estimation of soil carbon stock changes in japanese ...€¦ · inventory development after 2008...

Post on 07-Oct-2020

3 Views

Category:

Documents

0 Downloads

Preview:

Click to see full reader

TRANSCRIPT

Estimation of Soil Carbon Stock Changes in Japanese Agricultural Soils

using National Resources Inventoryg y

Yusuke TAKATA

National Institute for Agro-Environmental Sciences

Natural Resources Inventory Center,

takatay@affrc.go.jp3-1-3 Kannondai, Tsukuba, Ibaraki, 305-8604 JAPAN305-8604, JAPAN

ContentsContentsBackground 1. Past national report about soil carbon stock (SCS).2. Development of agro-environmental inventory after the past reportthe past report.

Body 3. Comparison accuracy among two SCS estimation methods. Categorical method and Hybrid-kriging method4. Changes of SCS (0-30cm) in Japanese agricultural land.

5. SCS in agricultural land at different depth

INTRODUCTION Paddy Fields

Soil carbon stock (tC/ha) in agricultural land on

Paddy Fields Orchard①

③ 1990 was summarized in NGGI using “Legacy data”

④⑤

Upland Fields

⑧⑨

Grassland⑩

⑫⑬⑬

INTRODUCTIONSoil carbon stock (tC/ha) in agricultural land on 1990 was summarized in NGGI using “Legacy data”

“Legacy data”; “Legacy data”;

1) Basic Soil-Environment Monitoring (Stationary Monitoring)(Stationary Monitoring)

2) Old cultivated soil map (1970’s),3) Old land use map (before 1970’s)3) Old land use map (before 1970 s)

Basic Soil - Environment Monitoring Project

(St ti M it i )(Stationary Monitoring)About 20 000 monitoring sitesAbout 20,000 monitoring sites.5-year-interval (Since 1979)

Figure . Location of stationary 

ContentsSite information, Soil characteristics , and Farmland managements g y

monitoring .Soil characteristics , and Farmland managements (fertilizer , organic matter and soil amendment application, kind of crops, yield, etc)

Dataset was divided into two groups; Validation dataset: Parameterization dataset = 1:10a dat o dataset a a ete at o dataset 0

INTRODUCTIONSoil carbon stock (tC/ha) in agricultural land on 1990 was summarized in NGGI using “Legacy data”

“Legacy data”; “Legacy data”;

1) Basic Soil-Environment Monitoring (Stationary Monitoring)(Stationary Monitoring)

2) Old cultivated soil map (1970’s),3) Old land use map (before 1970’s)3) Old land use map (before 1970 s)

Inventory Development after 2008Two versions of Soil land use map (1992 and 2001)Two versions of Soil-land use map (1992 and 2001).

04 04 03: Andosols

03  03 03  03  Andosols

04: Wet Andosols

Tsukuba Tsukuba

10  10 

1992 version 2001 version06: Brown

06  06 

Brown Forest soil 09: Red soils

09 09

Red soils 10: Yellow soils

Southwest Japan Southwest Japan1992 version 2001 version

Southwest Japan Southwest Japan

Cultivated Soil –l d M land use Map (2001 version)(2001 version)Japanese cultivated soil classification

Soil groups: 16Soil groups: 16Soil series groups: 60 Soil series: 320

Andosols

Soil series: 320

Wet AndosolsBrown Forest soilsGray Lowland soilsGray Lowland soilsGley soilsOther Soil Groups

Inventory Development after 2008Two versions of Soil land use map (1992 and 2001)

04 04 03: Andosols

Two versions of Soil-land use map (1992 and 2001).

03  03 03  03  Andosols

04: Wet Andosols

Tsukuba Tsukuba

10  10 

1992 version 2001 version06: Brown

06  06 

Brown Forest soil 09: Red soils

09 09

Red soils 10: Yellow soils

Southwest Japan Southwest Japan1992 version 2001 version

Southwest Japan Southwest Japan

Cultivated soil area in JapanSoil Group Area (x1000 ha)No. of

1973 1992 2001Gray Lowland 9 Fluvisols, 1275 1157 1072

Soil Group(WRB)

Area (x1000 ha)Soil Group

No. ofSoil Series Group

ysoils

9 ,Gley soils

1275 1157 1072

Gley soils 7 Fluvisols 1027 908 848Gley soils 7 Fluvisols 1027 908 848

Andosols 5 Andosols 1007 944 879

Brown Forestsoils

3 Cambisols 483 426 362soils

Wet Andosols 5 GleyicAndosols

384 419 397

Other soils 31 1498 1347 1231

Total 5675 5202 4790

Inventory Development after 2008Soil temperature regime map (1km grid)

3.0 

4.0 

T (o

C) Eq. 1

Eq 2 Eq 3

Soil temperature regime map (1km grid)

1.0 

2.0 

Soi

l T-A

ir Eq. 2 Eq. 3

0.0 

0 250 500 750 1000

Altitude (m)Diff

S

f(Diff Soil T- Air T (oC)) =0.0015*Altitude + 1.8 (N 108 R2 0 21) E 1

Map of Diff Soil T- Air T (oC)

Soil tempreturemonitoring sites(N=108, R2=0.21) Eq.1.

Diff Soil T- Air T (oC) = f(Diff Soil T- Air T (oC)) + ε Eq. 2ε; residuals of Eq. 1.(measured value –predicted value),

monitoring sites

Frigid Mesic Thermic

Hyperthermic

ε; residuals of Eq. 1.(measured value predicted value), and it was interpolated by Ordinary Kriging

Soil temperature map =Air temperature map + ff S ( C) Unknown (Altitude > 1000m)

Soil Temperature regime map

Diff Soil T- Air T (oC) Eq. 3

National Resources Inventory for Estimating SCS in Japanese Estimating SCS in Japanese

Agricultural Land 1) Basic Soil-Environment Monitoring (Stationary Monitoring)(Stationary Monitoring)

2) Updated “soil - land use” map (1992 and 2001 version),,

3) Annual agricultural census3) Annual agricultural census

4) Soil temperature regime map4) Soil temperature regime map

Comparison accuracy among t SCS ti ti th dtwo SCS estimation methods

Categorical MethodCategorical MethodV S

Hybrid-kriging MethodV. S.

Hybrid kriging Method

Categorical methodPedo transfer depth function

Soil carbon content (g/kg)

“Basic Soil-Environmental Monitoring Project” data Deficit data

Calculating the increase/decrease ratio of C content or bulk density for 0-1cm layer to each layer per soil groups and

Pedo-transfer depth function

Soil carbon content (g/kg) Bulk density 0 cm 0 cm 0 cm

layer to each layer per soil groups and land use type (64 SG_LU categories).

0 cm

1cm 1cmX

30 cm 30 cm

?

30 cm 30 cm30 cmAveraging per soil series group and land use type (SSG LU categories) in Th t t i bl f th l

Increase/  Decrease Ratio

(SSG_LU categories) in each 1cm layer.

The target variable of the layers was computed by Increase/Decrease Ratio by top layer’s variable

60 SSG and 4 land use (paddy, upland, orchard, grassland)240 SSG_LU categories

SCS (0-30cm; tC/ha) in the Major Soil Groups for 1st Interval (1979-1983)180 Groups for 1st Interval (1979-1983)

120 120

60

0

Paddy field Upland fields Orchard GrasslandCategorical Method

SCS (0-30cm; tC/ha) Map Delineated by Categorical Method Categorical Method

SCS Map Soil Map

0 540‐5454‐6969‐102 Andosols102‐150150‐250

Wet AndosolsBrown Forest soilsGray Lowland soilsGray Lowland soilsGley soilsOther Soil Groups

Spatial biased of the categorical method Residuals (tC/ha)( )= Measured SCS values – Estimated SCS values Residuals (tC/ha) > 0; Underestimation

( C/ ) 0 O

a b c d a b c c3030

C/h

a)

C/h

a)

Cool region; Underestimated Residuals (tC/ha) < 0; Overestimation

00

sidu

als

(tC

sidu

als

(tCCool region; Underestimated Warm region; Overestimated

1st interval (1979-) 3rd interval (1989-)a b c c a a b ab

-30-303030

Res

ha)

Res

ha)

Warm region; Overestimated

Frigid Mesic00

dual

s (tC

/h

dual

s (tC

/h

4th interval (1994-)2nd interval (1984-)Frigid Mesic

Thermic Hyperthermic

Soil temperature regime map-30-30 R

esid

Res

id

Soil temperature regime map

Residuals (prediction error) of SCS estimation using the categorical method have a negative correlation with soil temperature regime

Hybrid-kriging methodR id l = M d d t E ti t d d t E 1Residuals = Measured data – Estimated data Eq.1

Estimated data was provided by Categorical Method (SCS Point data) (Map data)(Point data)

ResidualsMap = Ordinary kriging (Residuals) Eq.2(Map data) (Point data)

SCS_Map = Estimated data + ResidualsMap Eq.3(Map data) (Eq. 2; Map data)(Map data)

E 3 Eq.3Eq.3

E 1 Eq 2Eq.1 Eq.2

ResidualsSCS map

(Categorical) SCS_Map (Hybrid)ResidualsMap

Number of the samples in each dataset

Method Parameter Residuals Validati

dataset

16 793 9351st int Categorical

ization Mapping tion

16,793 93516,793 8,698 935

1st int. Categorical(1979 -) Hybrid-kriging

16,774 81716,774 7,561 817

2nd int.(1984 -)

CategoricalHybrid-kriging

16,639 82516,639 5,740 825

3rd int.(1989 -)

CategoricalHybrid-kriging

14,750 60714,750 5,582 607

4th int.(1994 -)

CategoricalHybrid-kriging

Validation of the estimation of SCS in cultivated soil (0-30cm)SCS in cultivated soil (0-30cm).

RMSE MEMethod(tC/ha) (tC/ha)Validation #

1st int. Categorical 41.4 -10.5(1979 -) Hybrid-kriging 38.1 -0.7

(tC/ha) (tC/ha)

935935( ) Hybrid kriging 38.1 0.7

2nd int. 39.8 -8.4(1984 ) 36 5 1 9

CategoricalHybrid kriging

935

817817(1984 -) 36.5 -1.9

3rd int. 38.5 -8.8(1989 ) 35 7 3 1

Hybrid-krigingCategoricalH b id k i i

817825

(1989 -) 35.7 -3.14th int. 39.0 -6.5(1994 ) 36 3 1 0

Hybrid-krigingCategoricalH b id k i i

825607607(1994 -) 36.3 -1.0Hybrid-kriging

RMSE; Root mean square error, SQRT{1/N*Σ(mean error)2}

607

RMSE; Root mean square error, SQRT{1/N Σ(mean error) } ME; Mean error, 1/N*Σ(Measured values – estimated values)

SCS (0-30cm; tC/ha) Map

Th i

102‐150150‐250

Thermicregion

0‐5454‐6969‐102102‐150

Thermicregion

Categorical Hybrid kriging489 Tg 455 Tg

Categorical Method

Hybrid-krigingMethod

Changes of SCS (0-30cm; tC/ha) in each land use typeeach land use type

160

120

80

40

0

Hybrid-kriging Method

Changes of soil carbon content (0-30cm; Tg) in Agricultural land(0 30cm; Tg) in Agricultural land

480

180

240

240

360

120

180

120

240

60

120

0

120

0

60

0

Hybrid-kriging Method

SCS at different depthFor calculating the SCS at different depth per SSG_LU category (0-50 cm, 0-

The SCS dataset in each 5 year

p p _ g y ( ,100 cm) in each 5 year interval

yinterval were merged.

The average SCS in each 1-cm layers were re-calculated from 0 to 100cm per soil groupsNumbers of the !to 100cm per soil groups

samples were limited!

SCS (tC/ha) at different depth400 0-30cm

Third interval (1989-1993)250

300 30-50cm

50-100cm150

200

Area Area

200

100

150 5,193 x 1000 haTotal carbon content

1,195 Tg

24,294 x 1000 haTotal carbon content

4,750 Tg

100 50

100 , g , g

0 0

Agricultural land Forest

0-30cm 30-100cm

Forest soil data source; Morisada et al. (2004) Geoderma, 119, 21-32

ConclusionConclusion

Categorical method have spatial bias. Hybrid-kriging method can overcome the spatial bi d it id hi h ti ti

1.bias, and it provides high accuracy estimation.

SCS in agricultural land was gradually increased during the monitoring period, but soil carbon

t t i i lt l l d d d ith 2.

content in agricultural land was decreased with decreasing agricultural land area.

Changes of SCS (0-30cm; tC/ha) in each soil groupeach soil group160

120

80

40

0

Hybrid-kriging Method

Basic Soil - Environment Monitoring Project

(Stationary Monitoring) (Stationary Monitoring) 800

O5m

g/100g)

600

osp

hat

e (

P2O

4001979-

1984-

Ava

ilabl

e P

ho

200

0

1989-

1994-

1998

land use

greenhousepastureorcharduplandpaddy

Changes of available phosphate contents of surface soilsChanges of available phosphate contents of surface soils during 1979-1998. (Obara et.al 2004)

top related