!"#"$%&' )%*+"&,#- .

'*/&"#
lorm small groups of 2-3 people -> each group should conslsL of
people from dlerenL counLrles and/or laboraLorles
Lach group slgns up for a 8esearch ro[ecL Lo work on from
1hursday aernoon Lhrough lrlday uslng Lhe compuLer lab
faclllues.
Senlor lecLurers wlll be avallable Lo help wlLh pro[ecLs
lf you nlsh Lhe 8esearch ro[ecL early Lhen conslder oLher sclence
quesuons you mlghL address uslng Lhe avallable daLa seLs and
follow up on Lhese opuons. ulscuss posslble approaches wlLh
senlor lecLurers.
!"#"$%&' )%*+"&,#- )%"#"0,$1*0#
3:30 - 3:00 pm lrlday (or before) work on presenLauon of 8esearch ro[ecL.
1he presenLauon should conslsL of 8-10 slldes maxlmum (10 mlnuLe LoLal
presenLauon: 8 mln Lalk leavlng 2 mln C&A):-
a. Sclence ob[ecuve - whaL ls Lhe maln sclence quesuon LhaL you hope Lo
address? WhaL ls Lhe goal of your sLudy?
b. uaLa seLs - whaL ls Lhe spaual and Lemporal resoluuon of your daLa? ls lL from
models or observauons? Where ls Lhe locauon of your daLa seL (gure?). Show a gure
of Lhe raw daLa.
c. MeLhods - whaL approach dld you use Lo examlne Lhe daLa seLs (e.g. specLra?
SLausucs? auLocorrelauons?) Why dld you choose Lhls meLhod? WhaL are Lhe beneLs
and llmlLauons of uslng Lhls Lechnlque?
d. 8esulLs - Show a couple of gures of your maln resulLs, Make sure your gures
are leglble and have unlLs and legends and axes are marked and labelled.
e. Concluslons - dlscuss your maln concluslons and how Lhls answers your sclence
ob[ecuve and any posslble fuLure follow-up work.

resenLauons wlll be glven 08:30 - 12:30 on SaLurday Lo all paruclpanLs
Crlucal feedback wlll be provlded by senlor lecLurers and fellow sLudenLs
)*##/23" )%*+"&,#
45 6"$#*0$3/,7 *8 9:6 #,%$1;&$1*0 </,'/0 ,'" =0>*0"#/$0 #"$#

Sclence Cb[ecuve: 1he lnvesugauon of Lhe spaual pauern of seasonal change of Lhe LemperaLure
and sallnlLy (oxygen Loo) wlLhln Lhe waLer column
8ackground: ln AugusL 1993 and lebruary 1994 Lhe Arllndo mlxlng program measured Lhe
Lhermohallne sLraucauon of Lhe lndoneslan Seas. lor Lhe mosL parL Lhe 1994 sLauons were ln
Lhe same locauon as Lhe 1993 sLauons..
8esearch ro[ecL: map Lhe dlerences of LemperaLure and sallnlLy as a funcuon of pressure and
denslLy (slgma-o). ?ou mlghL wanL Lo map color-coded symbols aL Lhe sLauon locauon denoung
u1, uS aL a selecuon of pressures and denslLy LhaL span Lhe upper 1000 m of Lhe waLer column.
lor selecL reglons you can ploL 1/p and S/ proles and 1/S from boLh crulses.
Cpuonal: compare Lhe observauonal based seasonal pauern Lo LhaL slmulaLed ln model ouLpuL.
uaLa SeL: Arllndo_93_94_C1u. 1xL ls a LexL le of Lhe 1993 and 1994 sLauon daLa: LemperaLure,
sallnlLy, oxygen wlLh pressure (depLh), plus sLauons number and laL/long. 1he 1993 sLauons are
ln Lhe 3000 llsL, 1994 ln Lhe 4000 llsL (aer Lhe 3 or 4 ls Lhe sLauon number).
?5 @A<"33/0B $3*0B 6*C,' D$E$ .*$#,

SClLnCL C8!LC1lvL: SLudylng ocean processes uslng model and observauon. undersLand varlablllLy of Lhe
upwelllng and udal mlxlng lmpacL on Lhe waLer masses.
8LSLA8CP 8C!LC1:
1 Make secuon ploL of sea surface LemperaLure for Lhe Lwo maln season of upwelllng uslng Lhe model
slmulauon and observauon
2 Make verucal secuon along souLh [ava coasL for Lhe Lwo maln season of upwelllng uslng Lhe model slmulauon
and observauon
3 ln an average box souLh of !ava, make a depLh vs ume ploL (hovmuller dlagram)
4 Make horlzonLal gradlenL ploL aL Lhe surface souLh of !ava for Lhe Lwo seasons
uA1A SL1S :
- 1wo annual mean slmulauons, forced by CllmaLologlcal forclng, one wlLh udal parameLerlsauon
(1luLS_CLlM_ann_run.nc), one wlLhouL udal parameLerlsauon (8Ll_CLlM_ann_run.nc), wlLh LemperaLure and
sallnlLy varlables.
- Cbservauon : annual mean and monLhly mean 1emperaLure and sallnlLy from CA8S2009 daLa base
(salL_ann.cdf, Lemp_ann.cdf, salL_monLh.cdf, Lemp_monLh.cdf)
- Cne monLhly slmulauon, forced by lnLerannual forclng, wlLh 1ldal parameLerlsauon
(1luLS_ln1L8ann_monLhly_run.nc)
nC1L: lease puL legend for Lhe gures (you mlghL need several gures Lo answer one quesuon) and a shorL
analysls of your undersLandlng of whaL you see on Lhe ploLs.
F5 9/>$3 G/H/0B /0 ,'" =0>*0"#/$0 6"$#
SClLnCL C8!LC1lvL: SLudylng ocean processes uslng model and observauon. undersLand varlablllLy of Lhe
upwelllng and udal mlxlng lmpacL on Lhe waLer masses.
8LSLA8CP 8C!LC1:
1 Make secuon ploL of Lhe LemperaLure and sallnlLy aL 100 m, 400m for Lhe model slmulauons and observauon.
lease make nlce and same scale for colorbar. lease glve explanauon abouL Lhe dlerences beLween 2
slmulauons and comparlng Lhe 3 locauons
2 Make secuon ploL of Lhe LemperaLure anomaly beLween Lhe slmulauon wlLh and wlLhouL udal
parameLerlzauon aL 100 m, 400m. lease make nlce symmeLrlc Lo zero scale for colorbar. WhaL can you see ?
lease glve explanauon for Lhe slgn of anomaly
3 CreaLe 3 1S dlagram for each followlng reglons: Palmahera, 8anda sea and Makassar sLralL, uslng Lhe model
slmulauons and observauon on Lhe same ploL, wlLh dlerenL llne color for each. WhaL can you see ? lease glve
explanauon abouL Lhe dlerences beLween 2 slmulauons and comparlng Lhe 3 locauons
uA1A :
- 1wo annual mean slmulauons, forced by CllmaLologlcal forclng, one wlLh udal parameLerlsauon
(1luLS_CLlM_ann_run.nc), one wlLhouL udal parameLerlsauon (8Ll_CLlM_ann_run.nc), wlLh LemperaLure and
sallnlLy varlables.
- Cbservauon : annual mean and monLhly mean 1emperaLure and sallnlLy from CA8S2009 daLa base
(salL_ann.cdf, Lemp_ann.cdf, salL_monLh.cdf, Lemp_monLh.cdf)
- Cne monLhly slmulauon, forced by lnLerannual forclng, wlLh 1ldal parameLerlsauon
(1luLS_ln1L8ann_monLhly_run.nc)
nC1L: lease puL legend for Lhe gures (you mlghL need several gures Lo answer one quesuon) and a shorL
analysls of your undersLandlng of whaL you see on Lhe ploLs.
I5 J/K"%"0&"# /0 L00C$3 G"$0 M"3*&/,7

Sclence Cb[ecuve: 1o examlne spaual pauerns of Lhe annual mean ln Lhe l1l reglon
8esearch ro[ecL: CompuLe annual averaged ow eld (for 2011 or 2012, or over
boLh), lnspecL Lhe annual mean pauern of Lhe ow eld Lo ldenufy Lhe lnow and
ouow passages of Lhe l1l represenLed by P?CCM (esp. Lhe Makassar SLralL, Lombok
SLralL, Cmbal SLralL, and 1lmor passages). CompuLe Lhe mean and sLandard devlauon
of Lhe ow Lhrough each ma[or channel and dlscuss dlerences.
uA1A:
- Surface and 100 m veloclLy ln 2011 and 2012
P?CCM_2011uv_0m.nc (dally u and v for 2011) P?CCM_2012yv_0m.nc (dally u and v
for 2012) P?CCn_monLhly_uv_0m.nc (monLhly u and v for 2011 and 2012)
P?CCM_2011uv_10m.nc (dally u and v for 2011) P?CCM_2012yv_10m.nc (dally u
and v for 2012) P?CCn_monLhly_uv_10m.nc (monLhly u and v for 2011 and 2012)

nC1L: 1he veloclues are ln m/s.
1hey have been averaged from 1/12- Lo 1/4-degree resoluuon.
A MaLlab scrlpL (m-le) ls avallable Lo read Lhe eld and make an example ploL.
N =0,"%$00C$3 E$%/$1*0 /0 M"3*&/,7
Sclence Cb[ecuve: 1o examlne Lhe spaual pauerns of Lhe year-Lo-year varlauon ln Lhe
l1l reglon
8esearch ro[ecL: CompuLe Lhe mean and sLandard devlauon of Lhe model veloclLy
elds. 1ake Lhe dlerence beLween 2011 and 2012 annual mean Lo examlne
lnLerannual dlerences ln Lhe ow eld. Compare Lo Lhe mean, sLandard devlauon
and year-Lo-year dlerences of Lhe observed Makassar SL. moored veloclLy elds.
WhaL mlghL cause Lhe dlerences beLween Lhe model and daLa ow elds?
uA1A:
- P?CCM Surface and 100 m veloclLy ln 2011 and 2012
P?CCM_2011uv_0m.nc (dally u and v for 2011) P?CCM_2012yv_0m.nc (dally u and v
for 2012) P?CCn_monLhly_uv_0m.nc (monLhly u and v for 2011 and 2012)
P?CCM_2011uv_10m.nc (dally u and v for 2011) P?CCM_2012yv_10m.nc (dally u
and v for 2012) P?CCn_monLhly_uv_10m.nc (monLhly u and v for 2011 and 2012)
Makassar SLralL moorlng veloclLy aL Lhe surface and 100 m

nC1L: 1he veloclues are ln m/s.
1hey have been averaged from 1/12- Lo 1/4-degree resoluuon.
A MaLlab scrlpL (m-le) ls avallable Lo read Lhe eld and make an example ploL.
O5 !"B/*0$3 J/K"%"0&"# /0 6"$#*0$3 .7&3" *8 M"3*&/,7

Sclence Cb[ecuve: 1o examlne spaual pauerns of Lhe annual mean ln Lhe l1l reglon
8esearch ro[ecL: CompuLe seasonal anomalles of Lhe veloclLy eld by removlng Lhe
annual mean, compare Lhe veloclLy anomalles ln dlerenL seasons (e.g., Lhe souLheasL
monsoon (ASC) vs Lhe norLhwesL monsoon (u!l) vs Lhe monsoon Lransluons (AM,
Cn)) and ploL for Lhe lndoneslan reglon. Can you ldenufy Lhe anomalous ow from Lhe
SouLh Chlna Sea blocklng Lhe upper parL of Lhe l1l ln a cerLaln season? Pow does Lhe
seasonal cycle change for dlerenL reglons of Lhe lndoneslan seas? WhaL mlghL cause
Lhls varlablllLy? ls Lhe seasonal cycle Lhe same aL Lhe surface and 100 m depLh?
uA1A:
- Surface veloclLy ln 2011 and 2012
P?CCM_2011uv_0m.nc (dally u and v for 2011) P?CCM_2012yv_0m.nc (dally u and v
for 2012) P?CCn_monLhly_uv_0m.nc (monLhly u and v for 2011 and 2012)
- 100-m veloclLy for 2011 and 2012
P?CCM_2011uv_10m.nc (dally u and v for 2011) P?CCM_2012yv_10m.nc (dally u
and v for 2012) P?CCn_monLhly_uv_10m.nc (monLhly u and v for 2011 and 2012)

nC1L: 1he veloclues are ln m/s. A MaLlab scrlpL (m-le) ls avallable Lo read Lhe eld
and make an example ploL5
P5 Q$,"% R$## E$%/$2/3/,7 *8 ,'" =9S 8%*R ,'" =0T*< ,* UCV3*<

SClLnCL C8!LC1lvL: Pow does Lhe aclc lnow waLer dler from Lhe lndoneslan
1hroughow waLer LhaL ls found ln Lhe lndlan Ccean
8LSLA8CP 8C!LC1:
Lxamlne Lhe mean and varlauon of Lhe verucal 1-S properues ln Lhe lnow aclc
reglon and Lhe ouulow lndlan Ccean. Pow do Lhey change wlLh depLh? Pow do Lhey
change on denslLy surfaces? Compare Lo Lhe mean cllmaLology. WhaL mlghL be
responslble for Lhe dlerences beLween Lhe synopuc 1-S and Lhe cllmaLologlcal 1-S
properues? Look aL Lhe depLh versus ume varlablllLy ln an average box for Lhe aclc
lnow reglon and Lhe lndlan Ccean ouulow reglon. WhaL changes are evldenL over
ume? WhaL mlghL cause Lhese changes?
uA1A SL1S :
- Argo daLa from lnow and ouulow reglons (ln subdlrecLory argo_download)
- Cbservauon : annual mean and monLhly mean 1emperaLure and sallnlLy from
CA8S2009 daLa base (salL_ann.cdf, Lemp_ann.cdf, salL_monLh.cdf,
Lemp_monLh.cdf)
nC1LS: 1hls pracucal mlghL besL be done ln Cuv as easler Lo load A8CC daLa (ask
!aneL)
8. Coherence of =0>/$0 U&"$0 )%*A"%1"#

SClLnCL C8!LC1lvL: Lxamlne Lhe characLerlsucs of waLer mass properues and ow
characLerlsucs wlLhln Lhe lndlan Ccean.

8LSLA8CP ro[ecL: loL some of Lhe raw daLa ume serles and examlne Lhe mean,
sLandard devlauon and correlauon of Lhe properues and ow from Lhe lndlan Ccean
moorlng daLa aL dlerenL depLhs on Lhe same moorlng and aL slmllar depLhs beLween
dlerenL moorlngs. WhaL are Lhe slmllarlues and dlerence? WhaL mlghL cause Lhese?
Lxamlne Lhe specLra for a few properues. WhaL are Lhe maln frequencles? Lxamlne
Lhe lnLegral ume scale (hlnL: compuLe Lhe auLocovarlance, nd Lhe decorelauon scale
(e-foldlng)) Pow does Lhls dler beLween properues aL Lhe surface and aL depLh? Pow
does Lhls dler beLween moorlngs?

uA1A:
SS1 and veloclLy daLa ln subdlrecLory lndlan8AMAMoorlnguaLa
noLe: In folder Practical_Correlations, see the tutorial outlined in
Lecture2Correlations.ppt and adapt Lecture2b_CorrelationsTAO0147.m
For your available mooring data.

95 )%*A"%1"# *8 ,'" 6C%8$&" =9S </,'/0 ,'" =0>/$0 U&"$0

SClLnCL C8!LC1lvL: Lxamlne Lhe characLerlsucs of l1l properues wlLhln Lhe
lndlan reglon.

8LSLA8CP ro[ecL: Lxamlne Lhe mean, sLandard devlauon and correlauon of
Lhe properues of Lhe l1l SS1 and SSS from Lhe lndlan Ccean moorlng daLa.
Lxamlne a 1-S ploL aL each locauon. WhaL are Lhe slmllarlues and dlerence
aL each moorlng slLe? WhaL mlghL cause Lhese? Lxamlne a PCvMuLLL8
dlagram (LonglLude vs ume) gure. WhaL changes are evldenL wlLh ume?
WhaL mlghL cause Lhls varlablllLy? Lxamlne Lhe specLra of SS1 and SSS aL each
moorlng slLe. WhaL are Lhe maln frequencles?

uA1A:
SS1 (ALL8AMA_ssL_xyL_mon.cdf.gz) and SSS (ALL8AMA_sss_xyL_mon.cdf.gz)
daLa ln subdlrecLory lndlan8AMAMoorlnguaLa
10. .*'"%"0&" *8 )%*A"%1"# </,'/0 ,'" )$&/;& U&"$0

SClLnCL C8!LC1lvL: Lxamlne Lhe characLerlsucs of waLer mass properues and ow
characLerlsucs wlLhln Lhe lndoneslan reglon.

8LSLA8CP ro[ecL: loL some of Lhe raw ume serles daLa and examlne Lhe mean,
sLandard devlauon and correlauon of Lhe properues and ow from Lhe aclc Ccean
moorlng daLa aL dlerenL depLhs on Lhe same moorlng and aL slmllar depLhs beLween
dlerenL moorlngs. WhaL are Lhe slmllarlues and dlerence? WhaL mlghL cause Lhese?
Lxamlne Lhe specLra for a few properues. WhaL are Lhe maln frequencles? Lxamlne
Lhe lnLegral ume scale (hlnL: compuLe Lhe auLocovarlance, nd Lhe decorelauon scale
(e-foldlng)) Pow does Lhls dler beLween properues aL Lhe surface and aL depLh? Pow
does Lhls dler beLween moorlngs?

uA1A:
1emperaLure and veloclLy daLa ln subdlrecLory aclc1ACMoorlnguaLa
In folder Practical_Correlations, see the tutorial outlined in
Lecture2Correlations.ppt and adapt Lecture2b_CorrelationsTAO0147.m
for the other mooring data that are available


445 WH$R/0/0B .*'"%"0, 9/R" 6&$3"# /0 ,'" =0>*0"#/$0 6"$#
SClLnCL C8!LC1lvL: Lxamlne Lhe characLerlsucs of waLer mass properues and ow
characLerlsucs wlLhln Lhe lndoneslan reglon.

8LSLA8CP ro[ecL:
1. loL Lhe raw ume serles daLa and compuLe Lhe correlauon (r) for posslble palrs of
Lhe LemperaLure and pressure daLa from Lhe lnS1An1 SCA ln Cmbal and 1lmor
SLralLs (Lhe maLlab funcuon corrcoef ls an easy way Lo deLermlne r, buL check
Lhe help page Lo make sure you undersLand how Lo lnLerpreL lLs ouLpuL).
2. Assumlng LhaL each dally observauon ls sLausucally lndependenL, are Lhese
correlauons sLausucally slgnlcanL? Whlch palrs are mosL sLrongly correlaLed?
uoes Lhls seem surprlslng?
3. CompuLe and ploL Lhe auLocovarlance for all ume serles (use xcov). Where
does Lhe rsL zero-crosslng occur? lf you use Lhe rsL zero-crosslng as a
decorrelauon scale and ad[usL Lhe number of eecuve degrees of freedom n
e
,
how does Lhls change Lhe slgnlcance level r
slg
compared Lo your resulLs ln
quesuon 1? CompuLe alLernaLe decorrelauon scales by ndlng Lhe ume scale for
Lhe ACl Lo drop from 1 Lo 0.3 or exp(-1). Pow does Lhls change Lhe resulLs?
4. SpeculaLe a llule: WhaL seems llke a relevanL ume scale Lo you? WhaL mlghL
cause Lhe slgnlcanL correlauons beLween some sLralLs and noL oLhers?

uA1A:
ln folder racucal_Correlauons, see Lhe LuLorlal ouLllned ln LecLure2Correlauons.ppL.
lollow Lhe maLlab scrlpL ln LecLure2a_CorrelauonsSCA.m for daLa seL found ln
uallylnS1An1SCA

4?5 !"3$1*0#'/A 2",<""0 )'7#/&$3 $0> X/*3*B/&$3 )%*A"%1"#
SClLnCL C8!LC1lvL: Lxamlne Lhe relauonshlp beLween physlcal and blologlcal properues ln Lhe
lndoneslan reglon
8LSLA8CP ro[ecL:
1. loL Lhe raw daLa and compuLe Lhe correlauon (r) for mlxed layer depLh and LoLal
chlorophyll for Lhe wesLern aclc daLa from Lhe nASA Ccean 8logeochemlcal Model
(nC8M) (hlnL: use Lhe maLlab funcuon corrcoef").
2. Assumlng LhaL each observauon ls sLausucally lndependenL, are Lhese correlauons
sLausucally slgnlcanL? uoes Lhls seem surprlslng? Why or why noL?
3. CompuLe and ploL Lhe auLocovarlance for all ume serles (use xcov). Where does Lhe rsL
zero-crosslng occur? lf you use Lhe rsL zero-crosslng as a decorrelauon scale and ad[usL Lhe
number of eecuve degrees of freedom n
e
, how does Lhls change Lhe slgnlcance level r
slg

compared Lo your resulLs ln quesuon 1? CompuLe alLernaLe decorrelauon scales by ndlng
Lhe ume scale for Lhe ACl Lo drop from 1 Lo 0.3 or exp(-1). Pow does Lhls change Lhe
resulLs?
4. SpeculaLe a llule: WhaL seems llke a relevanL ume scale Lo you? Why?
3. CompuLe Lhe specLra of each ume serles. Pow are Lhey slmllar/dlerenL?
uA1A:
ln folder racucal_Correlauons, see Lhe LuLorlal ouLllned ln LecLure2Correlauons.ppL.
lollow Lhe maLlab scrlpL ln LecLure2c_CorrelauonsnC8M.m for daLa found ln
nC8M_CPLMLu_ac/nC8M_1chlMLu.maL


4F5 9%"0># /0 )'7#/&$3 $0> X/*3*B/&$3 )%*A"%1"#
SClLnCL C8!LC1lvL: Lxamlne Lhe Lrends ln physlcal and blologlcal properues ln Lhe lndoneslan
reglon
8LSLA8CP ro[ecL:
1. loL Lhe raw ume serles and a scauer ploL of Lhe mlxed layer depLh (y-axls) and LoLal
chlorophyll (x-axls) for Lhe wesLern aclc daLa from Lhe nASA Ccean 8logeochemlcal
Model (nC8M)
2. llL a llnear funcuon (leasL squares L) Lo Lhe relauonshlp.
3. CompuLe a llnear Lrend for Lhe MLu and chlorophyll ume serles. Are Lhe Lrends slgnlcanL?
Conslder Lhe slgnlcance lf ou calculaLe Lhe lnLegral ume scale
4. SpeculaLe a llule: WhaL mlghL cause Lhls Lrend?
3. llL a seasonal cycle Lo each ume serles uslng harmonlcs. Pow does Lhe ampllLude and phase
of Lhe seasonal cycle dler for Lhe MLu and chlorophyll? Why?
6. 8emove Lhe seasonal cycle from Lhe raw ume serles and recompuLe Lhe Lrend. Pow ls Lhls
dlerenL (or noL) from Lhe Lrend compuLed uslng [usL Lhe raw daLa?
uA1A:
ln folder 1rends see Lhe LuLorlal ouLllned ln LecLure3_llmngluncuons.ppLx for lnformauon on
how Lo underLake Lhls analysls.
lollow Lhe maLlab scrlpL ln LecLure3a_AnnualCyclenC8M.m and LecLure3b_Llnear1rendnC8M.m
1he daLa ls found ln racucals_Correlauons/nC8M_CPLMLu_ac/nC8M_1chlMLu.maL


14. ?our Cholce!

Choose a Loplc of lnLeresL Lo lnvesugaLe ln Lhe
lndoneslan seas uslng Lhe Lechnlques LhaL you
have learnL so far, and Lhe daLa seLs avallable
Lo you ln Lhe compuLer lab.
8efore you begln, formulaLe your sclence
ob[ecuve and approach, and Lhen dlscuss your
Loplc wlLh a senlor lecLurer for some guldance
as Lo how you mlghL underLake your sLudy.
8esearch ro[ecLs: Slgn-up SheeL
!"#"$%&'
)%*+"&,
YCR2"%
Z%*CA Y$R" Z%*CA )$%1&/A$0,# [0$R" $0> $\3/$1*0]