Decision support systems in practice - some observations. David Freebairn

Preview:

DESCRIPTION

Presentation from the WCCA 2011 conference in Brisbane, Australia.

Citation preview

Decision support systems in practice- some observations

David FreebairnRPS Brisbane, Australia

Outline

• Simplicity and transparency• Who says the world has to be complex?• Acknowledge stakeholders as experts• Modesty and limitations

• Ideas, comments, suggestions

Experience -background

And many good ideas from my colleagues in DPI, DERM and CSIRO

• Many decisions are simpler than we think

• Many analytic tools are complex, inaccessible or opaque

• Computers are good at simple tasks (e.g. arithmetic)

• Humans are good at complex tasks (e.g. decision making)

General observations

Complexity and Generality

Relativeaerial

applicability

Detail

RulesOf thumb

Simplesimulation

Complex simulationDSS

The struggle between usefulness (goodness) and complexity

http://www.dau.mil/pubscats/PubsCats/atl/2005_11_12/war_nd05.pdf

Soil management, water conservation, erosion

Rainfall simulation - a research and extension tool

Bare soilBare soilStubble coverStubble cover

Simplicity and transparency

• The simplest things generally work best, and the simpler the better.

• The easier a decision support tool is to use and support.

• More complex >> less transparent.• Active demonstrations are most effective

learning tools.

Who says farming is complex?

• Increased complexity is a common pathway for scientists.

• What challenge farm decision making though is uncertainty.

• There is a view that many models should be used in an “instructive” mode.

Acknowledge stakeholders as experts

• Remember who has the greatest vested interest in problem solving.

• The farmer is clearly the best expert, and expert farmers often use a range of other experts to support them.

• Being useful to decision makers requires getting into their shoes.

Modesty and limitations

• Acknowledge external “experts” have small roles to play

• System status-history (weather, previous crops)

-monitoring (soil water, weeds, disease)

• Weather futures - based on history

- forecasts

• Market futures

• Fit in the system

• Personal preferences

Decision point

Tactical decision making- where is the niche for improved information?

• System status-history (weather, previous crops)

-monitoring (soil water, weeds, disease)

• Weather futures - based on history

- forecasts

• Market futures

• Fit in the system

• Personal preferences

Decision point

Tactical decision making- how do farmers view this?

Importance of various elements in decision making – e.g. planting

Climate forecast

adjustment

Gut feeling

Weeds

Price

Soil N

Seed availability

Starting soil water

20%

15%

30%8%

8%

8%

8%

8%

Diseaserisk

Note:Use this figure to

focus discussion on what are the issues and their relative

importance

(no correct answers)

Estimating soil moisture- the simple “push” probe

“2 feet of moisture”

Simple vs. less simple

0

50

100

150

200

250

300

0 50 100 150 200 250 300

Observed (mm)

Acland

Capella

Greenmount

Wallumbilla

Warra

RMSD = 38 mm 1:1 Line

y = 0.76x + 47

R2 = 0.56

0

50

100

150

200

250

300

0 50 100 150 200 250 300

Observed (mm)

Pre

dict

ed (

mm

)

Acland

Capella

Greenmount

Wallumbilla

Warra

RMSD = 28 mm

1:1 Line

y = 0.82x + 29.6

R2 = 0.72

Fallow efficiency

-20% fallow rainfall

HOWWET?

-daily model

Soil cover (%)

0

10

20

30

40

50

0 20 40 60 80 100

Bare fallow

Stubble incorporated

Stubble mulch Zero-till Pasture

Average annual soil loss (t/ha)

Influence of stubble cover on soil erosion

Greenmount (Qld) 1978-88

Seeing, feeling, trialling

Some issues Queensland farmers consider

• What are the chances of a planting rain?• What are current moisture, nitrogen

conditions?• What are implications for yields? • Input needs?

Component questions for simple models

• What are current conditions (e.g. moisture heat sum)?

• What are the chances of a future event (e.g. planting rain, frost, wet harvest)?

• What is skill in a forecast?

• What are the implications of above, and what management options are there to adjust?

Recent Histor

y

Now(the decision

point)

Futureoutcome

RainfallTemperature

Previous crop Soil type

Management

Range of Options

andoutcomes

Current conditions

Soil waterNutritionDiseaseWeeds

-supported by new

observation

Linking conditions NOW and Future probabilities

Expected drivers• Rainfall

• Temperature

Based on • History

• Persistence• forecasts

Time line

Rainfall mm Temperature > OC Temperature < oC Heat sum oC days

What are the chances of getting …

50 3 30 200

In days, between 10

Occurs in % of years between 54 1912-2010

Maximum

in eachyear

Previous analysis

Rainfall Max. temp. stress days Min. temp days Heat sum oC days

How is the season progressing?

Between

Previous analysis

Season to date rainfall from dd/mm/yyyy to dd/mm/yyyy9th , 5th and 1st decile

Enlightened DSS design• Question focused, client focused

• Easy to use and ready access

• Multiple access points

• Transparency

• Information, not advice

• Efficient

• Recognise life cycle

How do we ensure we move 1, 2, 4?

Thankyou

Recommended