Spatial Indices of Upwelling

1) Coastal Topography

Premise

• Average spatial patterns of coastal ocean processes are strongly influenced by topography and bathymetry

How to quantify coastal topography?

‘Coastal Anomaly’

Broitman and Kinlan 2006 MEPS, In press

Bays Headlands

(Embayments)

0

-1.0

+1.0

(Headlands)

Norm

alized residual from sm

oothed coast

COASTAL STRUCTURE

Alo

ng

-co

ast

Dis

tan

ce (

km)

Scale of Smoothing (km)

Smoothing Scale=1000 km

COASTAL STRUCTURE

Smoothing Scale=50 km

COASTAL STRUCTURE

S. Africa WNAChile

Coastal Topographic IndexCoastal Topographic Index

• Easily calculated• Data readily available• Arbitrarily high

resolution• Can index processes

at multiple scales

• Indirect index; index of upwelling or any other process depends on nature of coupling

• Not dynamic (no time component)

• Oversimplified?

ConsPros

SST

Spatial Indices of Upwelling

2) SST-derived indices

time

longitude

lati

tude

Coastal SST

Latitudinal mean at each pixel

Topographic Index and dSST Spatial Series

C) Topo vs. Δ SST C) Topo vs. Δ SST

Correlation between Topo and dSST indices

Correlation with Bakun Regional Index

Ecosystem Structure

Spatial patterns of kelp and phytoplankton

Biomass

Areal cover

chl-a vs. kelp

Spatial Cross-Correlations

A) Topo vs. kelpA) Topo vs. kelp

B) Topo vs. chl-aB) Topo vs. chl-a

C) Topo vs. Δ SST C) Topo vs. Δ SST

Spatial Cross-Correlations

Jones et al 1988, Cont Shelf Res

Approximately the same characteristic scaleApproximately the same characteristic scale

Alongcoast scales: Variogram Analysis

 

Variable Nugget (SE) Sill (SE) Range (SE) (km)

Kelp 0.35 (0.09) 0.65 (0.09) 188 (100)

Chl-a 0.13 (0.08) 0.87 (0.08) 178 (34)

ΔSST† 0.008 (0.007) 0.992 (0.007) 151 (18)

Coast Anomaly 0.04 (0.03) 0.96 (0.03) 161 (40)

Alongcoast scales: Variogram Analysis

Onshore Community Patterns

South Africa58 sites along 2000 km of coast

WNA73 sites

along 6000 km of coast

Chile26 sites

1000 km of coast

100 200 300 400 500 600 700 800 900

-0.6

-0.4

-0.2

0

0.2

0.4

0.6

xcorr: Urchin GSI vs. Topographic Index

smoothing scale for topographic index (km)

corr

ela

tion

co

eff.

What scale of coastal features matter to the What scale of coastal features matter to the process you’re interested in?process you’re interested in?

Correlation between variable of interest and topographic index at each smoothing scale

-25 -20 -15 -10 -5 0 5 10 15 20 25

-0.2

0

0.2

0.4

0.6

Along-coast Lag Distance (km)

Cro

ss-C

orre

lati

onScale of pattern is just part of the questionScale of pattern is just part of the question

Peak @ 15 km in the poleward direction

Correlations with Coastal Anomaly – South AfricaCorrelations with Coastal Anomaly – South Africa

Correlation with Coastal Anomaly – MusselsCorrelation with Coastal Anomaly – Mussels

CHILE SOUTH AFRICA

Chile – Upwelling-Recruitment Dynamics Seen in Bare Space?

Coastal Topography and Physical ForcingCoastal Topography and Physical Forcing

CHILE SOUTH AFRICA

Correlation with Coastal Anomaly: MusselsCorrelation with Coastal Anomaly: Mussels

CHILE SOUTH AFRICA

Correlation with Coastal Anomaly: Trophic GrpsCorrelation with Coastal Anomaly: Trophic Grps

CHILE SOUTH AFRICA

Long-term Average Recruitment vs. Topography at 21 Coastal Sites

Green=Chthamalus

Black=Balanus

Balanus

r2=0.84, p<0.0001

Multiple regression with:

Topo10km, Topo150km, Coastal Coordinate (linear latitudinal trend)

Other processes correlated with topography

Aside… Hierarchical Bayesian Modeling for Data Integration & Complex Hypothesis Testing

Santa Barbara

Offshore Ocean Nearshore Ocean

Site 3

Site 4

Site 2

Site 1

Offshore

Larval

Pool

Santa Barbara

Offshore Ocean Nearshore Ocean

Recruitment

Offshore

Larval

Pool

Sea Temp (SST)

ChlorophyllA

Current

Current

Current

Current

Recruitment

Recruitment

Recruitment

16 points in time, each describing previous month

0, 1, m2

Rit~Poisson(FtHitei)

k,S

Likelihood:

Parameters:

Priors:N(0,10-3)

vague

Data: SSTt-TS GitRit

IG(10-3,10-3)vagueN(0,10-3)

vague

Links: Ft = k + S TS*SSTt-TS Logit(Hit)= 0+1Git+g

g ~N(0,m2)

i~N(0,2)

IG(2,10)informative

2

TS=0

2

i = space index , t = time index

OffshoreLarval Pool

OnshoreDelivery

Combine theseto get Recruitment

Chl , gate+errror, Gamma

SST , gate+errror, Gamma

SST , gate+errror,

Poisson

SST , gate+errror,

Poisson, site effects