Title: Empirical indicators of climate variability and ecosystem response since 1965
1Empirical indicators of climate variability and
ecosystem response since 1965
- Steven R. Hare1 and Nathan J. Mantua2
- 1International Pacific Halibut Commission, P. O.
Box 95009, Seattle, WA, 98145-2009. Email
hare_at_iphc.washington.edu - 2 Joint Institute for the Study of the Atmosphere
and Oceans, University of Washington, Box 354235,
Seattle, WA 98195-4235. Email
mantua_at_atmos.washington.edu
2Introduction
- Widespread agreement that a climatic regime
shift, with extensive ecosystem impacts, occurred
in the winter of 1976-1977. - The 1977 regime shift appears to follow earlier
shifts in 1925 and 1947. - Has a regime shift occurred since 1977?
- 1989 generally pointed at as most likely change
point - PDO index used to identify earlier regime shifts
winter PDO reversed in 1989 but summer/annual PDO
index did not. - Collected diverse array of time series to test
coherency of a 1989 regime shift (and revisit
1977)
3Data
- We assembled 100 indices of climate and
biological variability for the period 1965-1997. - 31 are climatic and 69 are biological time series
- Spatial range is California Current to Bering Sea
- Climate series are mostly winter averages,
biological series are annual averages. - All time series were normalized - most biological
series were log-normalized
4Methods
- Three methods used to detect regime signal
- Principal component analysis - interannual
variability - Generates climate patterns (EOFs) and temporal
indices (PCs) and amount of variance accounted
for by each EOF/PC pair. - Ebbesmeyer at al. (1991) step statistic -
interannual/decadal - Used to compute magnitude of step in composite of
all 100 time series. Two regime tests done -
around 1977 and 1989 - Difference maps - decadal
- SST and SLP averaged within each regime and then
the difference computed - done for summer and
winter
5Atmospheric Indices
6Air temperatures and Streamflows
7PDO, ENSO and SST indices
8Upwelling and miscellaneous
9Lower trophic level biology
10 Bering Sea groundfish recruitment
11GOA groundfish recruitment
12West Coast groundfish recruitment
13Alaska salmon catches
14Transition Zone salmon catches
15West Coast salmon catches
16Approximate centers of location of 100 time series
17PCA results - all data
18PCA results - separate data
PC scores - physical indices
PC scores - biological indices
19Loadings on PC1 - all data
20Loadings on PC2 - all data
21Loadings on PC1 - biological data
22Loadings on PC2 - biological data
23Loadings on PC1 - climate data
24Loadings on PC2 - climate data
25Loadings on PC3 - climate data
26Ebbesmeyer et al. regime test
27SST difference maps - winter
Regime 2 minus Regime 1
Regime 3 minus Regime 2
Regime 3 minus Regime 1
28SST difference maps - summer
Regime 2 minus Regime 1
Regime 3 minus Regime 2
Regime 3 minus Regime 1
29SLP difference maps - winter
Regime 3 minus Regime 2
Regime 2 minus Regime 1
Regime 3 minus Regime 1
30SLP difference maps - summer
Regime 3 minus Regime 2
Regime 2 minus Regime 1
Regime 3 minus Regime 1
31A 1998 sea change?
SLP
SST
Summer
Winter
32Summary and Conclusions
- Analysis of ecological and climate records
reproduces previously identified characteristics
of 1977 regime shift, and lends additional
support for (and insights into) a 1989 regime
shift - 1989 changes less widespread and clearly not a
simple reversal to pre-1977 Pacific climate and
ecology - especially important to Bering Sea and some NE
Pacific ecosystems - Arctic vortex persistently more intense, NE
Pacific and Bering Sea SSTs warmer (especially
in summer), and winter Aleutian Low circulation a
little weaker, than in 1977-88 period
33Summary and Conclusions (contd)
- within regimes, ecosystem indices are more steady
than climate indices - non-linear ecosystem responses to environmental
changes highlight the importance of ecosystem
monitoring because of a strong signal to noise
ratio - improved regime shift predictions still hinge on
an improved understanding of climate dynamics - a mechanistic understanding is required to know
what should be monitored, which models might be
useful, and whether or not there is any
predictability ...