Jean L. Steiner Jurgen D. Garbrecht Jeanne M. Schneider X. C. (John) Zhang M. W. Van Liew
description
Transcript of Jean L. Steiner Jurgen D. Garbrecht Jeanne M. Schneider X. C. (John) Zhang M. W. Van Liew
Integrating Climate Variability and Forecasts into Risk-Based Management Tools for Agricultural Production and
Resource Conservation
Jean L. SteinerJurgen D. GarbrechtJeanne M. SchneiderX. C. (John) Zhang
M. W. Van Liew
USDA-ARS Grazinglands Research Laboratory
Great Plains Agroclimate and Natural Resources Unit
El Reno, OK
Objectives
• Regional context of Southern Great Plains
• research focus• Methods • Assessing decision maker needs• Relevance to GECAFS
El Reno, OK
Research Focus
• Risk-based decision making• Climate variability as a primary risk
factor– Decadal scale cycles– Seasonal forecasts
• Levels of analysis– Regional, watershed– Farm-scale
Methods and Preliminary Analyses
0
20
40
60
80
100
120
140
160 M
ean
Pre
cipi
tatio
n, m
m
0
5
10
15
20
25
30
Mea
n Te
mpe
ratu
re, C
J F M A M J J A S O N DMonth
Precipitation Temperature
El Reno, Oklahoma – 1971 to 2000
Year
Prec
ipit
atio
n [i
n]
Dry PeriodsWet Periods
Annual Precipitation
5-yr weighted average CD3405; 1895-2003152025303540455055
1895 1915 1935 1955 1975 1995
Annual Precipitation in Central Oklahoma
USDA-ARS-GRL
Calendar Year
Annu
al S
trea
mflo
w [
cfs]
Annual Precipitation [in]Blue River, Oklahoma
Blue River Streamflow and PrecipitationPrecipitationStreamflow
5-yr weighted average
Average for1937-2003
R2 = 0.84USGS 07332500
USDA-ARS-GRL
Blue River StreamflowPr
obab
ility
of E
xcee
danc
e
1981-2002
1947-1980
Streamflow [cfs]USDA-ARS-GRL
USDA-ARS-GRL
USDA-ARS-GRL
USDA-ARS-GRL
CPC precipitation forecasts product
USDA-ARS-GRL
Dependability of Wet Forecasts, |DN| ≥ 10% Lead Time 0.5 months, 58 forecasts from JFM 1997 through OND 2001
< 50% 50-99% 100%
1/12/2
2/2
1/12/2
1/12/2
2/22/23/31/1
2/23/3
4/44/5
2/2
4/44/4
4/53/4
4/53/3
4/4
4/4
3/33/3
4/4
3/4 1/1
2/2
2/23/3
2/3
2/2
4/55/7
4/65/66/7
6/6
5/75/7
4/6
4/6
5/7
5/76/7
5/7
4/7
1/21/2
1/2
1/2
2/4
4/8
Dependability of Dry Forecasts, |DN| ≥ 10% Lead Time 0.5 months, 58 forecasts from JFM 1997 through OND 2001
< 50% 50-99% 100%
5/5
3/31/1
1/1
2/2
6/8
5/8 9/13
6/6
10/12
10/11
2/22/2
1/1
1/12/2
2/3
2/2
1/11/1
3/4
7/8
10/14
12/18
10/14
17/19
9/14
12/16
1/2
1/2 2/3
1/1
1/2
1/1
1/21/2
3/6
2/3
First: Downscale Forecasts toFarm and Monthly Scales
Second: Use Weather Generators to Produce Sequences of Daily
Weather
Third: Use Models to Produce Forecast Shifts in Odds for an
Application
Fourth: Incorporate Climate Information
in Decision Support Tools
location normal
location forecast =
location normal ++ forecast
anomalies
divisionforecast
locationforecast
divisionforecast
locationnormal
divisionnormal
Very WetVery Dry PRECIPITATION
PROB
ABIL
ITY
OF E
XCEE
DANC
E
forecast anomalies = division forecast - division normal
Spatial Downscaling of Forecasts
SEP
NO
V
JAN
MA
R
MAY JU
L
SEP
NO
V
JAN
Full cycle of 13 3-month forecasts
Desired set of 151-month forecasts
Simulated grain yield, kg/m2
0.1 0.2 0.3 0.4
Cum
ulat
ive
prob
abili
ty
0.0
0.2
0.4
0.6
0.8
1.0
Dry-40%Dry-70%Avg-40%Avg-70%Wet-40%Wet-70%
Month
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Mon
thly
mea
n pr
ecip
itatio
n, m
m
0
20
40
60
80
100
120
140
160
180
200Wet-NWSWet-CLIGENAvg-NWSAvg-CLIGENDry-NWSDry-CLIGEN
Evaluating a climate generator (CLIGEN) for daily precipitation…
… and wheat growth model sensitivity to precipitation terciles and initial soil water condition
50%
0%
100%
forecast
Very Wet
Very DryPRECIPITATION
PRO
BABI
LITY
OF
EXCE
EDAN
CE
normal
100%
Currentlyunknown…
forecastnormal
PRO
BABI
LITY
OF
EXCE
EDAN
CE
0%
50%
100%
Very Low Very High3-MONTH PRECIPITATION
50%
0%
forecast
yield
Very High
Very LowFORAGE YIELD
PRO
BABI
LITY
OF
EXCE
EDAN
CE
normalyield
100%
What is the relationshipbetween a sequence offorecasts and outcome?
50%
0%
forecast
yield
Very HighVery LowFORAGE YIELD
PRO
BABI
LITY
OF
EXCE
EDAN
CE
normalyield
100%Associate baseline and forecast odds for outcomes with economic factors to define “risks”.
Models Used
• Regional, watershed– SWAT– Neural Networks
• Farm/field Level– WEPP– CERES– Enterprise budgets, market tools
Identifying Decision Maker Needs
Workshops to present findings and engage in dialog
One-on-one discussions of specific issues
Exploratory work in form of “case studies”
Decision Making Case Study
Cropping/Grazing
Systems in Southern Great Plains
Month Agricultural Management Calendars for S. Great Plains Decision Points
Wheat SummerPerennial
WinterPerennial
SummerAnnual
StockerCattle
January graze
February graze grow to grain?spring fertilizer?
March bale wheat in May?
April
May
June harvest summer crop?
July contract for cattle?#, delivery date, $
August sell
September
October sow area to plant, which fields first,variety, seeding rate, fertilizeramount
November delivery
December graze
Decision Points: Wheat Grazing Systems
foragequality dip
graze
sow
graze
buy additional cattle?
sell cattle?
supplemental feed?
Agronomic Decisions• Crop selection
– e.g., maize/sorghum/millet– Long vs short season varieties
• Planting density and geometry• Fertility levels, dates, rates…• Area to be planted
Crop/livestock system Decisions
• Future stocking rates• Forage (grazed or hayed) vs
grain harvest• Intensity and timing of grazing• Supplemental feed• Purchase, selling, or movement
of animals
Business Decisions
• Marketing/hedging• Diversification of farm
enterprises• Off-farm income
Decision Maker Needs
• Work with individual farmers, extension, conservationists
• Identify their goals and priorities• Identify their resources and
characterize their systems• Develop climate scenarios relevant
to key decisions
Decision Maker Needs
• Focus on record keeping is essential
• A “journaling” tool will be used to analyze decision points, factors considered in taking decisions, building decision trees or decision rules
Regional Case Study
Water Release
from Reservoirs
Decision Maker Needs
• Work with agencies with management responsibilities (e.g., U.S. Bureau of Reclamation, U. S. Corps of Engineers)
• Understand stakeholders and issues• Analyze decision criteria and decision
trees specific to their situation• Incorporate climate variability and
climate forecast scenarios
Risks in Farming• Risk is an important aspect of the
farming business. The uncertainties of weather, yields, prices, government policies, global markets, and other factors can cause wide swings in farm income.
• Risk management involves choosing among alternatives that reduce the financial effects of such uncertainties.
http://www.ers.usda.gov/Briefing/RiskManagement/
Types of Risks• Production risk derives from the uncertain natural growth processes
of crops and livestock. Weather, disease, pests, and other factors affect both the quantity and quality of commodities produced.
• Price or market risk refers to uncertainty about the prices producers will receive for commodities or the prices they must pay for inputs.
• Financial risk results when the farm business borrows money and creates an obligation to repay debt. Rising interest rates, the prospect of loans being called by lenders, and restricted credit availability are also aspects of financial risk.
• Institutional risk results from uncertainties surrounding government actions. Tax laws, regulations for chemical use, rules for animal waste disposal, and the level of price or income support payments are examples of government decisions that can have a major impact on the farm business.
• Human or personal risk refers to factors such as problems with human health or personal relationships that can affect the farm business. Accidents, illness, death, and divorce are examples of personal crises that can threaten a farm business.
http://www.ers.usda.gov/Briefing/RiskManagement/
Relevance to GECAFS DSS
• Decision making is individualized process and may be approached as case study
• Decision makers have multiple objectives, some economic and some not, which must be balanced
USDA-ARS-GRL
Recognizing andAdapting to Change