Pdo Enso
Pdo Enso
JORGE A. RAMÍREZ
Associate Professor, Civil Engineering Department, Colorado State University, Ft. Colllins, CO
May 1, 2000
ABSTRACT
Decadal climate variability in the form of the Pacific inter-Decadal Oscillation (PDO) is analyzed
with the El Niño Southern Oscillation (ENSO) to determine impacts on seasonal water supply in
the Columbia River Basin. A basic overview of the ENSO and PDO phenomena is presented.
The cross-correlation functions suggest a significant correlation between seasonal runoff and both
the Southern Oscillation Index (SOI) and PDO suggesting a critical time period where ENSO
events must form to have a significant impact on the January through July volume runoff at The
Dalles Dam. Results show there is a statistically significant change in the mean January through
July volume runoff only when El Niño events occur during the cold phase of the PDO and when
La Niña events occur during the warm phase of the PDO. Precipitation and temperature patterns
are illustrated to explain this shift in the mean runoff.
1. Introduction
It is a forgone conclusion that periodic anomalies exist in the ocean-atmosphere system in the
Pacific Ocean. In fact, such anomalies exist in many of this planet’s major oceans (Enfield and
Mestas-Nuñez, 1999). The Pacific Ocean is singled out in this statement simply because of the
gross media attention given recently to the phenomenon known as the El Niño Southern
Oscillation (ENSO). The warmest ENSO event ever recorded occurred during the 1997-98
season. This event presented the opportunity to test the scientific community’s skill in applying
climate prediction models and estimating the effects of the event on global precipitation,
temperature, sea surface temperature (SST), etc.
The shifts in these climate parameters as a result of the larger-scale climatic oscillation are of
primary concern here. Drastic and persistent changes in these parameters may result in flooding,
drought, and/or changes in the form of seasonal precipitation (i.e. snow or rain). The latter may
play a vital role in determining the available water for a given year for those regions that depend
on water supplies where the majority of the volume comes from seasonal snowpack.
In this analysis, we will look specifically at the Columbia River Basin located in the
northwestern United States and southwestern Canada. This basin has been significantly
developed for hydroelectric power production since the early 1930s. The purposes of many of
these dams and reclamation projects include power production, irrigation, navigation, and
recreation. In recent years, resource managers have also been charged with the responsibility of
protecting fish and wildlife habitat while fulfilling each of these other uses. This balancing act
of the multiple-purpose reservoir network was further complicated in the late 1990s with the
deregulation of the wholesale and retail electric utility industry. This change placed a greater
economic value on Columbia River water that did not exist in previous years. This greater value
resulted in better efforts to plan the operation of the reservoir network to satisfy all needs while
maximizing the economic value of the water.
Water supply forecasts drive the overall planning of any reservoir system. As hydrologists and
engineers, we recognize that a large degree of uncertainty resides in predicting both weather and
the resulting hydrologic response. A better understanding of the role of climate in hydrologic
response may help to reduce some of this uncertainty.
In a previous study by the author, the effect of ENSO events on the hydrology of the Pacific
Northwest was analyzed assuming ENSO events were the only driving variable. This analysis
drew conclusions that demonstrated a fair correlation with ENSO events. Temperature and
precipitation patterns showed a general trend to warmer and drier winter conditions during the
warm phase of ENSO with the opposite trend exhibited during the cold phase. This resulted in a
trend toward below normal runoff volume during warm ENSO events and above normal volume
during the cold phase. There were, however, some outlying data points and nonconformities that
could not be explained with this methodology. This paper builds upon that analysis attempting
to refine the earlier study and explain some of the stray data points.
In addition to investigating the response of seasonal runoff in the Columbia River Basin to
ENSO events, we will overlay the response with an additional phenomenon called the Pacific
inter-Decadal Oscillation (PDO). This phenomenon, which manifests itself in the northern
Pacific Ocean, is hypothesized to explain some of the nonconformities from the previous
analysis. A brief discussion of each of these phenomena and the data used for the analysis
follows in the next few sections. The results are presented and discussed in Section 4, with
conclusions summarized in Section 5.
The following two subsections briefly describe some aspects of the phenomena and indices used
in this study. This is intended to give the reader only some understanding of the signatures of the
phenomena. The mechanisms causing these phenomena are not completely understood
(Gershunov and Barnett, 1998). Therein lies the difficulty in predicting the events themselves.
Here we only aim to describe the characteristic anomalies observed during these events such that
there is a general understanding of the synoptic level patterns when we address shifts in
precipitation and temperature patterns. As will be explained in the following sections, the end
effects of each phenomenon are similar, but the climatic signals that identify each one and the
time scales differ significantly.
The ENSO phenomenon is associated with anomalous sea-level pressure, surface winds, and
SST near the equatorial Pacific. Fishermen near Ecuador and Peru first noted the shifts in ocean
2
temperature. They noticed a warm current near their respective shorelines that occurred annually
around late December (NOAA, 1994; Mock, 1981). With the proximity of the arrival time to
Christmas, they termed this warm current El Niño, Spanish for “the Christ Child” (NOAA,
1994). Although this is an annual occurrence, some years exhibited warmer conditions and some
colder. Modern terminology designates an anomalous warm event El Niño with its colder
counterpart termed La Niña (girl child).
The primary signature of an ENSO event is in the sea-level pressure gradient currently measured
between Darwin, Australia and Tahiti (NOAA, 1994). This gradient is the primary variable used
to measure the magnitude of an ENSO event in the form of the Southern Oscillation Index (SOI).
Under normal conditions, the easterly trade winds create surface currents which pool warm water
near Indonesia and create an upwelling of cold, nutrient rich water along the South American
coast (NOAA, 1994). The warm pool of water provides a large energy source, which supports
seasonal, convective storms around eastern Asia.
In an El Niño episode, this gradient relaxes, leading to a weakening of the easterly trade winds,
thus reducing the surface ocean currents. The result is a reduction in cold upwelling and an
eastward migration of the warm sea-surface water along the equatorial region. This allows for a
greater extent of warm water for convection and shifts the location of convection east (NOAA,
1994). Figure 1 shows a schematic representation of an El Niño. Note in Figure 1 the reduced
level of cold upwelling in the eastern Pacific as a result of the decreased surface currents.
Figure 1 – Schematic representation of the eastward shift of sea surface temperature and
convective loop during an El Niño (right). (source: Canadian Meteorological Centre).
The La Niña exhibits opposite tendencies with a strengthening of the pressure gradient inhibiting
the normal migration of the warm water. The result is cooler SST anomalies in the eastern
Pacific and a constriction of convection in the western Pacific. Figure 2 illustrates actual SST
during recent ENSO events compared to a normal condition in 1993.
The oceans and the atmosphere are continuously engaged in feedback transfers (NOAA, 1994;
Peixoto and Oort, 1992). The surface and upper air patterns begin to change as a result of the
changes in SST. The temporal and spatial changes in surface and upper level winds affect the
same attributes of air temperature and precipitation. These may further compound the ENSO
event. Figure 3 illustrates typical winter upper air anomalies that have been observed in past,
strong ENSO events. The positive anomalies in North America during El Niño result in a high-
pressure ridge forming over the western coast of the continent, leaving the Columbia Basin in a
3
region of weak, inactive flow aloft. Storms transported along the jet stream tend to be split with
more energy being directed along the southern branch of the jet stream than would exist under
normal conditions.
Figure 2 – Typical sea-surface temperatures during recent El Niño and La Niña events
compared to normal. (source: NOAA/PMEL/TOA).
In La Niña winters, negative anomalies over the continent tend to consolidate jet stream energy
and direct it toward the Pacific Northwestern states and southwestern Canada.
Figure 3 – Anomalies in the 500 mb (~18,000 ft) geopotential height field associated with typical strong El Niños
(left) and La Niñas (right). The bold arrows show the approximate locations of the jetstreams. (source: Canadian
Meteorological Centre).
It becomes easy to see that the winds may cause the ENSO event or that the winds are a response
to the event. The causality relationship between the event and the signature changes in climate is
4
poorly understood (Gershunov and Barnett, 1998). As stated earlier, the scope of this analysis is
to examine the impacts of the changes in climate to the hydrology of the Pacific Northwest and
not to dwell on the mechanics of the events themselves.
Anomalous circulation oscillations occurring on a decadal time scale have been found in several
previous studies. Enfield and Mestas-Nuñez (1999) found evidence of three low frequency
oscillations by investigating SST anomalies and filtering out variability associated with ENSO
events. These included the North Atlantic Oscillation (NAO), Arctic Oscillation (AO) and the
phenomenon of interest here, the PDO. Livezey and Smith (1999) also identified three distinct
phenomena occurring in the Northern Hemisphere. Here, they recognized the inter-annual
variability associated with ENSO and identified the NAO as well as the third “global” climate
signal occurring with a decadal cycle (Livezey and Smith, 1999). Cayan, et al. (1998) showed
that decadal variations in sea level pressure and SST in the northern Pacific were responsible for
up to 50 percent of the annual variability of precipitation in western North America. Much of the
sea level pressure anomalies were associated with areas near the Aleutian low, a climatic feature
present in the north pacific winter months. It is believed most of the variability associated with
the Aleutian low can be attributed to variability in the PDO (Cayan, et al., 1998; Overland, et al.,
1999). Mantua, et al. (1997) demonstrated the low frequency cycle of the PDO and drew
conclusions as to the impacts on salmon in the Pacific. In a study similar to the one presented
here, Gershunov and Barnett (1998) disaggregated sea level pressure and precipitation anomalies
according to ENSO events and further by PDO phase. They showed that a relationship exists
between ENSO and PDO that leads to constructive and destructive enhancement of ENSO
events. The conclusions of their work are the basis of the investigation presented here.
The PDO is indexed by SST in the North Pacific (Mantua, et al., 1997). Like the ENSO, the
PDO exists in cool (positive) and warm (negative) phases. The positive phase of PDO manifests
itself in negative SST anomalies in the central north Pacific (Mantua, et al., 1997). The cool
temperatures accentuate the SST gradient near 30°N, which then enhance the westerly surface
winds. The invigorated westerlies advect more cool air throughout the north Pacific as to sustain
the cool SST anomaly (Enfield and Mestas-Nuñez, 1999). The added energy in the midlatitudes
results in an increased level of cyclonic activity in the region overlying the cold anomaly and a
weakening of the low latitude easterlies (Enfield and Mestas-Nuñez, 1999). This weakening is
similar to the signal associated with the El Niño phase of the SOI in that warmer than normal
waters surge east from the slackening of the east winds and the intensification of the westerlies.
The cooler anomaly and increased dynamic environment results in a deepening of the Aleutian
low with winter sea level pressure anomalies up to 4 mb (Mantua, et al., 1997; Enfield and
Mestas-Nuñez, 1999; Gershunov and Barnett, 1998).
Using the SST anomalies to index the PDO, Mantua, et al. (1997) showed the PDO switched
polarity in 1924, 1947 and 1977, with negative polarity during the period 1947 to 1976. With the
similarity of the signatures between positive polarity of the PDO and El Niño, it stands to reason
that one may hypothesize a constructive relationship when a warm ENSO event occurs while the
PDO is positive, and vice versa. Conversely, a destructive, or dampening effect may result when
the two phenomena are out of phase. This is exactly what Gershunov and Barnett (1998) showed
5
was the case with respect to heavy precipitation and sea level pressure. From here, we will
investigate whether these conclusions may be expanded to make general statements with respect
to the hydrology of the Columbia River Basin.
The data required to fulfill the goals of this study included the indices used to measure ENSO
and PDO magnitude as well as a representative value of the runoff for the Columbia River Basin.
Temperature and precipitation data are also needed to provide explanation of any results. The
following sub sections discuss the data used in this study and the manner in which they were
analyzed.
The SOI is used as the primary measure of the state of the ENSO. The SOI is computed as the
normalized difference in standardized SLP anomalies between Tahiti and Darwin relative to its
root mean square. Other indices exist to track the state of the ENSO, but this index has been
measured and/or reconstructed back to the late 19th century. Its complete and long period of
record make it very desirable for this study. The SOI data used in this study was retrieved from
the Joint Institute for the Study of the Atmosphere and Ocean (JISAO), University of
Washington, Seattle. Figure 4 shows the SOI time series and the high frequency variability
associated with ENSO. This study only uses SOI data after 1900. The PDO data was available
back to this date and therefore limits the analysis to the common record for all data sets.
4
-1
-2
-3
-4
1 9 0 0 1 9 1 0 1 9 2 0 1 9 3 0 1 9 4 0 1 9 5 0 1 9 6 0 1 9 7 0 1 9 8 0 1 9 9 0
Figure 4 – Unsmoothed SOI time series. Positive values indicate La Niña episodes.
Many researchers refer to specific years as El Niño or La Niña years based on the year of onset.
The El Niño or La Niña phase generally develops around September or October (NOAA, 1994).
The annual variability conveniently lends itself to this level of grouping. Also convenient is the
general coincidence of the regular development of ENSO events with the water year defined as
October to September. This study defines years as being El Niño or La Niña years based on
water year and the predominance of one polarity of the SOI during that year.
A 3-month moving average was computed to reduce some of the noise in the SOI time series. A
season was defined as being under the influence of an ENSO event if the smoothed SOI value for
that season or the prior season exceeded ±0.6. Four seasons were defined for the year in Table 1.
6
Table 1 – Definition of seasons used in this study
Autumn September – November
Winter December – February
Spring March – May
Summer June – August
A definitive list of ENSO event years does not exist. Different agencies and researchers declare
differing years depending on their individual criteria and assumptions. There is better agreement
in the last twenty years of record. Table 2 lists the water years declared as ENSO event years
used in this study. For the most part, there is good agreement with other agencies (Canadian
Meteorological Centre, National Oceanic and Atmospheric Administration, Australian Bureau of
Meteorology). There is often a one-year discrepancy that results from the standard declaration
using the year of onset versus the water year definition used here.
Table 2 – List of El Niño and La Niña water years used in this study.
The PDO is indexed by the leading North Pacific SST pattern (Mantua, et al., 1997). The PDO
data was also obtained from JISAO which are from the original work by Mantua, et al. (1997).
Figure 5 shows the unsmoothed PDO time series.
7
3
-1
-2
-3
1 9 0 0 1 9 1 0 1 9 2 0 1 9 3 0 1 9 4 0 1 9 5 0 1 9 6 0 1 9 7 0 1 9 8 0 1 9 9 0 2 0 0 0
Figure 5 – PDO time series showing the cold (positive) and warm (negative) phases. The dashed lines indicate the
changes in polarity determined by Mantua, et al. (1997).
Figure 5 clearly shows the lower frequency pattern in the PDO. There is some inter-annual
variability but it has been shown to be largely negligible with respect to the decadal variability
(Mantua, et al., 1997). The PDO phase prior to 1924 is unclear from the time series plot and
references do not make any statements prior to that date. Since this is a point of uncertain
phasing of the PDO, this study will only deal with water years 1924 to 1998.
In keeping with the conclusions from previous studies (Mantua, et al., 1997; Gershunov and
Barnett, 1998) the PDO was divided into cold and warm epochs marked by the changes in
polarity, denoted by the dashed lines in Figure 5. From this, the years in Table 2 may be further
grouped according to PDO polarity. This is shown in Table 3.
From these groups of water years, changes in the mean runoff and variability for each sample
were compared with the neutral years to see if there is a statistically significant shift resulting
from combinations of PDO and ENSO. Each combination of ENSO and PDO is referred to as a
8
climate condition. The neutral condition is defined as when the ENSO is neither in the El Niño
or La Niña phase irrespective of the PDO.
The Columbia River Basin is located primarily in four western states in the Pacific Northwest
(Oregon, Washington, Idaho, and western Montana) and British Columbia, Canada. Minor
portions of the basin extend into three surrounding states. Flow measurements for this study are
taken at The Dalles Dam, Oregon from the USGS modified streamflow records. The term
modified refers to a procedure used to adjust the flow data for irrigation levels, changes in
storage, etc. In other words, the modified flows are the best estimate of the natural flow
conditions at a specific depletion level, in this case 1990 levels. Records have been
reconstructed to the early 19th century, but are only considered past 1924 for reasons stated
earlier.
The Dalles Dam is a key forecasting point for water supply in the Columbia River Basin. Its
operational significance was the primary reason for choosing this as the study point. The Basin
is very large and is therefore climatically diverse. The western extreme of the Basin is on the
leeward side of the Cascade Range with fairly mild sloping loess-based soils and rocky outcrops.
This terrain gradually transitions into the western foothills of the Rocky Mountains. Because of
the diversity of climate and topography, it is prudent to divide the basin accordingly. This will
also aid in explaining the results of this study. Figure 6 shows how the Columbia River Basin
was subdivided for this study.
9
The Columbia River above Grand Coulee Dam (GCL) extends into northern Idaho, western
Montana, and southeastern British Columbia. This area is generally forested and of much higher
relief than the western half of the Basin. In the mean, 61 percent of the January through July
volume runoff is generated in this portion of the Basin. Twenty-seven percent of the January
through July runoff at The Dalles Dam comes from the Snake River above Lower Granite Dam
(LWG), in the mean. The latter area consists of drainage from the south and west sides of the
Bitterroot Range and the Southeast Highlands of Idaho surrounding the Treasure Valley in
Southern Idaho. This is largely forested in the higher elevations with agricultural activities
dominating the lower valleys. The remainder of the Basin will be referred to as the West Basin.
The annual water supply is stored for a large portion of the year as snow in the mountains. The
accumulation season starts in mid October to early November and continues until mid March.
Much of the seasonal snowpack is reduced to runoff by the end of July. The January through
July volume runoff is used as a general descriptor of the water supply for a given year. Many
operational policies and contractual obligations with respect to fish and wildlife management and
flood control depend on volume runoff over these seven months. Since this has the most
operational utility, this study concentrated on properties of this seasonal volume of water.
Of interest here is to determine patterns in the January through July volume runoff associated
with the ENSO and PDO phases. The years were grouped as in Table 3 and runoff data for each
climate condition were fit to standard statistical distributions. Using a skewness test for
normality (Salas, et al., 1980) the distribution of volume runoff for each climate condition
(including the neutral condition) were found to be well approximated by the normal probability
distribution. While we recognize that the normal distribution will give a nonzero probability of
less than zero runoff, this probability is so small that we may neglect the necessity to explicitly
account for the physically based lower bound of zero runoff.
Any shift in the mean of a given climate condition was tested using a t-test at the 95 percent
confidence level to determine if the difference in the mean was statistically different from the
neutral condition. The shifts in the mean volume runoff, or lack thereof, are explained by
investigating at the corresponding changes in precipitation and temperature patterns throughout
the Basin.
Temperature and precipitation data were taken from Department of Energy Precipitation and
Temperature Anomaly data set from the Climate Diagnostics Center. This is a global, gridded
data set that consists of monthly anomalies and is based on surface measurements between 1851
and 1990. Normal precipitation in this study was considered to be ±10 mm for the season.
Normal temperature is considered to be ±5 °C for the season.
4. Discussion of results
Before we look at the variability of the January through July volume runoff with the climate
indices, it is prudent to examine the auto-correlation function of the runoff. Any significant
10
auto-correlation in the runoff must be removed before further variability analysis can take place.
Figure 7 shows the auto-correlation function of January through July runoff.
1.0
0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
0 2 4 6 8 10
Lag (years)
The auto-correlation coefficient is insignificant beyond lag-0 and was neglected in all further
analysis. The annual scale variability associated with the SOI suggests there may be some
statistically significant correlation with the January through July runoff. Figure 8 shows the
cross-correlation functions of the seasonal SOI with runoff.
0.5 0.5
(c) (d)
0.4 0.4
0.3 0.3
0.2 0.2
0.1 0.1
0.0 0.0
-0.1 -0.1
-0.2 -0.2
-0.3 -0.3
0 2 4 6 8 10 0 2 4 6 8 10
Figure 8 – Lagged (in years) cross-correlation functions of January through July runoff at The Dalles Dam with
seasonal average SOI for Dec-Jan-Feb (a), Mar-Apr-May (b), Jun-Jul-Aug (c) and Sep-Oct-Nov (d).
11
The cross-correlation with seasonal SOI is insignificant in the long-term, however patterns are
evident in Figure 8 suggesting some near-term correlation. It is important to recall in
interpreting the seasonal patterns that the water year begins in October, or Figure 8(d).
The strongest cross-correlation appears in the summer and autumn seasons. The correlation is
significant to a lag of two years in the summer. Lack of correlation in subsequent seasons at
lag-1 suggests the correlation at lag-2 in the summer and autumn may have no basis. The lag-1
correlation in the summer is noteworthy. There appears to be a carryover effect from an ENSO
event that builds in the summer into the following autumn and winter. There is also a weakly
significant correlation in the spring at lag-1. This pattern suggests that for an ENSO event to
have a significant impact on the runoff, it must be well developed three to six months prior to the
snow accumulation seasons.
Similarly, we may look at the cross-correlation function with respect to the seasonal value of the
PDO. Figure 9 shows this function. In this case, there is not a significant cross-correlation
beyond lag-0 and is entirely insignificant in the autumn. Significant correlation exists at lag-0
for other seasons. The negative correlation coefficient supports the hypothesis that the runoff is
inversely related to the PDO phase.
0.3 0.3
(a) (b)
0.2 0.2
0.1 0.1
0.0 0.0
-0.1 -0.1
-0.2 -0.2
-0.3 -0.3
-0.4 -0.4
-0.5 -0.5
-0.6 -0.6
0 2 4 6 8 10 0 2 4 6 8 10
0.3 0.3
(c) (d)
0.2 0.2
0.1 0.1
0.0 0.0
-0.1 -0.1
-0.2 -0.2
-0.3 -0.3
-0.4 -0.4
-0.5 -0.5
-0.6 -0.6
0 2 4 6 8 10 0 2 4 6 8 10
Figure 9 – Lagged (in years) cross-correlation functions of January through July runoff at The Dalles Dam with
seasonal average PDO for Dec-Jan-Feb (a), Mar-Apr-May (b), Jun-Jul-Aug (c) and Sep-Oct-Nov (d).
The normal distribution was fit to the set of runoff values for each climate condition. Figure 10
shows the probability density functions (PDF) for each climate condition.
12
The neutral condition has a mean of 104 million acre-feet (MAF). The most dramatic shifts in
the mean of the distributions are when the warm ENSO event occurs during a cold PDO and vice
versa. Applying the t-test on the means of the different climate conditions compared to the
neutral condition, means of the El Niño during warm PDO and La Niña during cold PDO are not
statistically different from the neutral condition at the 95 percent confidence level. This
hypothesis is rejected for the remaining two climate conditions. The mean January through July
volume runoff for the El Niño during cold PDO and La Niña during warm PDO are 88.5 MAF
and 130 MAF, respectively.
In order to give some physical justification to these results, it is necessary to look at the climate
signatures associated with these coupled events. As stated earlier, one of the primary signatures
of the PDO are the sea level pressure anomalies in the northern Pacific (Mantua, et al.,1997;
Gershunov and Barnett, 1999; Overland, et al., 1999).
0.025
0.020
6 -1
Probability (ac-ft x10 )
0.015
0.010
0.005
0.000
0 20 40 60 80 100 120 140 160 180 200
6
Jan - Jul Volume Runoff at The Dalles Dam (ac-ft x10 )
Gershunov and Barnett (1999) showed the sea level pressure anomalies in the northern and
eastern Pacific to be most stable in during El Niño/cold PDO and La Niña/warm PDO events.
This stability allows any resulting precipitation and temperature shifts to become stable and more
persistent during the event. This is most important in the Columbia River Basin from autumn
through spring. Temperature and precipitation patterns that become established during this time
frame dictate the amount of snowfall that will occur during the accumulation season and the way
the snowpack will melt in the spring. Figures 11 through 14 illustrate the deviations from
13
normal precipitation and temperature during constructive phases of PDO and ENSO events.
These plots are largely qualitative indicating the relative frequency when the majority of the
region specified was normal (0), above (+) or below (-) normal with respect to each parameter.
There is a very predominant shift toward above normal precipitation and below normal
temperatures in La Niña episodes during warm (negative) PDO in all seasons. This has the
affect of extending the accumulation season while at the same time providing above normal
snowpack. Summer is included for completeness and has the least impact on the January
through July volume runoff.
El Niño events during positive PDO exhibit a mixed shifting in the precipitation pattern. The
precipitation anomalies do not lead to any conclusions by themselves. The temperature
anomalies show a slight shift toward above normal temperatures in the autumn and a very drastic
shift in the winter and spring. This is the driving mechanism causing the reduction in the mean
volume runoff. The precipitation may show no clear signal as to which direction it may shift, but
the temperature shift to above normal during the accumulation season suggests that snow levels
will decline as a result of higher temperatures. More of the late autumn and winter precipitation
will fall as rain, reducing the available snowmelt that drives the January through July volume
runoff. This also reduces the fraction of the basin that is in a frozen ground condition, allowing
more of the precipitation to be infiltrated over the season.
14
Figure 11 -- Historic Precipitation Anomalies Associated with La
Niña Events during Negative PDO
(anomalies based on 1851-1990 period of record)
Sep-Oct-Nov
0 0 0
+ 0 - + 0 - + 0 -
Dec-Jan-Feb
0 0 0
+ 0 - + 0 - + 0 -
Mar-Apr-May
0 0 0
+ 0 - + 0 - + 0 -
Jun-Jul-Aug
Above GCL Above LWG West Basin
1 1 1
0 0 0
+ 0 - + 0 - + 0 -
15
Figure 12 -- Historic Precipitation Anomalies Associated with El
Niño Events during Positive PDO
(anomalies based on 1851-1990 period of record)
Sep-Oct-Nov
0 0 0
+ 0 - + 0 - + 0 -
Dec-Jan-Feb
1
Above GCL Above LWG West Basin
1 1
0 0 0
+ 0 - + 0 - + 0 -
Mar-Apr-May
1
Above GCL Above LWG West Basin
1 1
0 0 0
+ 0 - + 0 - + 0 -
Jun-Jul-Aug
1
Above GCL 1 Above LWG 1 West Basin
0 0 0
+ 0 - + 0 - + 0 -
16
Figure 13 -- Historic Temperature Anomalies Associated with La
Niña Events during Negative PDO
(anomalies based on 1851-1990 period of record)
Sep-Oct-Nov
Frequency
0.6
Frequency
0.6 0.7
Frequency
0.4 0.5
0.4
0.4
0.3
0.5 0.2
0.5
0.3
0.2
0.5 0.3
0.2
0.1 0.1
0.1
0 0
0 0
0 0
+ 0 - ++ 00 - ++ 00 -
Dec-Jan-Feb
0.9
Above GCL Above LWG West Basin
Frequency
1
Frequency
0.7
Frequency
0.8 0.6
0.7 1 0.6
1 0.5
0.6 0.5
0.5 0.4
0.4
0.5
0.4
0.2 0.2
0.2
0 0
0 0
0 0
++ 00 -- + 0 - ++ 00 -
Mar-Apr-May
0 0 0
+ 0 - + 0 - + 0 -
Jun-Jul-Aug
Above GCL Above LWG West Basin
1 1 1
0 0 0
+ 0 - + 0 - + 0 -
17
Figure 14 -- Historic Temperature Anomalies Associated with El
Niño Events during Positive PDO
(anomalies based on 1851-1990 period of record)
Sep-Oct-Nov
0 0 0
+ 0 - + 0 - + 0 -
Dec-Jan-Feb
0 0 0
+ 0 - + 0 - + 0 -
Mar-Apr-May
0 0 0
+ 0 - + 0 - + 0 -
Jun-Jul-Aug
Above GCL Above LWG West Basin
1
1 1
0.5 0.5 0.5
0 0 0
+ 0 - + 0 - + 0 -
18
5. Summary of conclusions
The presence of decadal oscillations is well documented. Gershunov and Barnett (1998)
concluded that the long wave pattern of the PDO effectively modulates the impacts associated
with ENSO events. Using a similar methodology, this study has shown that the constructive and
destructive phasing of PDO and ENSO also translates to significant shifts in seasonal
precipitation and temperature anomalies. These then translate into significant shifts in the
distribution of January through July volume runoff at The Dalles Dam. This study concluded
that statistically significant shifts in the distribution of volume runoff in this time frame only
occur during El Niño/cold PDO and La Niña/warm PDO events. Figure 10 showed that although
the variance increased with alternate phasing of ENSO and PDO when compared to the neutral
condition, the means were not statistically different.
So how can this information be utilized? The intent of understanding the statistical behavior of
volume runoff as it varies with oscillating climate indices is to reduce the amount of forecasting
uncertainty. If the forecaster has reason to believe the climate anomalies for a given forecast
season will respond according to the distributions suggested here, he/she may be able to use that
information to refine probabilistic forecasts by eliminating “unlikely” runoff scenarios.
Hamlet and Lettenmaier (1999) documented a methodology where information regarding ENSO
and PDO, along with long-lead climate forecasts, may be used to conditionalize probabilistic
forecasts. There continue to be improvements in long-range forecasting an climate prediction
that can be used to further condition forecasts. Given the value now placed on Columbia River
water and the necessity to accurately plan the operation of the reservoir network, knowledge of
the statistical behavior of runoff as a function of climate indices will help to maximize the use of
the river.
19
5. References
Livezey, R. E. and T. M. Smith, 1999: Covariability of aspects of North American climate with
global sea surface temperatures on interannual to interdecadal timescales. J. Climate, 12, 289-
302.
Mock, D. R., 1981: The southern oscillation: historical origins. Unpublished term paper,
University of Washington, Seattle, Washington, 9 pp.
National Oceanic and Atmospheric Administration, 1994. El Niño and Climate Prediction.
Reports to the Nation, Spring 1994, 3, 25 pp.
Overland, J. E., J. M. Adams, and N. A. Bond, 1999: Decadal variability of the Aleutian Low
and its relation to high-latitude circulation. J. Climate, 12, 1542-1548.
Peixoto, J. P. and A. H. Oort, 1992: Physics of Climate, AIP Press, New York.
Salas, J. D., J. W. Delleur, V. Yevjevich, and W. L. Lane, 1980: Applied Modeling of Hydrologic
Time Series, Second Printing, Water Resources Publications, Littleton, Colorado.
20