Practical management of climate variabilityin a changing climate

Peter McIntosh, Senthold Asseng, Dean Thomas, Guomin Wang and Nirav Khimashia

Overview

• Simulate growing wheat crop in southwest WA using APSIM

• Explore two different N application strategies• maximise income (risky)• $1 N must return $2 (risk averse, realistic)

• Vary N rate using seasonal climate model forecast• May-Oct rainfall forecast (above/below median)

• Show that the forecast• is much more valuable for the realistic N management strategy• requires fewer good years for greater returns (climate change)

• Explore risk and payoff time for forecast

• Explore mixed wheat-sheep farming

Nyabing

Good years decreasing

1900 1920 1940 1960 1980 20000

50

100

150

200

250

300

350

400

450

Time

Nyabing (Rain)

May

-Oct

(m

m)

mean = 289, slope = -8.2 mm/decade

0 10 20 30 40 50 60 70 80 90 1000

20

40

60

80

100

120

% of years

% o

f lo

ng

-te

rm in

com

e

Nyabing

The importance of good years

80% of income in 40% of years

Simulated wheat crop 1980-2006

• Crop simulation model APSIM-Nwheat

• Local variety, clay soil

• Reset soil moisture to no plant-available stored soil water each year

• Reset to 50 kg mineral N per ha each year

• Plant after enough rain (10mm over 10 days)

• Apply nitrogen (N) at sowing (>80kg applied 4 weeks later)

• Gross Margin (GM) = income – operating cost – N cost

• Wheat price $200/t (± protein premium/penalty)

• Operating = seed, sprays, diesel, insurance, interest etc ($150/ha)

• N cost = $1 per kg

Applied N (kg N/ha)

0 20 40 60 80 100 120 140 160 180 200

Gross margins (A$/ha)

-200

0

200

400

Nyabing, clay soil, 1980-2006

$2 per $1 N

$1 per $1 N

How much N (nitrogen fertiliser) to add?

All yearsMaximize GM

Why be risk averse?

3 bad years in a row can lose the farm

1 really bad year can compromise the next 2-3

1980 1985 1990 1995 2000 2005-400

-200

0

200

400

600

800

Year

GM

($

/ha

)

risk aversemax GM

5 10 15 20 25

-200

-100

0

100

200

300

400

500

600

number of years

Gro

ss m

arg

in (

$/h

a)

Risk averse

5 10 15 20 25

-200

-100

0

100

200

300

400

500

600

number of years

Gro

ss m

arg

in (

$/h

a)

Maximise GM

Number of years to be 95% sure of breaking even

3 years 6 years

Applied N (kg N/ha)

0 20 40 60 80 100 120 140 160 180 200

Gross margins (A$/ha)

-200

0

200

400

Above median

Below median

Nyabing, clay soil, 1980-2006

What if we had a rainfall forecast?

All years

$2 per $1 N

Potential value of a forecast

StrategyN (kg/ha)

above/below

GM ($/ha)

above/below

ave

Farm Income ($/2500ha)

Climatology

(risk averse)0 / 0

166 / 27

94235,000

Climatology

(maximise GM)77 / 77

376 / -32

164410,000

Correct 2 category

(risk averse)60 / 0

378 / 27

196490,000

Correct 2 category

(maximise GM)80 / 0

410 / 27

211527,500

Gain $59 by not fertilising in bad years (max GM only)

Gain $212 (risk averse) or $134 (max GM) by fertilising in good years

Risk averse strategy has more to gain from a forecast

risk averse

gain $255K

max GM gain $117K

Seasonal forecast model - POAMA

• Predictive Ocean Atmosphere Model for Australia

• Models the earth system (atmosphere, ocean, ice, land)• Based on dynamics of fluids, range of physical processes• Start from measured state of the ocean• Step forward every 15 minutes for 9 months on global 250km grid• Predicts wind, temperature, rainfall• Climate change built-in

• Start 1 May, predict May-Oct rainfall• 1980-2006

Forecast skill

Longitude

Latit

ude

110°E 115°E 120°E 125°E 130°E 135°E 140°E 145°E 150°E 155°E45°S

40°S

35°S

30°S

25°S

20°S

15°S

10°S

-1

-0.8

-0.6

-0.4

-0.2

0

0.2

0.4

0.6

0.8

1

Longitude

Latit

ude

110°E 115°E 120°E 125°E 130°E 135°E 140°E 145°E 150°E 155°E45°S

40°S

35°S

30°S

25°S

20°S

15°S

10°S

0

10

20

30

40

50

60

70

80

90

100

r=0.32 at Nyabing

(significant at 90% but not 95%)

70% at Nyabing

(significant at 95%)

correlation (r) two-category hit rate

Use POAMA forecast to determine N

1980 1985 1990 1995 2000 2005

Gross margins (A$/ha)

-200

0

200

400

600

Wrong forecast

Below median

With POAMA forecast

Risk averse strategy

StrategyN (kg/ha)

above/below

GM ($/ha)

above/below

ave

Farm Income ($/2500ha)

Climatology

(risk averse)0 / 0

166 / 27

94235,000

Climatology

(maximise GM)77 / 77

376 / -32

164410,000

Correct 2 category

(risk averse)60 / 0

378 / 27

196490,000

Correct 2 category

(maximise GM)80 / 0

410 / 27

211527,500

POAMA 2 category

(risk averse)60 / 0

313 / 20

161402,500

POAMA 2 category

(maximise GM)80 / 0

335 / 13

168420,000

risk averse gain $167K

max GM gain $10K

Forecast value

ClimatologyPOAMA forecast

Correct

forecastForecast efficacy

Risk averse $235,000 $402,500 $490,000 66%

Maximise GM $410,000 $420,000 $527,500 9%

Planted area 2500 ha

How long for a forecast to pay off?

5 10 15 20 25-150

-100

-50

0

50

100

150

200

years

Gro

ss M

arg

in (

$/h

a)

95% chance of exceeding (risk averse)

climatologyforecastperfect

7 years

How long for a forecast to pay off?

5 10 15 20 25-150

-100

-50

0

50

100

150

200

years

Gro

ss M

arg

in (

$/h

a)

80% chance of exceeding (risk averse)

climatologyforecastperfect

3 years

Forecast makes the most of good years

0 20 40 60 80 1000

20

40

60

80

100

120

140

160

180

200

% of years

% o

f lo

ng

term

clim

ato

log

y

no forecastforecast

80% of income in 40% of years

same income in 15% of years

Managing mixed farms with seasonal forecasts

all crop all crop

+FC

crop/ pasture

crop +FC/ pasture

crop +FC/ pasture +FC

dry wet

5 options

Farm

1

Farm

3

Farm

4

Farm

5

Farm

2

Gross margins (A$/farm) (,000)

0

200

400

600

800

All Cro

p

All Cro

p

+ fo

reca

stCro

p /

Pastu

re

Crop +

fore

cast

/

Pastu

re

Crop +

fore

cast

/

Pastu

re +

fore

cast

Nyabing

3000 ha farm, duplex soil

Mixed farming benefits from a POAMA forecast

GM from sheep

Forecast benefit

Summary

• Get much better value from a seasonal forecast using a realistic conservative management strategy

• A moderately skilful forecast (19 out of 27 years correct) can increase farm profit from $235K to $402K

• Using such a forecast pays off in• 7 years (at 95% certainty)

• 3 years (at 80% certainty)

Thank you

CAWCR(The Centre for Australian Weather and Climate Research - A partnership between CSIRO and BoM)Peter McIntoshPrincipal Research Scientist

Phone: +61 3 6232 5390Email: peter.mcintosh@csiro.auWeb: www.csiro.au/cmar

Contact UsPhone: 1300 363 400 or +61 3 9545 2176Email: Enquiries@csiro.au Web: www.csiro.au