).
formed such an analysis on the Iowa River in the United • Regional representativity: The selected stations should
States, Piechota et al. (1997) on groups of several rivers
in the western United States, and Kim and Lee (2000) on
two Korean rivers.
drain large areas, making them representative of their
climatic regions and less sensitive to local meteorologi-
cal events.
In this paper, we take advantage of the availability of • Geographic distribution: The chosen stations are dis-
a very large database of historical long-term discharge
series for numerous rivers in the world at the Global Run-
off Data Centre (GRDC), to relate these discharge (and
especially their high and low values) to the different phases
of ENSO; i.e., to test the significance of the differences in
discharge distributions for normal and El Niño years. Al-
though data are available for rivers in all continents, the
present study considers only Asian and Oceanian rivers.
It is believed that the relative magnitude of discharge is a
good integrated index for a possible teleconnection as it
results from the magnitude and time of occurrence of pre-
cipitation over the whole basin, despite its convoluted trans-
tributed throughout the whole Asia-Pacific region ac-
cording to the availability of long-term series within
the database and to their adherence to the selection
criteria. They are grouped into five regional geographic
subsets to allow possible regionalization of the obtained
results. These five subsets are: Oceania-Pacific (19
rivers), South-East Asia (9 rivers), Far East Asia (25
rivers), the Indian Subcontinent (11 rivers), and Cen-
tral Asia (13 rivers). The locations of the gauging sta-
tions are shown in Figure 1.
According to GRDC procedures, countries provide
portation within the terrestrial part of the hydrologic cycle. their discharge data for storage in the database and are
This study will investigate whether the different years solely responsible for the quality of these data. The se-
can be statistically differentiated on the basis of their dis- lected stations of the Asia-Pacific region used for this study
charge responses to a SOI index. El Niño is currently are presented in Table 1. The table presents, for each river,
monitored in almost real time and forecasts are made regu- its GRDC station number, an identification code for loca-
larly on its development; thus, if the regions under El Niño tion on Figure 1, the country location code, the name of
influence were known, as well as the temporal pattern of the river and the related gauging station, its longitude and
IWRA, Water International, Volume 27, Number 2, June 2002