ASIA AND PACIFIC COMMISSION ON AGRICULTURAL STATISTICS · St atis tic s U sing A rea S am pling F rame A pproac h Current Condition and Problem 1 New Method: Area Sample Frame 2 Result:

ASIA AND PACIFIC COMMISSION ON

AGRICULTURAL STATISTICS

TWENTY-SEVENTH SESSION

Nadi, Fiji, 19 – 23 March 2018

Agenda Item 9.2

Improvement of Indonesian Rice Statistics Using Area

Sample Frame (ASF) Approach

Contributed by: Kadarmanto, Head of Food Crops Statistics Division

BPS-Statistics Indonesia

Indonesia

[email protected]; [email protected]

APCAS/18/9.2.3P

OUTLINE

2

Improvement of

Indonesian Rice

Statistics Using

Area Sampling

Frame Approach

Current Condition and

Problem 1

New Method:

Area Sample Frame 2

Result:

The First Round of Jan 2018 3

Conclusion 4

Background

3

• The current key annual agriculture surveys for the annual

esti ates of I do esia’s Food Crops, Estate Crops, a d Horticulture Crops Surveys are not based on sound, scientific

statistical methods and practices.

• There is a pressing need for the methodology of the surveys

to be reviewed.

• BPS is asked to replace them with objective probability

sample surveys based on sound statistical practices.

• There is a will of government (President Executive Office, Vice

President Office) to improve the existing method of rice

statistics in respect to accuracy and timeline

Current Condition and Problem

4

Harvested Area Yield Production

Subjective Measurement:

Eye Estimate Inaccurate Result

Objective Measurement:

Area Sample Frame

(ASF)

Crop Cutting of 2.5 x 2.5 m2 of

plot size not yet covered new

type of planting system.

Improvement in

Methodology

Collected by :

Agricultural Extension

Services at sub-

district

Frequency of data

collection : monthly

(A) Hectar (ha)

Through Crop Cutting

Survey

Collected jointly by BPS Staf

and Agricultural Extension

Services

Frequency of data

collection :

a time of harvest

(B) (ton/ha)

Calculated and

Reported every

four months

District Level

Estimate

A X B (ton)

Hopefully,

after applying

new method

more

accurate

Legal Basis of ASF Use

Estimation of

harvested area

must be obtained

through objective

measurement

Statistical Society

Forum

recommended Area

Sampling Frame

should be used Based on a letter

from

President

Executive Office,

estimation of

harvested area

must use ASF in

2018.

ASF was developed by

The Agency for The

Assessment and

Application of Technology

(BPPT)

ASF

President Executive Office

Construction of Area

Frame

6

Data Input:

Topography Map (GIS Board)

Administration Map

Paddy Field/Wetland Area Map

Land Use Map

Contruction of Paddy Field Area Frame

(Paddy Field Stratification)

Grid Formation:

(6 km x 6 km) dan (300 m x 300 m)

Sampling Model Creation

(Random Sampling)

Segment Sample Extraction

(Stratified Random Sampling)

Overlay of Field Sample Frame into

Extracted Segment Sample

Segment Selection

Selected

Segment Putting Atribut:

location code,

name, etc

Maps of

Segment

Location

Stratification

7

Data used : 1. Paddy Field Map 1 :

10.000

2. Topography Map 1 :

25.000

0 Strata-0 (S-0) : Non-Agricultural Land NO SEGMENT SAMPLE

1 Strata-1 (S-1) : Irrigation Rice Field

2 Strata-2 (S-2) : Non-irrigation Rice Field

3 Strata-3 (S-3) : Possible Rice Field (Dryland Arable)

Stratification in West Java Prov

8

S-1 (Green) : Irrigation Rice Field

S-2 (Blue) : Non-Irrigation Rice Field

S-3 (Red) : Possible Rice Field (Dry Land Arable)

S-0 (White) : Non-Agricultural Land (NO SEGMENT SAMPLE)

Definition of Segment

9

400 square sizing of

300m x 300 m each.

Each square is called segment.

6 km

6 km

- Area of study is divided into square block of 6 km x 6 km .

- The square block is further divided into square segment of

300 m x 300 m. The boundary of the segment is based on

geographical coordinate with fix location.

5% Systematic Random Sampling

10

Replicated to each

square block of

6 km x 6 km

20 segments are

selected among

400 segments in

square block.

11

300 m x 300 m

6 km x 6 km

6 k

m x

6 k

m

Extraction of sample segments steps by systematic random sampling

1. Gridding Areas of Study by 6 km x 6 km

2. Sub-gridding by 300 m x 300m

3. Random start extraction-1:

4. Replicate the pattern of extracted sample segment to other grids

Definition of Dimension

•Segment size: (300x300) m2

Random Sampling Model & Square of 6km x 6km

12

Overlay Random Sampling Model & Paddy Field Area

13

NEXT STEP:

• select 5 % sample segment or

proportional to the size of paddy area in

sub-district systematically.

• sample segments on S-0 are eleminated

• 1 km distance threshold.

Overlay Random Sampling Model & Paddy Field Area

14

Result of Segment

Selected in West Java

Province of Java Island

Segment and 9 observation points

15

Segment of 300 m x 300m = 9 Ha Description: 1. In each selected Segment (300 m x 300m = 9 Ha) and

then further divided into 9 Sub-Segment of 100 m x 100 m

2. The centroid of each Sub-Segment will be the observation

point which is regularly visited every month using

ANDROID BASED APLICATION

Observation of Rice

Growing Phase

16

METHOD OF TAKING PHOTO OF RICE GROWING

PHASE ON THE OBSERVATION POINT

Schedule of Field Observation

17

• The last 7 (sevent) days every month

• The first 5 (five) days of the following month, the result will be

validated by supervisor based on photo of rice growing phase

sent by the surveyor.

The last 7 days

of current month

Day 1-5 of the following

month

Day 6-10 of the following

month

Field Observation Result Validation by

Supervisor Data processing

Field Observation

18

Observation of Rice Growing Phase on selected segment Using Android Mobile Phone

Observation of Rice Growing Phase:

GENERATIVE Rice Field is being planted another type of plant

Field Observation

19

Field Supervision by Deputy Chief Statistician

Effort to observation point

Effort to observation point

Effort to observation point Effort to observation point

Steps of Implementation

20

Area Frame Construction

Survey Preparation

• Survey equipments (field map,

aerial photo/image, GPS,

Survey Form)

• Surveyor (training)

Field Survey

Observe and record growing

stages of rice at each sample

point

Field data delivery

Data processing and

presentation

• Stratification of Area

• Sample Size Determination

• Ekstraction of Sample Segment

AREA SAMPLE FRAME

is constructed by utilizing satellite

imagery and wetland paddy area

map

Observation of growing stage of

rice in the field is done by surveyor

every month

Visualization of rice growing phase

PL V1 V2

G P

B

LL

1-35 day 35 -55 day

55-105 day Uncultivated rice filed

Rice field for other crops

Land preparation for rice Vegetative-1 Vegetative-2

Generative Harvest

Number of Sample Segment

22

Field Data

Number of Sample Segment

23

REGION Number of Segment

S1 S2 S3 Total (1) (2) (3) (4) (5)

Sumatera 2.034 3.444 674 6.152

Java 5.383 2.156 690 8.229

Bali & Nusa Tenggara 432 868 289 1.589

Kalimantan 564 1.643 424 2.631

Sulawesi 908 1.689 330 2.927

Maluku 21 80 112 213

Papua 39 89 52 180

Total Indonesia 9.381 9.969 2.571 21.921

Indonesia 22.087

24

PROVINCE Surveyor Supervisor Total

(1) (2) (3) (4)

11. Aceh 172 59 231

12. Sumatera Utara 296 99 395

13. Sumatera Barat 140 49 189

14. Riau 90 29 119

15. Jambi 83 28 111

16. Sumatera Selatan 257 83 340

17. Bengkulu 63 21 84

18. Lampung 171 58 229

19. Kep. Bangka Belitung 22 6 28

21. Kep. Riau 12 5 17

31. Dki Jakarta 6 3 9

32. Jawa Barat 454 155 609

33. Jawa Tengah 532 175 707

34. Di Yogyakarta 60 22 82

35. Jawa Timur 691 247 938

36. Banten 86 28 114

51. Bali 42 14 56

52. Nusa Tenggara Barat 120 40 160

53. Nusa Tenggara Timur 163 53 216

61. Kalimantan Barat 203 67 270

62. Kalimantan Tengah 114 31 145

63. Kalimantan Selatan 220 74 294

64. Kalimantan Timur 45 17 62

65. Kalimantan Utara 21 8 29

71. Sulawesi Utara 64 24 88

72. Sulawesi Tengah 106 36 142

73. Sulawesi Selatan 271 101 372

74. Sulawesi Tenggara 84 29 113

75. Gorontalo 36 10 46

76. Sulawesi Barat 46 14 60

81. Maluku 28 10 38

82. Maluku Utara 28 12 40

91. Papua Barat 12 9 21

94. Papua 35 19 54

INDONESIA 4 772 1 635 6 407

Number of Surveyor and Supervisor

by Province

Dot Sampling Method

25

• A New Survey Method for Area Estimation introduced by Issei

Junguji.

• Improvement of Point Sampling Method:

– Dr. Frank Yates If accurate large-scale maps showing the field

boundaries are available, the point method of sampling is very suitable

for crop surveys in which contact with farmer is not necessary. The

fields will then act as sampling units, and selection will be with

probability proportional to size. Provided the whole of a selected field

is under a single crop, all that is necessary for acreage estimates is to

ascertain the crops, no determination of area being required. If more

than one crop is being grown on a selected field, the proportions of

area under the different crops must be determined, but eye estimate

will usually be adequate for this purpose.

Procedure of Dot Sampling Method

26

• identify the category of land use at sample dots,

• count the number of dots by category to estimate the proportion

Proportio Esti ator . • Do ’t have to measure.

• The number of survey items at a sample dot is only one item.

• Can estimate land use areas by category based on statistical formula, and

calculate reliabilities, namely sampling error.

• Estimator : 𝑇 = 𝑛1𝑛 𝑊 = 𝑝 𝑊 where W= whole area of target; n=number of

sample dots; n1 = number of sample dots which are identified on the

survey item (e.g. rice field)

Procedure of Dot Sampling Method

27

• identify the category of land use at sample dots,

• count the number of dots by category to estimate the proportion

Proportio Esti ator . • Do ’t have to measure.

• The number of survey items at a sample dot is only one item.

• Can estimate land use areas by category based on statistical formula, and

calculate reliabilities, namely sampling error.

• Estimator : 𝑇 = 𝑛1𝑛 𝑊 = 𝑝 𝑊 where W= whole area of target; n=number of

sample dots; n1 = number of sample dots which are identified on the

survey item (e.g. rice field)

Illustration of Areas Estimation

28

The proportio of sa ple ust be sa e as that of populatio . If you distribute sa ple dots,………….

×

It is a basic mathematics. The area of red rectangle is 200cm2. Please estimate the

total area of green figures using the formula. The number of dots on the green areas must be proportional to the size of green area in the population.

The proportio of sa ple ust be sa e as that of populatio . If you distribute sa ple dots,………….

Where n=200

W=10 cm x 20 cm = 200 cm2 𝑇 = 700×200 = 0.365×200=73

24

15

12

9

13

Total 73

Estimator Used

29

jn

1i

ij

j

j pn

1p

j

j

n

i

jij

jj

p ppnn 1

22 )()1(

1

jjj pDA jpjj DAVar 2)(

m

j

jAA1

Pj : average proportion of rice in segment-i, strata-j

nj : sample size in strata-j

Dj: Total area of strata-j

Aj: Estimated rice area in strata-j

A: Total area of rice in the whole Distict

j: number of strata

i: Sample segment i-th

δ: variance

Illustration of recieved field data

30

Indramayu District, total area: 207.675 Ha

Total Segment : 54 Recieved data: 52 data Area frame : 145.750,00 Ha

No Sub-district Segment PL V1 V2 G P LL B H No Sub-district Segment PL V1 V2 G P LL B H

1 Sindang 321470803 0 0 0 0 20 5 0 0 28 Gantar 321462003 0 0 0 0 0 6 19 0

2 Kandanghaur 321470603 0 0 0 0 21 4 0 0 29 Terisi 321466401 0 0 0 0 23 2 0 0

3 Kandanghaur 321466406 0 0 0 0 22 3 0 22 30 Gantar 321462106 0 0 0 0 0 5 20 0

4 Kandanghaur 321470504 0 0 23 0 0 2 0 0 31 Gantar 321462004 0 0 0 0 0 14 11 0

5 Losarang 321466503 0 0 0 0 23 2 0 11 32 Gantar 321457906 0 0 0 0 0 8 17 0

6 Losarang 321470601 0 0 0 0 0 25 0 0 33 Balongan 321466706 0 0 0 0 0 25 0 0

7 Bangodua 321462403 0 0 0 0 20 5 0 0 34 Widasari 321462406 0 0 0 0 25 0 0 0

8 Cantigi 321470704 0 0 0 0 0 21 4 0 35 Kroya 321466301 0 0 0 0 0 0 25 0

9 Bangodua 321462304 0 10 0 0 6 4 5 0 36 Kroya 321462206 0 0 0 0 0 3 22 0

10 Sliyeg 321466601 0 0 0 0 24 1 0 0 37 Kroya 321462103 0 0 0 0 0 5 20 0

11 Sliyeg 321466604 0 0 0 0 5 20 0 0 38 Kroya 321462104 0 0 0 0 0 0 25 0

12 Kertasemaya 321462503 0 0 0 0 13 12 0 0 39 Cikedung 321462306 0 0 0 0 0 5 20 0

13 Krangkeng 321462603 0 0 0 0 0 2 23 0 40 Lohbener 321466606 0 0 0 0 23 2 0 0

14 Krangkeng 321458406 0 0 0 0 0 25 0 0 41 Lelea 321466501 0 0 0 0 12 1 12 0

15 Anjatan 321470401 0 0 0 24 0 1 0 0 42 Lelea 321466504 0 0 0 0 6 2 17 0

16 Juntinyuat 321466704 0 0 0 0 20 0 5 0 43 Losarang 321466404 0 0 0 0 25 0 0 0

17 Juntinyuat 321466701 0 0 0 0 18 7 0 0 44 Losarang 321466403 0 0 0 0 0 25 0 0

18 Cikedung 321462201 0 0 0 0 0 5 20 0 45 Tukdana 321458206 0 0 0 0 12 13 0 0

19 Cikedung 321462204 0 0 0 0 0 8 17 0 46 Tukdana 321462401 0 0 0 0 25 0 0 0

20 Anjatan 321466206 0 0 0 20 0 5 0 0 47 Patrol 321470506 0 0 0 0 9 16 0 0

21 Anjatan 321470403 0 0 18 0 0 7 0 0 48 Patrol 321470404 0 2 22 0 0 1 0 0

22 Haurgeulis 321462006 0 0 0 0 0 0 25 0 49 Sukra 321474601 17 5 0 0 0 3 0 0

23 Haurgeulis 321466204 0 0 0 0 0 2 23 0 50 Sukagumiwang 321458203 0 0 0 0 21 4 0 0

24 Haurgeulis 321466203 0 0 0 0 0 8 17 0 51 Bongas 321470503 0 0 6 15 0 4 0 0

25 Haurgeulis 321466201 0 0 0 0 0 5 20 0 52 Bongas 321470501 0 0 5 14 0 6 0 0

26 Gabuswetan 321466303 0 0 0 0 23 2 0 0 TOTAL 17 17 74 73 412 340 367 33

27 Gabuswetan 321466304 0 0 0 0 16 9 0 0

31

Total observed points: 1.300

Total frame area : 145.750,00 Ha

Rice growing stage Total obs.

point

Proportion

(%)

Area

(Ha)

Land Preparation (PL) 17 1,3 1.895

Vegetative-1 (V1) 17 1,3 1.895

Vegetative-2 (V2) 74 5,7 8.308

Generative (G) 73 5,6 8.162

Harvest (P) 412 31,7 46.203

Other land cover (LL) 340 26,2 38.187

Uncultivated rice field (B) 367 28,2 41.102

Harvest in between 2-

survey (H) 33 2,5 3.644

Harvest next 2-month (Prediction) 16.470

Harvest next 4-month (Prediction) 20.260

Observed Subsegment in Jan 2018

32

22,087 21,154

Target Realisasi

Each surveyor observes 9 point of

observation in each segment.

Total number of observed point visited

by all surveyors is

21.154 x 9 = 190.386

TOTAL SAMPLE SEGMENT

95,90%

33

Harvested Area of Rice in Indonesia Using Area Sampling Frame (ASF) Method

in January 2018

Month Jan Feb Mar Apr

Harvested

Areas (Ha) 573.757,01 ? ? ?

Potential of Rice Harvested Area during Jan-March

Month Jan Feb Mar Apr Jan-Apr

Harvested

Areas (Ha) 573.757,01 1.116.310,23 1.350.640,00 1.514.660,84 4.555.368,08

Description:

January Harvest

February Generative

March Vegetative 2

April Vegetative 1

Compared with the conventional method

34

0

200000

400000

600000

800000

1000000

Har

vest

ed A

rea

(Ha)

ASF Conventional Jan 2018 Conventional Jan 2017

35

Calculation of Surplus/Deficit of Rice Production in Indonesia In January 2018

Method Harveste Area (Thousand Ha)

Production 1)

(Million Ton of Rice)

Consumption 2) 3)

(Million Ton of Rice)

Production -

Consumption (Million Ton of Beras)

ASF 573,76 1,80 2,58 -0,78

Assumption:

1. Production = Harvested Areas x Yield; using 2017 figures

2. Consumption per Capita = 114 kg per capita per year

3. Number of population = Mid-Year Population in 2018

4. Poin 1 to 3 is calculated from provincial base

Surplus/Deficit of

Rice Production

by Province

in January 2018

(Ton of Rice)

36

Conclusion

37

ASF Survey Result Map of West Java Provinci : Jan 2018

Conclusion

38

• No need to contact farmers.

• Very easy field implementation.

• Calculation of Surplus/Deficit of Rice Production can be done

quickly (Less time to data processing).

• ASF Results are statistically more objective than those

obtained by conventional method.

• Can estimate harvested areas until the next 3 months.

• Need more effort to reach the observation points in certain

areas, especially in mountain areas

References

39

• Frank Yates, (1949), SAMPLING METHODS for CENSUS AND

SURVEYS

• Jinguji, Issei. Dot Sampling Method for Area Estimation.

Ministry of Agriculture, Forestry, and Fisheries. Kamakura,

Japan.

• Mubekti. Sampling Frame of Square Segment by Points for

Rice Area Estimation. The Agency for the Assesment and

Application of Technology. Jakarta, Indonesia.

Thank you

40