plantation intelligence analysis of commercial data for

OBERTHÜR, ThomasDirector South East Asia Program, International Plant Nutrition Institute

Plantation Intelligence®

Analysis of Commercial Data for Yield and Fertilizer Management in Oil Palm

Conceptual Background

IPNI 4R Nutrient Stewardship ®

OberthÜr et al.IFA Crossroads Asia-Pacific

Kuala Lumpur, Malaysia

Decision Making Uncertainty

Complex interactions in an agronomic system render outcomes of any management

decision

Uncertain

®



STRUCTURAL Uncertainty from internal factors that influence fertilizer efficiency, e.g. EFB applications

TRANSLATIONALUncertainty from external factors that reduce fertilizer performance, e.g. harvest, mill and transport efficiency

TEMPORALUncertainty about

timing of fertilizer applications, e.g.

drought interference

METRIC Uncertainty about rate & placement of

fertilizer to support a yield target

Cook et al. Better Crops 97, 17 - 20 (2013)

Decision Making Uncertainty

Example:

ROI in Fertilizer

®



Hans Peter Luhn, IBM Journal, 1958A Business Intelligence System

http://www.bireports.co.uk/blog/tag/hans-peter-luhn/

(business) intelligence is “the ability to apprehend the interrelationships of presented facts in such a way as to guide action towards a desired goal.”

Business Intelligence ®



http://www.bireports.co.uk/blog/tag/hans-peter-luhn/

Business Intelligence ®





System is Monitored in Extraordinary Detail

Plantation Intelligence® ®

Devisespecific

performance intervention

options(EVALUATE)

Organizeexisting

performance data

(VISUALIZE)

Quantify performance change for

management (DECIDE)

Generate performance

indicators and metrics

(ANALYSE)


Exogenous

Exogenous

®




An adaptive learning process

based on the analysis of

plantation data

®



MeasurePerformance

PlantationIntelligence®

Business Opportunity

PALMSIMModel

Continuous Improvement

Best Management Practices

EstateNetworking

TheBenchmark Club

Plantation Intelligence® ®



Application Examples

Current Protocols

Yield AgeProfiling

Naïve Gross Margins

Yield Trends Yield Soil ClimateInteractions

Yield Labor Interactions

Yield SoilInteractions

Fertilizer ResponseAnalyses

®



0

40

30

20

50

60

3 4 11975 13 14 15

10

6 8 10 12 TREEAGE

FFB

(t/h

a)

Yield Age Profiling ®

TREEAGE0

40

30

20

50

60

10

FFB

(t/h

a)

3 4 11975 13 146 8 10 12

Yield Age Profiling

15

®

TREEAGE

PALMSIM ModelAgricultural Systems, 131:1-10 (2014)The Planter, 91: 81-96 (2015)

0

40

30

20

50

60

10

FFB

(t/h

a)

3 4 11975 13 146 8 10 12

Yield Benchmarking

15

1. Average yield profile mirrors that of potential

2. Yield gap between average and potential

3. Blocks approach yield potential

4. Blocks above potential require further assessment

®

Aver

age

naïv

e gr

oss

mar

gin

in U

SD

per

hec

tare

Individual Block0

4000

3000

2000

5000

1000

3500

2500

1500

4500

5500

500

Naïve Gross Margins

Cost = 500 $US per ha per year“i.e. a favorable benchmark”

®

Aver

age

naïv

e gr

oss

mar

gin

in U

SD

per

hec

tare

Individual Block0

4000

3000

2000

5000

1000

3500

2500

1500

4500

5500

500

Naïve Gross Margins ®

Response to Fertilizer

NPKMg

8~10BREAK–EVEN

kg kg

point

fresh fruitbunches per kg nutrients

Return on Investment ®



2.5 years to 0.5 year before harvestSum of NPKMg

Return on Investment

NPKMg

®



20

15

10

25

30

100 300 1100900700500

5

600 800 1000 1200

NPKMg3yW_IN

Fres

h fru

it bu

nche

s in

tons

per

hec

tare

35

Steep Ascent : Palm age from 3-5

200 400

25

20

15

30

35

10

40

300 400 800700600500 550 650 750 850350 450

45

Fres

h fru

it bu

nche

s in

tons

per

hec

tare

NPKMg3yW_IN

y=22.233 + (0.005*x)y=12.908 + (0.011*x)

Plateau : Palm age from 6-13

Tree Age Effects ®



10

34

30

26

22

18

14

Fres

h fru

it bu

nche

s in

tons

per

hec

tare

NPKMg3yW_IN

12

32

28

24

20

16

Annual Response 2012

360 400 640600560520480440340 380 620580540500460420 680660

®



10

34

30

26

22

18

14

Fres

h fru

it bu

nche

s in

tons

per

hec

tare

NPKMg3yW_IN

12

32

28

24

20

16


360 400 640600560520480440340 380 620580540500460420 680660

y=21.142 + (-0.003*x)

y=32.489 + (-0.015*x)y=14.481 + (0.024*x)

y=36.217 + (-0.022*x)

SMG ASMG BSMG CSMG D

®



NPKMg3yW_IN


8

30

26

22

18

14

12

32

28

24

20

16

10

34

360 400 600560520480440340 380 620580540500460420

Fres

h fru

it bu

nche

s in

tons

per

hec

tare

®



NPKMg3yW_IN


8

30

26

22

18

14

12

32

28

24

20

16

10

34

360 400 600560520480440340 380 620580540500460420

Fres

h fru

it bu

nche

s in

tons

per

hec

tare

y=3.256 + (0.036*x)

y=50.634 + (-0.054*x)y=13.470 + (0.025*x)y=13.972 + (0.025*x)

SMG ASMG BSMG CSMG D

®



Yield Response (t FFB ha y) = F1 (PV1, PV2, … PVN, PVNPKMg)Yield Response (t FFB ha y) = F2 (PV1, PV2, … PVN)Response Contrast (t FFB ha y) = F2 – F1

Visualizing Local Response

‘Contrast Method’

®



2013 2014

Visualizing Local Response

1.40

0.00

-0.75

Contrast in t FFB per ha per year relative to yield

without fertilizer

®



“Disclaimer”

10

15

5

30

25

20

40

35

15 4525 355 10 20 30 40

45

50

55

Harvest Man days

Fres

h Fr

uit B

unch

Yie

ld in

t pe

r ha

20032004200520062006200720082009201020112012

Confounding Effects ®



10

15

5

30

25

20

40

35

15 4525 355 10 20 30 40

45

50

55

Expected yield level

Fres

h Fr

uit B

unch

Yie

ld in

t pe

r ha

20032004200520062006200720082009201020112012

Harvest Man days




10

15

5

30

25

20

40

35

15 4525 355 10 20 30 40

45

50

55

Fruit is grown but not harvested

Only with enough labor, all fruit is harvested

Expected yield level

Fres

h Fr

uit B

unch

Yie

ld in

t pe

r ha

20032004200520062006200720082009201020112012

Harvest Man days




VS.

Yield Taken

Yield MadeEXPECTATION

REALITY




4WA

IPNI Southeast Asia Program

IFA Crossroads Asia-Pacific20-22 October 2015


plantation intelligence analysis of commercial data for

Documents