Variability of United States Runoff and its Climate Teleconnections

1
Variability of United States Runoff and its
Climate Teleconnections E.P. Maurer(1), D.P.
Lettenmaier(2) and N.J. Mantua(3) (1)Department
of Civil Engineering, Santa Clara University,
Santa Clara, CA 95053 (2)Department of Civil and
Environmental Engineering, Box 352700, University
of Washington, Seattle WA 98195 (3)Climate
Impacts Group, Joint Institute for the Study of
the Atmosphere and Ocean, Box 354235, University
of Washington, Seattle, WA 98195
Poster H22B-0917
3
Principal Components of Distributed Runoff and
Climate Teleconnections
Abstract Understanding the space-time
variability of runoff has important implications
for climate because of the linkage of runoff and
evapotranspiration, as well as practical
implications for the prediction of drought and
floods. In contrast to climate variables like
precipitation and temperature, there has to date
been relatively little work evaluating climate
teleconnections of runoff, in part because of the
absence of data sets that lend themselves to
commonly used techniques in climate analysis like
principal components analysis (PCA). We examine
the space-time variability of runoff over North
America between latitudes 25 and 53 degrees
north, which is the spatial domain of the North
American Land Data Assimilation System (N-LDAS)
for which a 50-year retrospective data set of
runoff and other land surface water cycle
variables has been produced at 1/8 degree
latitude-longitude spatial resolution. Past
efforts to investigate spatial patterns of runoff
variability in the United States have utilized
streamflow observations, which have three
important drawbacks first, the location of the
observations is highly non-uniform, second,
routing effects associated with the conversion of
runoff to streamflow confound the interpretation
of the observed variables, and third, river
impoundments and diversions affect the
observations by varying amounts. By using derived
spatially distributed runoff which represents
natural (no effects of routing, or diversions or
impoundments) conditions, we are able to avoid
these shortcomings. Using the 50-year 1/8 degree
data set, accumulated to monthly amounts, we
determine climatic teleconnections using common
climate indices (such as Niño3.4 and NAO), by
season for lead times of months to a year. High
and low values of climatic indices are evaluated
separately, which allows independent
interpretation of the telconnections of different
climatic anomalies to the runoff variability. We
identify patterns of runoff variability that are
not revealed with observed datasets, especially
where observations are sparse. A greater number
of significant climate-runoff relationships are
exhibited for runoff patterns on the east and
west coasts and southern interior, with fewer for
Northern interior runoff patterns in the Upper
Mississippi and Missouri river regions. Rarely do
both the positive and negative phases of any
climatic index show significant teleconnection
with a particular pattern of runoff variability,
lead time and season.
Climate signals exhibiting 95 significant
teleconnection to the indicated pattern of runoff
variability for the designated season and lead
time. A () indicates the positive phase of this
index is significantly correlated with the runoff
pattern, and a (-) indicates the negative phase.
Rows indicate runoff season, and columns indicate
lead time in seasons.
PC loadings are plotted below, illustrating
coherent modes of runoff variability 10 PCs were
rotated (using the Varimax technique). Variability
is revealed in the northern Great Plains that is
not detectable in studies using sparse gauge
observations. Two of these patterns are the Upper
Missouri/Canadian prairie and Lower Missouri,
with both regions experiencing devastating floods
within the past 10 years.
Season Lead 0 Lead 1 Lead 2 Lead 3 Lead 4
East/Mid-Atlantic/Gulf East/Mid-Atlantic/Gulf East/Mid-Atlantic/Gulf East/Mid-Atlantic/Gulf East/Mid-Atlantic/Gulf East/Mid-Atlantic/Gulf
djf Nino3.4(-) nino3.4(-), pdo(-) ao(-),nao(-)
mam nino3.4() nino3.4()
jja Pdo() pdo(-)
son ao(-) ao(-)
Far West/Great Basin Far West/Great Basin Far West/Great Basin Far West/Great Basin Far West/Great Basin Far West/Great Basin
djf np()
mam ao(-)
jja ao() nao(-), np(),pdo(-) nino3.4(-)
son ao() np(-)
Ohio/Tennessee Basin Ohio/Tennessee Basin Ohio/Tennessee Basin Ohio/Tennessee Basin Ohio/Tennessee Basin Ohio/Tennessee Basin
djf np(-) ao(-)
mam pdo(-) nino3.4(-)
jja pdo(-),pdo() np(),pdo(-)
son ao()
Southern Plains Southern Plains Southern Plains Southern Plains Southern Plains Southern Plains
djf pdo(-) nino3.4(-), pdo(-) nao()
mam nino3.4(), pdo(-) nino3.4(), pdo() nino3.4()
jja pdo(-) ao() ao(-)
son Nino3.4(-)
New England/Quebec New England/Quebec New England/Quebec New England/Quebec New England/Quebec New England/Quebec
djf np(-)
mam Nao() ao(-),nao(-) nao(-)
jja nao(-) pdo(-) pdo(-) nao(-)
son Pdo() ao(), nino3.4() ao(),pdo(-)
Southwest/Mexico Southwest/Mexico Southwest/Mexico Southwest/Mexico Southwest/Mexico Southwest/Mexico
djf np(-) nino3.4(-) pdo(-) pdo(-) pdo(-)
mam nino3.4(-), nino3.4(), pdo(-) ao(),nao(-), nino3.4(-), nino3.4() nino3.4(-) nino3.4(-)
Upper Mississippi Upper Mississippi Upper Mississippi Upper Mississippi Upper Mississippi Upper Mississippi
djf pdo(-) np(-) pdo()
Upper Missouri/Canadian Prairie Upper Missouri/Canadian Prairie Upper Missouri/Canadian Prairie Upper Missouri/Canadian Prairie Upper Missouri/Canadian Prairie Upper Missouri/Canadian Prairie
djf nao(-)
mam pdo() pdo() pdo() ao()
jja
Great Lakes Great Lakes Great Lakes Great Lakes Great Lakes Great Lakes
mam ao(-) nao()
Pacific Northwest Pacific Northwest Pacific Northwest Pacific Northwest Pacific Northwest Pacific Northwest
jja pdo(-) pdo(-) nino3.4(-), pdo(-) nino3.4(-), nino3.4()
Lower Missouri Lower Missouri Lower Missouri Lower Missouri Lower Missouri Lower Missouri
jja ao()
son ao()
Runoff season
Runoff variance explained by pattern
For seasons and locations such as these examples,
where no climate indices show a significant
relationship to runoff, any land surface
influence may be more important (see box 4 below)
1
Motivation and Research Questions

• Improved understanding of hydrologic variability
  is needed to advance prediction of floods and
  droughts.
• Principal components (PC) analysis (PCA) provides
  a tool to characterize variability of earth
  system variables.
• PCA has been used to examine streamflow
  variability, using observed gauge flows to
  determine modes of variability.
• PCA on spatially distributed runoff can avoid the
  problems in using observed streamflows, such as
  non-uniform spatial distribution of observations,
  effects of river impoundments or diversions, and
  routing through the contributing basin.
• PCs of spatially distributed runoff be examined
  to reveal climatic teleconnections and
  relationships of runoff to the land surface, to
  show potential sources of runoff predictability.
• These inspired the following questions
• Do different patterns of runoff variability
  emerge with spatially distributed runoff data as
  compared to the irregularly-spaced stream gauge
  data?
• What are the SST and climatic teleconnections to
  the patterns of runoff variability?
• Where and at what lead times are the soil
  moisture and snow conditions significantly
  related to the modes of runoff variability?

10 highest and lowest values of anomaly in
50-year record selected PC magnitude for each
set of 10 compared for significant difference
from 0

• Climate Teleconnection Analysis
• Significant climate teleconnections at different
  lead times for each pattern are shown in the
  table to the right.
• High and low signals were evaluated separately

• Climate teleconnections are generally weaker at
  longer lead times.
• More climate teleconections exist for Southern
  patterns in DJF and MAM, and for Northern
  patterns for MAM and JJA.
• In most cases where one phase of a climate signal
  has a significant teleconnection its opposite
  does not.
• AO and NAO are significantly related to runoff at
  long lead times, and can provide additional
  runoff predictability.

4
5
Bonus Section Land Surface Influence on Runoff
Variability
Summary and Conclusions
Soil Moisture Low magnitude DJF runoff has high
SM relationship. Low values (dry conditions) show
stronger relationships to runoff.
Snow water equivalent shaded blocks indicate
grid cells with significant relationships between
SWE and runoff pattern. Relationships follow snow
melt snow produces MAM runoff variability in the
East, MAM and JJA runoff variability in the West.

• By including a large domain and using a
  spatially distributed runoff data set, runoff
  variability was revealed that could not be
  obtained directly using observed streamflow
• Through a lead of 1 season, snow water equivalent
  is a significant predictor for runoff
  variability, and is strongest in the Pacific
  Northwest and Far West.
• Soil moisture provides widespread predictability
  at short lead times, and is more frequently a
  significant predictor in its dry state than wet.
• By a lead of 1 season, soil moisture has its
  primary influence on runoff variability in the
  Western U.S.
• Both soil moisture and snow explain significant
  portions of runoff variability where climate
  signals do not, showing that the two sources of
  predictability can be complementary.

2
Data and Domain
The runoff, soil moisture, and snow water
equivalent data used are those presented by
Maurer et al. (2002 J. Climate 15(22)3237-51),
derived by driving a hydrologic model with
gridded observed meteorology. The output
including runoff, soil moisture, and snow,
compared favorably to available observations.
The domain is North America between latitudes
53ºN and 25ºN, coinciding with the Land Data
Assimilation System-North America (N-LDAS)
region. Climate indices were obtained from
publicly available sources and include Niño 3.4
(for ENSO signal) Pacific Decadal Oscillation
index (PDO) Arctic Oscillation index (AO) North
Atlantic Oscillation index (NAO) North Pacific
index (NP), which mirrors the Pacific North
America index (PNA)