Korea Ocean Satellite CenterKorea Ocean Research & Development Institute

GOCI Data Processing System : Algorithm and Cal/Val

Joo-Hyung Ryu with KOSC Colleaques

Research/Application

GDPS SW+Cal/Val

GOCI data

Maximize the GOCI application

Next GOCI

GOCI coverage

China(? Billion)

Korea(70 million)

Japan(100 million)

Taiwan

Russia

→ GOCI coverage / Earth surface = 6,250,000 / 520,000,000 km2 = 1.2 %

(Considering the geographical size, coverage area is very small) → East Asia : South and North Korea, China, Japan, Russia, Taiwan Population : large (a heavily populated district) Economy, Commerce : big (container traffic) Defense, Military : important Environments : variable, changeable Climate change : serious

Water Leaving Radi-ance(Lw)

Normalized Water Leav-ing

Radiance(nLw)

Optical Properties Water

Chlorophyll

TSS

CDOM

Redtide

Fishing Ground Informa-tion

Under water Visibility

Water Current Vector

Atmosphere & Earth En-vironment Monitoring

Water Quality Level 1-5

Primary Productivity

Earth Environ-ment Climatic

Change

Long & Short term Ocean Envi-ronment Monitor-

ing

Atmosphere & Weather

Land & Disaster Preven-

tion

Military Applica-tion

Secret Intelli-gence Commu-

nity

Ocean Environ-mental

Monitoring

DisasterMonitoring

GOCI Product

Applications Objects

+Prediction &

ForecastCooperation with Others

Missions of GOCI

GDPS introductionGOCI Data Processing System

Ocean Color SW

• SeaDAS : SeaWiFS, MODIS • BEAM : MERIS• GDPS : GOCI

– A standard operational system of KOSC– Basic data processing system for GOCI user– Well-designed data processing structure – World’s first Geostationary Ocean Color Data

Processing SW

★ GOCI : Geostationary Ocean Color Imager★ COMS : Communication Ocean and Meteorological Satellite

★ It shall be operated in a staring-frame capture mode onboard its COMS.

★ The mission concept includes eight visible-to-near-infrared bands, 500 m spatial resolution, and a coverage region of 2,500*2,500 km centered at Korea.

★ The instrument is expected to provide SeaWiFS quality observa-tions for a single study area with imager frequency of 1 hour from 9 am to 4 pm (8 times a day).

★ GOCI Application : LEO mission + Operational mission harmful algae bloom (HAB), health of marine ecosystem, move-

ment of suspended sediment and current, and to produce marine fisheries information for fishing communities + ocean forecast-ing (with modeling)

A brief overview of GOCI

GEO vs. LEOGEO/GOCI LEO/SeaWiFS

Altitude 35,857 km 705 km about 50 times-far

Sensor typeStaring-frame

capture 1-axis scanning

Spatial resolu-tion 500 m 1000 m 4 times better

Spectralrange 400-900 nm 400-900 nm Almost same

Temporal resolution 1 hour 1 day 8 times better

Sun-Satellite position variable stable BRDF

SNR > 1,000 ~ 300 3 times better

MTF > 0.33 ~ 0.3 Better Image Quality

Coverage local global limitation

Bio-opticalalgorithm local global New local algo-

rithmGEO is about 50 times farther from the Earth than LEOGEO spatial resolution is 4 times better than that of LEOGEO temporal resolution is 8 times betterTo be considered and prepared sensor type, geometry & local coverage for overcoming GEO characteristics

Korean Seas : Case-I & II water► In the GOCI swath (==CoastColour site 11)

― The coastal area of Korea & China and the ECS(seasonal variation) is a typical Case-2 waters

― East/Japan Sea, central YS & Pacific come under the Case-1 water

Definition of Case-1 and 2 waters using in situ <chl> and <SS> data of the Korean Sea

In situ measurement were performed during a lot of cruises in the Korean terrestorial seas and neighboring waters through the years 1998-2010 onboard the KORDI research vessel and fisher boats.

120 oE 125 oE 130 oE 135 oE 140 oE

25 oN

30 oN

35 oN

40 oN

45 oN

50 oN

RUSSIA

KOREA JAPAN

• Total points: 1348 points

• IOP and AOP

Issue 1 : Geometry variation and Local cov-erage

Ferrybox

Year Month Vessel

2001 5,7,8,9,10 Chunghaejin

2002 6,7,8,9,10 Chunghaejin

2003 5,7,9,10,11 Chunghaejin

2004 8,9,11 Ohamana

18 times during 4 years Operation time : 19:00-10:00(next

day) Instruments : YSI(SS, chl, T, S,

DO), ChelSea(chl), SeaPoint(chl), McVan(SS)

Water sample : about 40 (every 20 min after departure) <SS>, <chl>, adom

Incheon

Jeju

For understanding the ocean environmental parameter of the YS, we conducted the ferrybox project during 4 years.

Concentration of SS, <chl>, aDOM in the YS

SS : 0-186 g/m3

<chl> : 0 – 16 mg/m3

adom : 0-0.9 m-1

GDPS algorithm

IMPS

GeometricalCorrection &

Mosaic

LEVEL-0Sector Image

LEVEL-1ADigital values

LEVEL-1BTotal

Radiance

Waterleaving

Radiance

LwN(B1-B8)

Bi-DirectionalCorrection

AtmosphericCorrection

RadiometricCalibration

LookupTable Reference

Target SPData

ImageDisplay

ASCII-DataGeneration

Red-TideFishing Ground

InformationCDOMTSSUnder water

VisibilityWater Current

VectorAtm. & EarthEnvironment

CASE-I&II & Fluor. Algorithm

1-BandAlgorithm

AbsorptionCoefficient

FishingGroundIndex

EmpiricalAlgorithm

Image patternComparison

3-BandsAlgorithm

Red-tideIndex

Chlorophyll

Water QualityLevel 1-5

Yellow DustForest FireInland flood

Vegetation IndexHeavy snowfall

Optical Properties

Water

K & InherentOptical

algorithm

Land Mask

CloudMask

Turbid waterMask

K-coefficientAbsorption

coeff.Backscattering

Coeff

New & oldmodel

NOAAImage

KOSC

PrimaryProductivity

GDPS

New model

Accessory Data file Sub-routine

Low levelData file

Analyzed data file Option

GOCI Data Processing System flow

CalibrationCoefficients

GOCI L2 Product

GDPS GUI

GOCI L2 Display

Setting Process Schedule

Band-math

(operational) GDPS Requirement

• Algorithm Programming– Developed atmosphere/bio-optical algorithms are integrated in

the software and working in real time.• Timeliness

– less than 30 minutes from L1B to L2 generation • Automation

– All data processing function can work automatically– Pre-configured operating support

• OS– PC Windows

• Output Format– HDF-EOS 5 format

• Development Tool– Microsoft Visual C++

GDPS - R1A

Duplicationl

GDPS – R2

GDPS - R1B

GDPS - UI

<GDPS System>

Processing Server SPEC:3.0GHz4core 2cpu, 4GB ram

Product (Level)

• Level 1B– Radiometric & geometric corrected Total Radiance

• Level 2– Environmental properties derived from Ocean

signal(Lw)– For GOCI, L2 data will be generated each hour.(8

times/day)• Level 3

– Secondary derived data from L2 like Fishery Ground in-formation, Primary Production.

– Cloud-free(reduced) ocean environmental data by daily composite of L2

GDPS result

Level1b L2 reflectance RGB-642 com-posite image (background: L1b band6)

Suspended Sediment Concen-tration (background: L1b band6)

Band 1 Band 2 Band 3 Band 4

Band 5 Band 6 Band 7 Band 8

GDPS productsLw

CHL SS CDOM

Kd

GDPS products

NDVIAbsorption Coeff.

GDPS Cal/Val plan

in situ measurements Research vessel, Ferrybox(with KORDI), Glider(with KO-

RDI) Buoy, Ocean research station :

• To use Korea Operational Oceanography Network(with KORDI)

• To cooperate neighboring countries (with Japan, China, Taiwan)

• To join International Group (with IOCCG, OCR-VS, Aeronet-OC, ESA CoastColor)

Inter-satellite Cal. Existing OC : MODIS, MERIS HICO (with D. Curtiss)

New system Kite, aerostat, airborne (with KARI) Argo-type buoy

Uniform land Cal/Val site Desert, Ice, Playa

Raw data

Level-0

Radiance(Lw, LwN)

Products(Rrs, chl, SS, DOM, Red-tide, Fishing

ground information, Under water visibility, water current vector,

and so on.)

Level-2

Level-1Radiometric / Geometric Calibration

Atmospheric correction CAL/VAL

Algorithm CAL/VAL

GOCI data processing and CAL/VAL activities GOCI data processing and CAL/VAL activities

GOCI Cal/Val step

Mr. Cho

Mr. Ahn

Dr. Park

Lessons from MERIS and MODIS Cal/Val

• From temporal discontinuity measurement (by ship)

• To continuous measurement (by buoy and tower) : Aeronet-OC

0.1

1

10

100

0.1 1 10 100

Seaborne CHL(mg/m3)

ME

RIS

CH

L(m

g/m

3)

Ship measurement : 10pts/2.5years

Tower measure-ment: 200pts/

6.5years

Guideline of KOSC Cal/Val

• To operate optic lab for radiometer(instrument)

validation : under construction

• To establish a serise of Cal/Val system

• To apply previous KORDI fixed system including

buoy, tower, ferry

• To cooperate international Cal/Val group and

neighboring country

Ieodo Station(2003)

Gageocho Station (2009)

Yellow Sea Buoy(2007)

Dokdo Buoy (2009)

Korea Operational Oceanography Network

East(Japan) Sea Station (2012)

Two buoys and two ocean stations have already been constructed by KOON project

Korea Operational Oceanography Net-work

Ieodo Station(2003)

• Case-1/2• Depth : 41m• 33 instruments

• Satlantic HyperSAS• 2003-2004• Managed by Govern-

ment Office

Gageocho Station(2009)

• Case-1/2• Depth : 15m• 31 instruments

• Aeronet-OC(2011)• TriOS(2010) : broken• Fluorometry(2010)• Turbidimeter(2010)

Dokdo station(2012)

• Case-1• 2012 (plan)

• Aeronet-OC(2013)• Fluorometry(2013)• Turbidimeter(2013)

Yellow Sea buoy(2007)

• Case-1/2• Depth : 80m• Diameter : 10m• Weight : 50ton

• Fluorometry• Turbidimeter

Dokdo buoy(2009)

• Case-1• Depth : 300m

• Fluorometry : lost

•

These systems are not for remote sensing only

Ieodo Ocean Research Station► Meteorological Instruments (13)► Environment Observing Systems

(6)► Ocean Monitoring Systems (22)

– Spectroradiometer – Wave radar– Directional Waverider– Self Contained Ultrasonic Sensors– Sea Level Monitor– Acoustic Doppler Velocimeter– ADCP– CTD– CTR7– Ultraviolet Fluorometer– Chlorophyll Fluorometer– Current Profiler– Ultrasonic Level Meter– etc.

Ieodo station is managed by NORI (Government offfice)KORDI : design, construction, installation, test NORI : operational work

Satlantic HyperSAS OCR-3000

Lsky(λ)

Ed(λ)

LwT(λ)

Comparison between HyperSAS and SeaWiFS water-leaving radiance in each

band

All visible bands were well correlated with SeaWiFS excepted at longer wave-

lengths

30

AERONET-OC (2002-present)

Current management and responsibilities

Potential sitesPlanned sites

Current sites

• NASA manages the network infrastructure (i.e., handles the instruments cali-bration and, data collection, processing and distribution within AERONET).

• EU JRC has the scientific responsibility of the processing algorithms and per-forms the quality assurance of data products.

• PIs are responsible for establishing and maintaining AERONET-OC sites.

26m

Gageocho Ocean Research Station

► Meteorological Instruments (12)► Environment Observing Systems (7)► Ocean Monitoring Systems (12)

► TriOS (installed on Jul. 6, 2010): taken as 1-min acquisition sequences every 15min from 8 am to 5 pm► Aeronet-OC (was tested on June 2011)

EdLsky

LwT

In-Situ TriOS Rrs(λ) : Gageocho Platform

Aug. 20 2010 - TriOS Rrs(λ)- GOCI Rrs(λ)

- TriOS Rrs(λ)- GOCI Rrs(λ)

- TriOS Rrs(λ)- GOCI Rrs(λ)- MODIS Rrs(λ)

10:35 11:35, cloudy

12:35 13:35

- TriOS Rrs(λ)- GOCI Rrs(λ)

- TriOS Rrs(λ)- GOCI Rrs(λ)

14:35

1. Cross Comparisons with time-series : in-situ TriOS, GOCI, MODIS Rrs(λ)

2. Rrs of shorter wavebands is still problem in early morning and late af-

ternoon

- Optical Path is different with time

- During In-Orbit Test (IOT), this problem is our homework

Comparison between In-Situ TriOS Rrs(λ) and GOCI Rrs(λ) : Gageocho Platform

Rrs(412) Rrs(443) Rrs(490)

Rrs(555) Rrs(660) Rrs(680)

Rrs(745) Rrs(865)

Statistics GOCI and in-situ matching data(Gageocho station)

Parameters Number of points RMSE AR RPD APD

Rrs(412) 8 0.212 1.61 60.81 60.81

Rrs(443) 8 0.189 1.51 50.57 50.57

Rrs(490) 8 0.090 1.08 8.20 17.59

Rrs(555) 8 0.138 1.23 23.02 27.48

Rrs(660) 8 0.425 2.66 166.09 166.09

Rrs(680) 8 0.401 2.51 150.92 150.92

Rrs(745) 8 0.731 5.53 452.59 452.59

Rrs(865) 8 0.735 5.45 445.02 445.02

N

YXRMSE

Ni

iii

1

21010 loglog

Ni

i i

i

Y

X

NAR

1

1

Ni

i i

ii

Y

YX

NRPD

1

1100

Ni

i i

ii

Y

YX

NAPD

1

1100

Xi : ith satellite-derived valueYi : ith in-situ-derived valueN : number of pointsRMSE : root mean square errorAR : average ratio of satellite to in-situ dataRPD : average relative percent difference (%)APD : average absolute percent difference (%)

Statistics of GOCI and in-situ matching data(Gageocho station)

Rrs(l)

All bandsRMSE=0.436

Rrs(l)

All bandsRMSE=0.436

Comparison of Chl. algorithms

Equation Reference RMS

OC2v2O’Reilly et al.

(1998)0.382

OC2v4O’Reilly et al.

(1998)0.399

OC4v4 (SeaWiFS standard)

O’Reilly et al. (1998)

0.413

OC3(MODIS

standard)

O’Reilly et al. (2000)

0.387

MERISpigment CASE-I

MERIS ATBD(2007)

0.450

GOCI 0.328

0929.01032 0077.08358.02429.22974.0 RRR

555

490log10

rs

rs

R

RR

071.01032 135.0879.0336.2319.0 RRR

555

490log10

rs

rs

R

RR

432 532.1649.0930.1067.3366.010 RRRR

432 403.1659.0457.1753.2283.010 RRRR

555

510,490,443log10

rs

rsrsrs

R

RRRMaxR

551

488,443log10

rs

rsrs

R

RRMaxR

432 75312.148061.544357.56303.340657.010 RRRR

560

510,490,443log10

rs

rsrsrs

R

RRRMaxR

1473.46497.1 R

555

412490443

rs

rsrsrs

R

RRRR

<chl> comparison of each algo-rithmGOCI OC4v4 OC2v2

GOCI OC4v4 OC2v2 in situ

Aver-age

1.065 2.483 1.964

0.96 (Feb. 2003) : same area

0.69 (Mar. 2004) : ECS0.70 (Mar. 2005) : ECS

vs. OC4v4

57.1 %

vs. OC2v2

45.8 %

Selection area: Lat. 33 – 34, Long. 124 – 125 (Red Box)Unit : mg/m3

- Local algorithms are well matched with in situ measurements- <chl> of local algorithm decreases about 50 % than that of NASA standard alg.

North

Google Earth

HICO Data

Inter-satellite Cal : GOCI vs. HICO (by D. Curtiss)

HICO ImagePusan, South Korea: Nov.18, 2009

GOCI(2010 09 24 01: 16:43) (R(680), G(555), B(412))

MERIS(2010 09 24 01: 43:51) (R(681.55), G(560), B(412.5))

*GOCI image is geo-corrected by MERIS Geometric information

Inter-satellite Cal : GOCI vs. MERIS

Ferrybox (Containership box)

Hanjin

Co.

2,821-

ton

4,538-

ton

Length ;

93.3m

Width ; 15.6m

Draft ; 5.60m

Speed ; 11

knot

Incheon→Kwangyang→

Pusan→Incheon

Ice-breaking Ship (Araon)

GOCI Cal/Val Discussion

• After slot and atmospheric correction…• No international OC Cal/Val site in GOCI coverage

• Indirect method : Inter-satellite Cal using MERIS,MODIS, NPP, HICO

• Existing Ocean Research stations need to be stabilize for GOCI Cal/Val • Our efforts (with KORDI) • Supporting of International groups (instrument Cal/Val)

• AERONET-OC site selection : Gageocho Station (with TriOS)• Already registered to NASA • After typoon season …• “Once started, termination is irreversible”

• No land Cal/Val site in GOCI coverage• No desert, ice, playa : uniform, wide, long term-radiometric

stability • Tidal flat (if possible) : uniform(except SMC), wide

Recent application of GOCI

- SSC hourly variation monitoring - Green algae bloom in ECS and YS- Oil spill in Bohai Sea

Hourly and daily variations in SSC during 3 days

Southwest ocean of Korea : 16-18 Sep. 2010

2010. 09. 16 01:16(UTC)2010. 09. 16 02:16(UTC)2010. 09. 16 03:16(UTC)

2010. 09. 18 06:16(UTC)

2010. 09. 16 04:16(UTC)2010. 09. 16 05:16(UTC)2010. 09. 16 06:16(UTC)2010. 09. 17 00:16(UTC)2010. 09. 17 01:16(UTC)2010. 09. 17 02:16(UTC)2010. 09. 18 03:16(UTC)2010. 09. 18 04:16(UTC)2010. 09. 18 05:16(UTC)2010. 09. 18 06:16(UTC)

Applied algorithm

BRDF correction : GOCI algorithm & relative algorithm Atmospheric correction : coastal standard correction & SSMMEmpirical SS algorithm developed based upon the in-situ data sets obtained on the Yellow Sea 2005 – 2010

Hourly SS variations10:16 11:16 12:16

13:16 14:16

0 10SS (g/m3)

There is no notable difference over the study area from 10:30 to 12:30. However, white box area is gradually decreased time after time until high tideand then suddenly decreased.

High tideFlood tide Flood tide

Ebb tide Ebb tide

SSC profiles

W E

S

N

SSC profile showed a lot of fluctuations in the image at 10:30 near the coastal area, which came to be stable as time passed along both the lines.

2008 Green algae : Enteromorpha Pro-liphera

Total removal quantity in Qingdao : about 1 million ton

Sea

During 2008 Beijing Olympic Game, sailing stadium of Qingdao was attacked by green algae.

2008 Green algae monitoring

Background image : May 20, 2008MODIS aqua band1 image

제주도

흑산도

Qingdao

양쯔강 하구

군산

목포

6 월 13 일

Yellow Sea

East China Sea

Korea

China

GOCI June 13, 2011(13:15)

제주도

흑산도

Qingdao

양쯔강 하구

군산

목포

Yellow Sea

East China Sea

Korea

China

GOCI July 18, 2011(14:15)

한국

제주도

흑산도

Qingdao

양쯔강 하구

군산

목포

Yellow Sea

East China Sea

Korea

China

GOCI July 19, 2011(16:15)

2011 Green algae (July 19-20)

A

A

B B

Bohai Bay Oil Spill(by Chosunilbo)

• The sea area polluted in an oil spill in China's Bohai Bay was five times as large as Beijing previously announced. A probe conducted by the Chinese State Oceanic Administration found that some 4,240 sq.km of water, or seven times the size of Seoul, were polluted by oil leaks from the Peng Lai 19-3 oilfield in Bohai Bay, the daily Xin Jing Bao reported Wednesday.

Beijing admitted the oil spill for the first time on July 5, a month after two oil leaks occurred at China's largest marine oilfield on June 4 and 17, saying only 840 sq.km were polluted. But the water quality of a 3,400 sq.km area nearby dropped from Grade 1 to Grade 3.

China National Offshore Oil Corp. and ConocoPhillips, the joint operators of the oilfield, said the oil spill was quickly contained and cleaned up, but ear-lier this week Beijing admitted that oil continues to leak out.

The Chinese government on Wednesday ordered the operators to suspend production until there is no more danger of further spills. Concern is in-creasing about the safety of seafood from the West Sea. The city of Yantai in Shandong Province near the ill-fated oilfield has set up an observation post on the coast to check for pollution.

Oil Spill (GOCI June 13, 2011)

Bohai

다롄

펑라이

Spilled point

Spilled point

7 days

5 days3 days

9 days

GOCI and Model comparison

GOCI discussion

• Enhanced temporal resolution, high perfor-mance of MTF and SNR of GOCI show better ef-fectiveness than we expected.

• GOCI has an excellent capability to monitor ocean environment and disaster.

GOCI-II Mission

• Succession and expansion of the GOCI missions

• Global Area (Full Disk)– Establishment of Ocean Observation System to monitor long-term cli-

mate change Evaluation of the Primary Productivity in Ocean -> CO2 absorption ca-

pability of Ocean -> Estimating ‘global warming’- Ocean Environment Monitoring

Variation of eco-system

• Local Area– Environment Monitoring for the efficient management of coastal wa-

ters Fresh water/Polluting material drifts & spreads, Pollution of coastal

waters – Production of fishing ground environment information

Searching fishing ground, Monitoring of aquaculture environment in coastal waters

• Common– For reducing the damage by disaster and catastrophe in Ocean, real

time ocean environment monitoring Spreads of red tide, Monitoring of oil spill & tidal wave

Comparison of GOCI-I and GOCI-II

GOCI-I GOCI-II

Orbit Geostationary Geostationary

Channels 8 13

Special Resolution 500m x 500m250m x 250m

1km x 1km

Covered Area LocalLocal

Global

S/N ~1000 ~1500

Observation Frequency Hourly Daytime Hourly Daytime+ Night

[Local and Global Area][Comparison of Resolution among Ocean Sensors]

Special Resolution(m)

Channels (Bands)

Obser-vation

Fre-quen-cy

Bits

Future ProspectGlobal observation in a day!

GOCI-II/KORDI

GEO-CAPE/NASA

OCAPI/ESA

Thank you