The Value of ENSO Forecast Information

1

• The Value of ENSO Forecast Information
• To Dual Purpose Winter Wheat Production
• In the U.S. Southern High Plains
• Steve Mauget
• USDA-ARS Plant Stress Water Conservation Lab,
  Lubbock, TX
• John Zhang
• USDA-ARS Grazinglands Research Lab, El Reno, OK
• Jonghan Ko
• USDA-ARS Agricultural Systems Research Unit, Ft
  Collins, CO

2
Analog Years Method
1) Given set of analogous years in historical
record marked by a certain forecast condition
over a growing region
2) For each analog year, conduct cropping
simulations
3) Repeat simulations for a range of management
practices
4) Determine which practice is optimally
profitable for that forecast condition, assuming
certain price and cost conditions
Net Profit Distribution (Best Forecast Practice)
P()

3
Analog Years Method Forecast Value

• Define a second set of analog years, that include
  the entire historical record (e.g. 1915-1999)
• Repeat process 1-4 to define a best management
  practice for climatological (i.e., No Forecast
  ) conditions
• Form a distribution of profit outcomes for the
  forecast analog years, using the best No-Forecast
  practice

Profit Distribution (Best No-Forecast Practice)
P()

4
Average Forecast Profit Effect
ltFVgt
ltFVgt lt (Forecast)gt - lt(No-Forecast)gt
Profit Distribution (Best No-Forecast)
Profit Distribution (Best Forecast)
Where, lt(Forecast)gt Average profit from
best management practice for
the specified forecast condition.
lt(No-Forecast)gt Average profit from best
management practice when no
forecast information is available.

5
NIN-3 ENSO Phase Forecast System
Correlation of December-January-Februrary (DJF)
Panhandle Rainfall with DJF SSTA
6
May-June-July (MJJ) Niño-3
SSTA Phase
vs. November-March (NDJFM) Panhandle
Precipitation Tercile
(85 Years 1915-1999)
NDJFM Panhandle Precipitation
Dry (lt 66 mm)
Normal
Wet (gt 96 mm)
Total
Cold (lt -0.5 C)
10 3 1 14
MJJ Niño-3 SSTA
13 18 15 46
Neutral
Warm (gt 0.5 C)
5 8 12 25
28 29 28 85
Total
7
Dual Purpose Winter Wheat Production



Aug. Sep. Oct.
Nov. Dec. Jan. Feb. Mar. Apr. May
Jun. Jul.
Planting
Dormant Period Grazing
Heading Grain Filling
8
Tactical vs. Strategic Forecast Value
ltFVgt lt (Forecast)gt -
lt(No-Forecast)gt
Profit Distribution (Best No-Forecast)
Profit Distribution (Best Forecast)
Forecast Value Distribution
Min
Max
Median
33rd
66th
9
Q Why Tactical Forecast Value ?
A To provide a probabilistic Track Record of
the consequences of using forecast
information in a single year.
Yakima River Valley (1977) Glantz, M.H., 1982
Consequences and Responsibilities In Drought
Forecasting The Case of Yakima, 1977, Water
Resourc. Res., 18(1), 3-13
Zimbabwe (1997) Patt, A.G. et al., 2007
Learning from 10 Years of Climate Outlook Forums
in Africa, Science, 318, 49-50.
10
Q Why Tactical Forecast Value ?
A Seasonal climate forecasts are
probabilistic The profit effects of
forecast information are also probabilistic
11
Methods Dual Purpose Simulations

• DSSAT winter wheat model grazing subroutine
  (J. Zhang)
• 85 years of simulation (1915-1999) at 3 farm
  sites using USHCN daily weather records.

12
Methods Management Options

• 5 planting dates
• Aug. 24, Sep. 8, Sep. 23, Oct. 8, Oct.
  23.
• 5 nitrogen (N) application rates
• 30, 60, 90, 120, or 150 kg ha-1
  applied at planting.
• 5 stocking rates (SR)
• 0.5, 1, 1.5 or 2 head ha-1, or no
  grazing (SR0.0 head ha-1).

13
80 Dual Purpose Grain Grazing Profits
25 Grain Only Grain Profits Only
20 Grazing Only Live Weight Gain Profits Only
1 Fallowing Option Net Profit 0.0 / ha



Aug. Sep. Oct.
Nov. Dec. Jan. Feb. Mar. Apr. May
Jun. Jul.
14
(No Transcript)
15
Analog Years NIN-3 Phase Forecasts Forecast
Dry, Normal, Wet Years
16
Analog Years Perfect Dry, Normal, Wet Years
17
Price Cost Conditions
Wheat Prices 3.22 / bu Historical (1978-2006)
Mean 7.00 / bu Elevated Price (Sept. 2007)
Live Weight Gain (LWG) Value 0.75 / kg LWG -
Leased Pasture Rental Rate 2.42 / kg LWG Wheat
Producer Owns Cattle
Production Costs Texas Coop Extension 2007
dryland wheat and cow-calf budget.
18
Four Production Scenarios

• Historical Wheat Price Leased Pasture
• Historical Wheat Price Own Cattle
• Elevated Wheat Price Leased Pasture
• Elevated Wheat Price Own Cattle

19
Historical Wheat Prices - Leased Pasture
Conditions ( 3.22 /bu )
(0.75 / kg LWG)
No-Forecast Profit (/hectare)
Planting Date Applied N Stocking Rate
Forecast Value (/hectare)
Perfect Wet Perfect Normal Perfect Dry Forecast
Wet Forecast Normal Forecast Dry
Best Management Practice By Forecast
Condition
20
Elevated Wheat Prices Leased Pasture
Conditions ( 7.00 bu )
(0.75 / kg LWG)
21
Q Commodity Price Determines Forecast Value ?
No-Forecast Profit (/hectare)
Planting
Date Applied N Stocking
Rate
3.22/ bu Wheat 0.75/ kg LWG
Forecast Value (/hectare)
Perfect Wet Perfect Normal Perfect Dry Forecast
Wet Forecast Normal Forecast Dry
Best Management Practice By Forecast
Condition
No-Forecast Profit (/hectare)

7.00/ bu Wheat 0.75/ kg LWG
Forecast Value (/hectare)
Perfect Wet Perfect Normal Perfect Dry Forecast
Wet Forecast Normal Forecast Dry
22
A Profit Margin Determines Forecast Value
7.00/bu, 0.75 / kg LWG Production Costs 2
No-Forecast Profit (/hectare)
Planting Date Applied N Stocking Rate
Forecast Value (/hectare)
Perfect Wet Perfect Normal Perfect Dry Forecast
Wet Forecast Normal Forecast Dry
Best Management Practice By Forecast
Condition
23
Forecast Skill Forecast Value?
Forecast Value ( 3.22/bu, 2.42/kg LWG)
Perfect Wet Perfect Normal Perfect Dry Forecast
Wet Forecast Normal Forecast Dry
24
Farm Level NDJFM Precipitation By Analog Years
NDJFM Precipitation (mm)
25
Specific Conclusions
Profit margins can influence forecast value
effects Improved regional forecast skill may not
lead to increased tactical forecast value at the
farm level Value of best no-forecast practices
See Mauget, S.A., Zhang, J. and Ko, J., 2009
The value of ENSO forecast information to dual
purpose winter wheat production in the U.S.
Southern High Plains. Journal of Applied
Meteorology and Climatology, In Press.
26
General Conclusions
Profit Effects of Forecast Information are
Probabilistic
Forecast information may not Pay Off every
year.
27
(More) General Conclusions
Similar analyses could be done in any area
sensitive to climate-related risk But while
seasonal forecasts may re-define climate related
risk they will never eliminate it To ease
adoption, provide a probabilistic track record
of how forecast information re-defines that
risk.
28
Conclusion (cont.)
Mauget, S.A., Zhang, J. and Ko, J., 2009 The
value of ENSO forecast information to dual
purpose winter wheat production in the U.S.
Southern High Plains. Journal of Applied
Meteorology and Climatology, In Press.