Transcript S5_memox

Summary session 5
Structure & outcomes
• Before coffee break
– Surface subsurface climate variability
• After coffee break
– Oxygen & nutrient
– Fisheries
• Outcomes
– Updated synthesis of climate processes in the N-Pac
– Interpret IPCC5 projections in the context of N-Pac
Cliamte dynamics
Mat Collins
• Anomalies 1986-2005 average.
• Transient climate response is likely in the
range 1-2.5C and extremely unlikely greater
than 3c. (box 12.2)
• My interest: Tropical atmospheric circulation is
weakened, subtropics creeps poleward.
• Five consistent models show faster reduction
of sea-ice.
Mat Collins 2
• SAT change 2081-2100 minus 1986-2005
– Global mean, land-sea contrast, polar amplification
(themodynamic processes) are removed. Then PICES
region is fastest warming
– Dynamical SAT changes seem much smaller – although
of crucial in the tropics (Xie et al. 2010)
• Vertical winds (omega) = precipitation (P) /
humidity (q)
– This relation is good for global, but not in Asian
– Chadwick et al. (2013)
Mat Collins 3
• Global mean changes are sub-ClausiusClapeyron,
• Confidence in storminess is low.
Mat Collins 4
• Chapter 14: climate phenomne and their
relevance for future regional climate change
• Any specific projected change in ENSO
variability in the 21st century remains low.
Associated precipitation changes will be
intensified.
• Extreme El Ninos (Cai et al, Nature Climate
Change 2014) will be doubled.
Mat Collins
• Changing El Nino teleconnections associated
with Walker Circulation Change
– Chung et al. Cdyn 2014
– Power et al. Nature 2013
Mat Collins
• Global warming ‘pause’ or ‘hiastus’ (Box 9.2)
• Chris Roberts, Matt Palmer, MO
– Pseudo ensemble hiatus from piControl + forced
response
– 2001-2010 negative trend (11.5%). Hiatus can
occur by chance.
– 2001-2015 negative trend (0.8%)
– Composite mean pattern of hiatus has PDO
pattern in the Pacific (with amplitudes in Indian
and Atlantic Oceans), consistent with observation.
Mat Collins
• Trade wind strengthening: England et al.
Nature Climate Change 2014
– See also Kosaka and Xie 2013
Mat Summary
• Large-scale thermodynamic response of temp
relatively well understood; global + land/sea + polar
amplification. Could build a quantitative theory.
– Interestingly, after removal of them SAT change is largest
PICES region. Can be associated with polar amplification
(just masking)
• Precip: weakened circulation and increase humidity
• Challenge is to combine information from imperfect
models with our (sometimes quite good)
understanding of physical processes.
Nate Mantua
• 1998-2013 minus 1977-1997
• Max corr. Is 0.53 Beween negative pole in C-NPac and pole in E-N-Pac.
• ER-SST (1900-2012) USHCNv2 GHCNv3 (19002012), flux SODA (1900-)
• EOF east of dateline (max corr 0.82): horse
shoe-oval still (ship track is good in this region)
• SLP PC1 max at 150W, 35N (shifted east than
NPI)
Nate Mantua 2
• SST = 0.81 SST(t-1) + 0.27 SLP(t) + eps(t) (monthly)
– Good reconstuction =0.79 (monthly)
• SST tendency due to latent heat flux, horizontal Ekman
transport, vertical T-advection
• Coastal air-temperature はannual meanでよく一致.
• SLP forcing accounts for all of the SST trend & most of
SAT trend (from northern California to Washington)
• Circulation changes looks to be free (not forced) from 4
CMIP5 models (31 members), consistent with Deser et
al. (2012, 2014), Oshima et al. (2012)
Mat Newman 1
• PDO is reddened ENSO + lagged KOE due to
Rossby waves
– PDO (yr) = 0.6 PDO (yr-1) + 0.6 ENSO(yr) + eps
• Multivariate red noise (Ornstein-Uhlenbeck)
– dx/dt = B x + fs
– If B is nonnormal (not symmetric), transient anomaly
growth is possible even through exponential growth is
not.
– Detemine B with LIM from lagged covariance (space
and time) statistics of X
Matt Newman 2
• LIM, State variable
– Pacific SST
– Tropical thermocline depth (20C isodepth)
– N-pac mixed layer (30-100 m)
• Seasonality is ignored
• B determined from 3 month lag
• Result is flat to 0.02 cycle/yr
Mat Newman 3
• After removal tropical forcing, Oyashio SST
anomalies stand out
• Trend + KOE-related + decadal ENSO (aka CP
ENSO)
• PDO may generally provide only weak forcing
of the atmosphere (Kumar et al. 2012)
• Smirnov et al. (2014) HR model KOE -> G
Alaska
Bo Qiu 1
• Unstable states lagged by 4-yrs to PDO, 3-yr to
NPGO. Due to Rossby waves.
– Stable -> strong gradient?
• KE index: sum of four indices, extended using
OFES to 1956 (just SSH anomaly in KE region).
• two way prediction model has a better skill 45 years
• Northward-shifted and stable KE will last
further several years. (Qiu 2014 JC)
Nikas 1
•
•
•
•
Taguchi & Schneider JC 2014 submitted
0-300 m HC propagate eastward
Higher vertical mode
Density-spiciness split
– δT = d[T]/dρ δρ + δ T’ (vertical movement +
spiciness)
– Spiciness is dominant
– Meridional displacement produces spiciness
Niklas 2 summary
• Westward propagating SSHa
• Eastward propagating OHCa
– Generation meridional migration
– Mean advection
– Damping produces higher vertical modes
Krovnin 1
• Azores high & icelandic low longitudes and
Z500
– 1950-1977:
– 1978-2005:
• Atlantic influence -> Eurasia -> North Pacific
• Ocean tracers as integrators
– Long memory -> red noise (Ito & Deutsch 2010;
2013)
– Double integration (Di Lorenzo & Ohman 2013)
Taka Ito 1
• Declining oxygen
– Coastal hypoxia (Diaz and Rosenberg 2008; Bograd
et al. 2008)
– Expanding/shaling oxygen minimum zone (OMZ)
(Stramma et al., 2008; 2010; Helm et al. 2011)
• What controls subsurface O2
– Oxygen <-> productivity -> respiration = O2
utilization
Taka Ito 2
• Helm et al. (2011) WOCE vs historic data 19921970 in 100-1000 m
• 1990-2008 minus 1960-74 diff (Stramma et al.
2010) O2 decline <- AOU increase in the topical
oceans
• Warming
• Weakening ocean vertical exchange
– Weaker physical O2 -> weaker nutrient upwelling ->
weaker productivity
• Tropics incease oxygen (Cocco et al. 2013)
Taka Ito 3
• Tropical O2 paradox
– There are a number of suggestions including
decadal variability of Deutsch et al. (2011)
– This study : anthropogenic nutrient flux
– Mineral Fe : acidification aerosol makes solble iron.
– Anthropogenic Fe decomposition promoves ocean
productivity (Krishnamurty 2009?)
Taka Ito 4
• Global 1x1, L23 (MITgcm –Ecco)
• 2 biogeochemistry models
• Soluble Fe flux 100 years-> macro nutrient
down -> productivity up -> O2 down
– This effect is not in CMIP5 models.
• This can explain the reduction of topical
oxygen reduction.
Sung Yong Kim (KAIST)
• CalCOFI region
• 6 harmonics T, T/2, T/3 … T/6
• Fitted with seasonal cycle, ENSO, PDO, NPGO,
trend
Di Lorenzo
• Good coastal hypoxia map (Diaz and
Rosenberg Science 2008)
• Isopycnal 26.5
• d O’shelf /dt = -(1+r) w’ [Osub] + [w] Osub’ –
eps O’shelf
• Northward poleward undercurrent bring low
O2 (high NO3) from the south
• Gyre brings high O2 (low NO3)
Di Lorenzo 2
• Salinity anomalies in ECMWF ORA3 on
isopycnal 26.5 is used as a proxy of Oxygen
– Good in ColCOFI and Statin P
– CalCOFI papers: Bograd et al 2008; Koslow et al
2011; Peterson et al. 2013
• Subtropical gyre in the central Pacific N-Pac
oxygen (index) can predict CalCOFI oxygen
with lag of 8-10 years
Art Miller 1
• CCE-LTER founded in 2004
• Rykaczewki and Checkley (2008): off-shore
Ekman pumping is important as coastal
upwelling
• Warm water coming from depth
Shin-ichi Ito
• Pacific saury: subarctic feeding, subtopics
spawning, Surf riding theory
• Examine using NEMURO.FISH
• Previous 3box -> Euler model
– Velocity (Ambe 08) 1/3 deg, temperature
(MODIS/Terra 1/12 deg), Chl-a SeaSiFS (1/12 deg)
• Feeding migration: highest grwoth, spawning
migration,
• Order is good, but interannual variation 20022009 is not good enough
– Adjust westward migration speed. ->
Shin-ichi Ito 2
• Westward migration is low in 2004 high in
2008
– SST west is warmer (cooler in east) -> faster
westward migration
– Saury may see light (cloudiness)
Sukgeun Jung
• Pacific cod (cold water spices) in Korea waters
increase from 1999 to 2011.
• Common squid (warm water) dominant
• Increase Sardine (warm) and Cod (cold) since
the late 1990s.
• Corresponding analysis -> dim1 & dim2
– Dim1 is correlated to temp 100-500 m
Sukgeun Jung, Summary
• Upper layer (50-100)
– Temp increase 1988-89
– Warm water species (Sardin -> Squid)
• Deep layer 100-500 m
– Cold water species (Filefish -> Cod & Herring)
• Time lag between the shifts in the upper and
dep layer was 5-6 years.