Dr. Takeuchi Lab. IIS Tokyo Univ. xx

The MRV system of greenhouse gaseous emission

in the world high-carbon reservoirs

Hironori AraiMasayoshi ManoNaoko SaitohKazuyuki Inubushi

Dr. Takeuchi Lab. IIS Tokyo Univ. xx

Establishing an integrated MRV system ofgreenhouse gas emission from wetlands with

Japanese earth-observation/modelling technologies and a data assimilation technique

RESPONSIBLECONSUMPTIONAND PRODUCTION

CLIMATEACTION

Hironori Arai1,2)

Koji Terasaki2), Takemasa Miyoshi2),

Hisashi Yashiro2)

Wataru Takeuchi1), Kei Oyoshi3),

Lam Dao Nguyen4),Kazuyuki Inubushi5)

1) 2) 3) 4) 5)

Outline

2

０．Motivation to DA (Story taking me here today)

１．Background & Objective

2．Ground observation of greenhouse gas emission

and statistical modeling

3. Satellite remote sensing of GHG emitters

- Cropping calendar & the adjacent fallow length

- Paddy soil/water covered by rice plants

- Top down verification with GOSAT

4. My next work with DA

3

Drainage on peatlands in SE asia

0

400

800

1200

Peat decomposition Peat Fire Japan

CO2 emission (Mt C/year) from peat in south east Asiaand Japanese total emission.

(2005) (2005)(1997-2006)

Hooijer et al., 2006; Hatano, 2009

CO

2 e

mis

sion

(Mt

C /

year)

Riely, 2008

4

Target fields

Natural Forest Drained Forest

Agricultural land Burnt Forest

0

1000

2000

Toma et al. 2011 Arai et al. 2014

IndonesiaPalangka Raya

World(-2003)

N2O

(kgN

ha

-1year-

1)

0

1500

1000

500 0

200

400

600

800

0 6 12 18 24 30

P＜0.01

・▲ - Drained forest ・● - Natural forest・× - Burnt Forest ・■ - Agricultrual land

N2O(mgN m-2 hour-1)

CO

2

(mgC

m-2

hour-

1)

5/11

Takeuchi, 2013

Takeuchi, 2013

三好2017

Modified from Yasuoka 2015

Cycle from Observation to Countermeasure

To target To solution To target

Observ.Survey

ModellingSimulation

Forcast・eval.Solutiondesign

optimization

management

Evaluation

DATAinformation

Estimation

Strategy Policy

Effect

Observation of the effect

GOSATK-computer

Modified from Yasuoka 2017

Ourcurrent studies

The rollI wanna play

Outline

11








- Top down evaluation with GOSAT


CH4

21

AR21996

Global Warming Potential of CH4 (IPCC)-on a 100-year horizon-

23

AR32001

25

AR42007

34

AR52013

Shindell et al. science, 2012

Schaefer et al. science, 2016

13

% o

f w

orl

d t

ota

lCharacteristics of Agriculture

in Monsoon Asia

0102030405060708090

100

Area Population Nitrogen

Fertilization

Agricultural

GHGs

emissions

Rice

production

GHGs from

paddies

FAO-STAT 2013

Yan et al. 2009

Kg CH4

km-2 year-1

Monitoring/Reporting

long-term changes of

rice cropping frequency,

fallow season management

and inundation status

Verification with the GOSAT

and atmospheric simulation

Monitoring

present status ofwater management

Unveiling

the potential of

CH4 reduction

and the baseline

Future prediction

of CH4 emission

in global scale

Development economic assessment to realize scientific decision making

Outline

15










1631st / Oct / 2011

Obtained annual CH4 emission data so far

17

EF baselines in philippines

(GoP 2014, Basak 2016)

・Single crop (rainy season)・Single crop (dry season)・double crop

(n=3)

alluvial soil

1.1 Mha, 28%

acid sulfate soil

1.1 Mha, 28%potentially acid sulfate soil

0.5 Mha, 13%

Xuan and Matsui, 1998

18

Characteristics of the Mekong delta

19


20


0

4

8

12

16

20

土壌還元焼却持ち出し

0

30

60

90

120

- CF AWD CF AWD CF AWD

0

700

1400

2100

2800

3500

土壌還元焼却持ち出し

■Conventional ■ Straw-use

■AWD

- Reduction of irrigation rate & GHGs (2012-2016)- Increase of rice grains and its quality

GH

Gs

em

issi

on

(t- C

O2

eq.h

a-1ye

ar-1

)

60%reduction

30%reduction

Irrigation r

ate

(mm

ha-1

year

-1)

Gra

in y

ield

(t- 1

4%

mo

istu

re h

a-1Ye

ar-1

)

C/N

ratio o

f gra

ins

0

7

14

21

28

3510%increase

ImprovedRice quality

Strawincorporation

Strawburning

StrawTaken out

Strawincorporation

Strawburning

StrawTaken out

Strawincorporation

Strawburning

StrawTaken out

Strawincorporation

Strawburning

StrawTaken out

Philippines CF

Philippines AWD

PALSAR-2 Lv.1.1

quadruple6m

Volumediffusion

Coherency matrix [T3] generation

Speckle filtering

Lee-refined (7×7)

Field water level & cropping calendar

CH4 flux data & Soil map

Sampling (25-30 pixel）from each ROI

・Statistical analysis

CH4 emission estimation

model

Inundated / non-

inundated paddies

characterization

Freeman-Durden

decomposition

Single scattering

Doublebounce

Field data collection

Hierarchical bayesian modelling

MODIS (MOD13Q1)Normalized Vegetation/Water index,

Days after sowing, 1km

AMSR-218.7 & 23.8 GHz (V), 10km

Normalized frequencyIndex

Local Maximum Fitting-Kalman filter

Validation&

Data Fusion

HH, HV σ0Enhanced Lee filter (3×3)

PALSAR-2 Lv.1.1

SCAN-SAR25-100m

Multi-looking

Co-registration

De-Grandimulti-temporalspeckle filtering

Geocoding,Radiometric calibration

HH, HV σ0

Local Incidence

angles

Inundated / non-

inundated paddies

classification

Flow chart

CH4 emission map

ValidationGOSAT

(Retrieval)10km

NICAM-Chem(LETKF)

Field observation & flux modeling

Remote sensing on field water & vegetation

Validation & integration

with daily EO data

Verification with GOSAT

Straw incorporation timeand amount

Water regime prior to rice cultivation

Water regime duringrice cultivation

A. ①

②

B.

③

CROPflood

CROPNon-flood

CROPNon-flood

>30 days

<180 days

CROP30days

CROP30days

>180 days

IPCC guideline (Tier1)[Emission factor × Scaling factor in IPCC guideline]

Sow

ing

day

s (d

ays

afte

r 2

00

1/0

1/0

8)

The ad

jacent fallo

w’s len

gth (d

ays)Cropping calendar evaluation with

MODIS－NDVI (LMF-KF)

Samples of paddies

Paddy A (x:184, y:229)Double cropping → Triple cropping

Paddy B (x:185, y:220):Triple cropping

DAS←days after sowing

β ←acid sulfate・coastal sandy soil coefficient

CH4 emission on a specific date= γ * carbon_management / non-inundated_fallow / inundated_fallow * water_management *α*β

carbon_management=[exp(-DAS*δ)-exp(-DAS*(δ+ω))+κ]

non-inundated_fallow

= [1+EXP(-1* ζ *(DAS – ι* days of nonflooding days of the former fallow))]

inundated_fallow= EXP(ε * days of flooding days of the former fallow)

water_management= EXP(η * inundated days during the last 10days)

γ, η, δ, ε , ω , ζ , ι, κ←constant (>0)

α ←straw incorporation coefficient

Semi-empirical daily CH4 flux (mg C m-2 day-1) Model

0 days non-flooding fallow 5 days non-flooding fallow

30 days non-flooding fallow 60 days non-flooding fallow

Ob

serv

ed 100

100101

10

1

0.1

0.01

0.001

0.01

Estimated

Da

ily C

H4

fluxe

s(k

g C

km

-2h

-1)

0

50

100

0 20 40 60 80100Days after sowing

CH

4fl

ux

Days after sowing

CH

4fl

ux

Outline

26










Satellite remote sensing of soils

ALOS-2/PALSAR-2– Lband-Synthetic Aperture Radar –

Scattering model decomposition

MITSUBISHI CO, LTD.

C-SAR

X-SAR

L-SAR

High transparency

Double bounce

Volumediffusion

ALOS-2/PALSAR-2– Lband-Synthetic Aperture Radar –

Xuan and Matsui, 199830paddies × 1village

5paddies × 4villages

Alluvial soils

1.1 Mha, 28%

Acid sulfate soils

1.1 Mha, 28%potential acid sulfate soils

0.5 Mha, 13%

Strip map

SCAN SAR

Coherency matrix [T3] generation

Polarimetric decomposition

Speckle filteringLEE refined

(7×7)

Sampling (25-30pixel）from each ROI

&Statistical analysis

Modified from Avtar et al. 2012

Freeman-Durden

Cloud-Pottier

PALSAR-2 Lv.1.1(quad. CEOS)

23 scenes

PALSAR-2 Lv.1.1(SCANSAR CEOS)

105 scenes

Multilooking

Co-registration

De Grandimulti-temporal

filtering

Geocoding&

Radiometriccalibration

HH HVIncidence

angle

Rice paddy masking&

Statistical analysis

Classification of inundated paddies and non-inundated paddieswhich is covered by rice plants

Inundated(fallow)

Inundated(cropping)

Non-inundated (cropping)

Non-inundated (fallow)

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Single scattering(%)1009080706050403020100

10

20

30

40

50

60

70

80

90

100

100

90

80

70

60

50

40

30

20

10

0

Dry fallow (+rice stumps)

Fallow after plowing or flooding fallow

0-20 days after sowing


Inundated Non-inundated



-Freeman-Durden decomposition-

0

Single scattering Single + VolumeSingle (+ Volume)

Specular reflection Volume + Double Double (+ Volume)

Dominant scattering type

SCANSAR (intensity - HHσ0)

31

Flooding season (2015 Oct. 23)Dry season (2015 Apr. 10)

Double bounce detection by SCANSAR (intensity - HHσ0)

32

Dry season (2015 Apr. 10) Rainy season (2015 Jul. 03 )

Flooding season (2015 Oct. 23)Flooding season (2015 Oct. 30) -LANDSAT-8-

Full-polarimetry (3m)

Fallow paddies





Inundated Non-inundated

HV

(d

B)

HH (dB)

-40

-30

-20

-10

-40 -30 -20 -10 0

(a) Quadruple-polarimetry

HV

+H

H t

hre

shold

(d

B*1000)

HH

thre

shold

(d

B*1000)

cosine(incidence angle)

cosine(incidence angle)

HH threshold (dB) = 0.550*HV +12.9*cosine(IA) -11.2

y = 242983x2 - 360742x + 77749R² = 0.7936

-60000

-57000

-54000

-51000

-48000

0.6 0.7 0.8 0.9 1

R² = 0.9321

-20000

-15000

-10000

-5000

0

0.6 0.7 0.8 0.9 1

HV -25dB

103.5

100.5

104.0

103.0

102.5

102.0

101.5

101.0

100%

10%

1%

0.1%

0%

AL

OS

-2/P

AL

SA

R-2

LS

WC

8%6%4%2%0%

AMSR-2 NDFI (Ascending, 18V & 23V)

10%8%6%4%2%0%-2%

MODIS-NDWI

100%

10%

1%

0.1%

0%

103.5

100.5

104.0

103.0

102.5

102.0

101.5

101.0

AL

OS

-2/P

AL

SA

R-2

LS

WC

Floodability analysis

(Cumulative LSWC/observation scenes)

2019 2020 20212015 2016 2017 20182011 2012 2013 20142007 2008 2009 20102004 2005 20062002 2003

AMSR-E, AMSR-2,3 (10km, 1day, cloud-free)

MODIS (250-1000m, 1day, cloud-biased)

ALOS-2

ALOS-4

NISAR

Sentinel-1A

Sentinel-1B

GOSAT-2

Sentinel-1C

MRV and available data

GOSAT-1

GCOM-C (250m)

Monitoring & Verification

Reporting

SCIAMACHY

AIRS, OCO-2

Sentinel-5P

Daily ALOS2-LandSurfaceWaterCoverage estimation

y = 1.00x + 0.31R² = 0.63

0

2000

4000

6000

8000

10000

0 2000 4000 6000 8000 10000

100%

80%

60%

40%

20%

0%0% 20% 40% 60% 80% 100%

Estimated daily ALOS2-LSWC (10km-res.)

Ob

serv

ed A

LO

S2

-LS

WC

(10

km

-res

.)

= (ALOS2floodability*ω+ ζ )* exp(AMSRNDFI*δ-MODISLSVC*δ)

Downscaleto MODIS

spatial resolution

(Thuy 2016)

Paddy distribution Daily inundation status

Daily croppingcalendar

Estimate daily CH4

emission REPORT (250m res., 2002-)

MONITORING with ALOS2 (since 2014)↓

REPORTING with AMSR, MODIS, GCOM-C/W (since 2002,daily)

Need to beVERIFIED !

NICAM-TM # with different altitudes（2000/Jan.）

Direct comparison between GOSAT and emission data is meaningless…

→Need transport model! But,,,,

Check spin-up status

40

2000/2/1 (1 month after) 2002/1/21 (2 years after) 2003/1/16 (3 years after)

80m 80m 80m

7,590m 7,590m 7,590m

Long years are needed for spin up

+ strong dependency on initial condition,,,

→DA is essential!

GOSAT + NICAM-TM GOSAT-2 + NICAM-TM-4DVAR

GOSAT-4DVAR ?

Flux estimation from atmospheric concentrationby omitting multi-collinearity

• No direct emission or apriori info. is required!

Transparent MRVwith NICAM-LETKF!

Back ground covariance matrices

Outline

43










And if possible…

Soil moisture/Drought assessment/GHG emission estimation with AHI-LST and its DA with atmospheric observation data

PM2.5 emission status estimation with AHI & NICAM-LETKF

My next work with DA

Kalnay et al. 2017

Economic assessment of GHG mitigation under various uncertainties

The MRV system of greenhouse gaseous emission

in the world high-carbon reservoirs

Hironori AraiMasayoshi ManoNaoko SaitohKazuyuki Inubushi

Thank youfor your attention

Documents

Dr. Takeuchi Lab. IIS Tokyo Univ. xx