1. Characteristics of large scale climate indices
Nate Mantua
University of Washington
Aquatic and Fishery Sciences GLOBEC/PICES/ICES ECOFOR Workshop, Friday Harbor
Sept 7-11

2. Motivation for identifying large-scale climate patterns?
Capture large fractions of field variability with a small number of spatial patterns and associated time series
Large-scale climate indices are tailored to questions of large-scale climate variations, not the climate variability of any single location or any particular region

3. Modes of variability in the atmosphere
Various analyses of historical data fields (like Sea Level Pressure or upper atmosphere pressure fields) have identified a relatively small number of geographically fixed patterns that explain significant fractions of the total monthly or seasonal field variance (mostly in N. Hemisphere winter). Prominent modes include:
The North Atlantic Oscillation and Artic Oscillation
The Aleutian Low/Pacific North America Pattern
The North Pacific Oscillation
The Southern Oscillation
atmospheric modes, generally speaking, appear to be intrinsic to the atmosphere
Some arising in part as instabilities in the mean climatological flow and/or arising from eddy/mean flow interactions
However, they can be excited via external forcing (solar, greenhouse gases, volcanic aerosols, stratospheric ozone depletion …), or changes in surface boundary conditions (SST, sea-ice), which can alter their temporal patterns and potentially enhance their predictability

4. November-April SLP change
Cool season Aleutian Low variability (Nitta and Yamada 1989, J. Met. Soc. Japan; Trenberth 1990, BAMS)
November-April SLP change
( ) - ( )
The NP index is the area weighted SLP anomaly from 30-65N, 160E-140W (Trenberth and Hurrell 1994, Clim Dyn)

5. A simplified ocean’s red noise response to the atmosphere’s white noise forcing (Frankignoul and Hasselman 1977: Tellus)
Here, the predictability or persistence of SST anomalies is limited to the timescale associated with the thermal inertia of the mixed layer
Ocean mixed-layer acts as a low pass filter with an enhanced response at low frequencies, but no preferred time scale 20 40 60
SST (H=50m) 20 40 60
These mld values span the typical range of observed MLD in the extratropical oceans.
SST (H=500m) 20 40 60
year
Figure from Deser et al. 2010: Ann. Rev. Mar. Sci.

6. The PDO pattern and index are derived from an EOF analysis of monthly North Pacific SSTa from after removing the monthly global average anomaly.

7. Schematic of Pacific Oceanic Response to Decadal Forcing by the Aleutian Low
Canonical
SST Pattern
Rossby waves
2 - 5 yrs
Lagged
KOE
SST
Pattern
(Miller and Schneider, 2000, Prog. Oceanogr.)

8. Canonical Pattern of Decadal SST Response (Aleutian Low Strengthening)
Schematic
SST
Warming
SST Cooling
Driven by surface atmospheric forcing
Canonical Pattern of Decadal SST Response
Equator
From Miller, Chai, Chiba, Moisan and Neilson (2004, J Oceanogr.)

9. Lagged Pattern of Decadal SST Response (Aleutian Low Strengthening)
Schematic
SST Cooling
sCooling
Driven by thermocline changes
via wind-stress curl
Lagged Pattern of Decadal SST Response
Equator
From Miller, Chai, Chiba, Moisan and Neilson (2004, J Oceanogr.)

10. Basin-Scale Pattern of Decadal Thermocline Response
(Aleutian Low Strengtherning)
Schematic
Thermocline
Thermocline Shallowing
Deepening
Lagged response in west due to
Rossby wave propagation
Basin-Scale Pattern of Decadal Thermocline Response
Equator
The PDO SST pattern is a consequence of multiple processes
From Miller, Chai, Chiba, Moisan and Neilson (2004, J Oceanogr.)

11. ENSO Impacts on cool-season climate
Two El Niño-related processes promote warming and poleward coastal currents in the NE Pacific:
Atmospheric teleconnection: the Aleutian Low tends to be more intense, and its location shifted south and east
Oceanic teleconnection: Northward propagating coastally-trapped kelvin waves originating in the equatorial Pacific also alter nearshore currents over the continental shelf.

13. The scale issue
Vol 430, 1 July 2004
Soay Sheep population dynamics are better correlated with the NAO than with “local climate”?
local weather events drive winter mortality: yet cold temperatures, high winds, and heavy rainfall all appear as causal factors in different years
One-dimensional view of climate (e.g. temperature) is simply too narrow to capture climate impacts on Soay sheep, and the NAO index (roughly) captures many dimensions
Hallett et al. (2004) describe an interesting case study of the links between Soay Sheep population changes, local climate, and changes in the North Atlantic Oscillation (NAO) index. The NAO index is based on anomalous sea level pressure changes between Iceland and the Azores, a see-saw in atmospheric pressure that is the dominant pattern of SLP variability over the North Atlantic sector (e.g. Hurrell et al. 19xx). Over the period from 19xx-20xx, the NAO index is better correlated with Soay Sheep population variability than are indices tracking Soay Islands precipitation, winds, or temperatures, respectively. Examining the local climate records with more scrutiny than provided by correlation analysis reveals the source for this apparent paradox. The link between the NAO index and Saoy Sheep comes with a range of different weather impacts on the limited food resources that sustain Soay Sheep in the winter season. In each case, local weather events influence wintertime mortality events for Soay Sheep; in some years it is extreme cold temperatures, in others it is high winds, and in others it is heavy precipitation events that appear as causal factors for mortality events. The NAO index is modestly correlated with each of those factors, and it therefore offers a better correlation with Soay Sheep population numbers than any one local weather does index alone. A take home message from this study of climate impacts on Soay Sheep is that a one-dimensional view of climate (e.g. temperature) is simply too narrow to identify and understand climate impacts on Soay Sheep.

14. Key points
Large scale indices are not meant to represent local/regional variability in any single place
Large scale indices for atmospheric patterns typically look like white-noise, with substantial intrinsic variability
Coordinated atmospheric forcing over large regions with a broad spectrum of time scales
Large-scale indices for upper ocean patterns (PDO, NPGO, AMO, ENSO, etc.) have more variance at lower frequencies
typically integrate atmospheric forcing and involve ocean processes too – multiple processes at work, some with time lags
Large-scale indices correlate with multiple dimensions of habitat, and this may favor improved correlations with biological/ecological variables