Translating Climate Forecasts into Agricultural Terms:

Advances and Challenges

James Hansen, Andrew Challinor, Amor Ines, Tim Wheeler, Vincent Moron

presented at theInternational Workshop on Climate

Prediction and Agriculture – Advances and Challenges

WMO, Geneva, 11 May 2005

Motivation

• Information relevant to decisions

• Ex-ante assessment for credibility and targeting

• Fostering and guiding management

Overview• Six years ago

– Dominance of historic analogs

– Doubts about crop predictability

• Recent advances– The challenge, and potential approaches

– Synthetic weather conditioned on climate forecasts

– Use of daily climate model output

– Statistical prediction of crop simulations

– Downscaling and upscaling

• Opportunities and challenges– Embedding crop models within climate models

– Enhanced use of remote sensing, spatial data bases

– Robustness of alternative coupling approaches

– Forecast assessment and uncertainty

– Climate research questions

Six Years Ago:Dominance of Historic Analogs

• Advantages– Intuitive probabilistic interpretation– Accounts for any differences in “signal strength”– May incorporate useful higher-order statistics

• Concerns– Small sample size,

confidence, artificial skill– Are differences in

distribution real?– How to use with dynamic

prediction systems without discarding information?

0%

10%

20%

30%

40%

50%

3 6 12 16

Number of "phases"

Var

ian

ce e

xpla

ined

cross-validated no cross-validation

Six Years Ago:Doubts About Crop Predictability

• Spatial variability of rainfall limits predic-tability at farm scale

• Accumulation of error from SSTs, to local climatic means, to crop response

• Impact of wrong fore-cast on farmers’ risk Barrett, 1998. Am. J. Agric.

Econ. 80:1109-1112

The Challenge

• Nonlinearities. Crop response to environment can be nonlinear, non-monotonic.

• Dynamics. Crops respond not to mean conditions but to dynamic interactions:– Soil water balance

– Phenology

• The scale mismatch problem.

0

1

2

3

4

5

Gra

in y

ield

(M

g/h

a)

0 200 400 600 800OND precipitation (mm)

The Scale Mismatch Problem

• Crop models:– Homogeneous plot spatial scale– Daily time step (w.r.t. weather)

• GCMs:– Spatial scale 10,000-100,000 km2

– Sub-daily time step, BUT... Output meaningful only at (sub)seasonal scale

– Tend to over-predict rainfall frequency, under-predict mean intensity

• Temporal scale problem more difficult than spatial scale.

Effect of Spatial Averaging

Inverse-distance interpolation of daily weather data, north Florida, at a scale comparable to a GCM grid cell.

Hansen & Jones, 2000. Agric. Syst. 65:43-72.

Effect of Spatial Averaging

• Spatial averaging distorts variability, increases frequency, decreases mean intensity.

• Similar spatial averaging occurs within GCM.

0

50

100

150

Rain

fall t

ota

l (m

m)

Jan MarMay Jul SepNov Month

interpolatedobserved

0.0

0.2

0.4

0.6

0.8

Rela

tive f

req

uen

cy

Jan MarMay Jul SepNovMonth

0

4

8

12

Inte

nsit

y (

mm

/d)

JanMarMay Jul SepNovMonth

Effect of Spatial Averaging

Simulated maize yields, CERES-Maize

Information Pathways

predicted crop yields

observed climate predictors

?

Information Pathways

crop model

analogyears

predicted crop yields

observed climate predictors

categorize

Information Pathways

downscaleddynamicmodel

stochasticgenerator

crop model(hindcast weather)

analogyears

predicted crop yields

statisticalclimatemodel

observed climate predictors

categorize

Information Pathways

downscaleddynamicmodel

stochasticgenerator

crop model(observedweather)

crop model(hindcast weather)

analogyears

predicted crop yields

statisticalclimatemodel

statisticalyield model

observed climate predictors

categorize

Approaches

• Classification and selection of historic analogs (e.g., ENSO phases)

• Synthetic daily weather conditioned on forecast: stochastic disaggregation

• Statistical function of simulated response

– Nonlinear regression

– Linear regression with transformation or GLM

– Probability-weighted historic analogs

• (Corrected) daily climate model output

Advances:Synthetic Weather Inputs

Two Approaches:

• Adjusting generator input parameters:– Flexibility to produce statistics of interest

– Assumed role of intensity vs. frequency

• Constraining generator outputs:– No assumptions re. frequency vs. intensity

Option 2. Constraining generated output

First step:First step:-- Iterative procedure – Using climatological parameters, accept the first realization with Rm near target:

|1-Rm/Rm,S|j <= 5%

Second step: Second step: - - Apply multiplicative rescaling to exactly match target monthly target.

Hansen & Ines, Submitted. Agric. For. Meteorol.

Constraining generator outputs reproduces correlations better than adjusting inputs.

  Tifton, Georgia 

Gainesville, Florida

Scenario RM vs. π RM vs. μI μI vs. π RM vs. πRM vs. μI μI vs. π

Observed daily rainfall 0.649 0.577 -0.165

 0.668 0.706 0.046

Disaggregated monthly rainfall

constrain RM 0.681 0.676 -0.004 

0.649 0.697 0.014

condition π 0.822 0.473 0.013 

0.831 0.121 0.052condition μI 0.491 0.856 0.071

 0.458 0.837 0.052

Constraining generator output requires fewer replicates for given accuracy.

0.0

0.5

1.0

1.5

2.0

2.5

0.0

0.2

0.4

0.6

0.8

1.0

1 10 100 1000

0.0

0.2

0.4

0.6

0.8

1.0

1 10 100 1000

0.0

0.5

1.0

1.5

2.0

2.5

RM

SE

, M

g ha

-1

0.0

0.2

0.4

0.6

0.8

1.0

1 10 100 1000

R

Gainesville Katumani

constrain RM

condition μI

condition π

d

b

f

a

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

c

e

Tifton

Number of realizations

1 0 0 0

1 2 0 0

1 4 0 0

1 6 0 0

1 8 0 0

2 0 0 0

1 1 0 1 0 0 1 0 0 0

N o . o f r e a l i z a t i o n s

RM

SE

, kg

ha-1

R m

p i

0 . 1

0 . 2

0 . 3

0 . 4

0 . 5

0 . 6

1 1 0 1 0 0 1 0 0 0

N o . o f r e a l i z a t i o n s

R

R m

p i

0

1000

2000

3000

4000

5000

6000

7000

1960 1965 1970 1975 1980 1985 1990 1995 2000 2005

Year

Yie

ld,

kg h

a-1

Simulated Hindcasts

R=0.44

0

1000

2000

3000

4000

5000

6000

7000

1960 1965 1970 1975 1980 1985 1990 1995 2000 2005

Year

Yie

ld,

kg h

a-1

Simulated Hindcasts

R=0.41

Rm

π

Maize simulated from disaggregated monthly GCM hindcasts, Katumani, Kenya

Advances:Use of Daily Climate Model Output

Options

• Calibrate simulated yieldsChallinor et al., 2005. Tellus 57A:198-512

• Correct GCM mean bias– Additive shift for temperatures– Multiplicative shift for rainfall

• Rainfall frequency-intensity correctionInes & Hansen, In preparation

1

F(xGCM=0.0)

F(xhist=0.0)

00

GCMHistorical

Correcting Bias in Daily GCM Output: Rainfall Frequency

calibratedthreshold

Correcting Bias in Daily GCM Output: Rainfall Intensity

GCMHistorical

0

Daily rainfall (x), mm

1

00

F(x

)

x'i

F(xi)

xi

1, ,( ( ))i obs m GCM m ix F F x

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1970 1975 1980 1985 1990 1995

Mean

rai

nfa

ll fre

quency

, w

d d

-10

2

4

6

8

10

12

14

1970 1975 1980 1985 1990 1995M

ean

rain

fall inte

nsity, m

m w

d-1

0

1

2

3

4

5

6

7

8

1970 1975 1980 1985 1990 1995

Year

Mea

n r

ainfa

ll a

mount, m

m d

-1

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1970 1975 1980 1985 1990 1995

Obs

EG

GG

Uncorr

Rm

μ

π

• Katumani, Kenya

• ECHAM4 & observed OND daily rainfall (1970-95)

• Intensity corrections:

• EG: empirical (GCM) to gamma (observed)

• GG: gamma (GCM and observed)

Corrects rainfall total, frequency, intensity.

Predicts yields from GCM, perhaps better than stochastic disaggregation

0

1

2

3

4

5

6

7

8

1970 1975 1980 1985 1990 1995

Year

Mean

mo

nth

ly r

ain

fall

(R m

), m

m d

-1

rMOS=0.59

rGG =0.74

0

1

2

3

4

5

6

7

8

1970 1975 1980 1985 1990 1995

Obs

MOS

GG

• CERES-Maize simulated with:• Disaggregated MOS-corrected monthly hindcasts• Gamma-gamma transformation of daily rainfall

Advances: Statistical Prediction of Crop Simulations

• Seasonal predictors of local climate potential predictors of crop response

• Predictand: Yields simulated with observed weather

• Eliminates need for daily weather conditioned on climate forecast

• Poor statistical behavior

Nonlinear Regression

Katumani maize prediction example:

• Yields as f(PC1)

• Mitscherlitch functional form:

• Cross-validation

ex py a b c x 1y=3.33+1.34(1-exp(-0.133x))R2 = 0.400

K Nearest Neighbor

• Unequally-weighted analogs

• Weights w:

– Based on rank distance (predictor state space)

– Interpreted as probabilities

• Forecast ŷ a weighted mean:

• Optimize k

• A non-parametric regression

1

1/

1/j k

i

jw

i

1,

ˆn

t i ii i t

y w y

Linear Regression & Transformation:Regional-Scale Wheat, Qld, Australia

• Wheat simulations: water satisfaction index

• ECHAM4.5, persisted SSTs, optimized (MOS)

• Yield prediction by c-v linear regression

• Box-Cox normalizing transformation

• Forecast distribution:

– Regression residuals in transformed space

– n antecedent X n within-season weather years

Hansen et al., 2004. Agric. For. Meteorol. 127:77-92

Linear Regression & Transformation:Regional-Scale Wheat, Qld, Australia

N

200 0 200 400 km

1 July

1 June

1 August

1 MayCorrelation

<0.34 (n.s.)0.34-0.450.45-0.550.55-0.650.65-0.750.75-0.85 > 0.85

Linear Regression & Transformation:Regional-Scale Wheat, Qld, Australia

0.5

1.0

1.5

2.0

0.5

1.0

1.5

2.0

0.5

1.0

1.5

2.0

Gra

in y

ield

(M

g ha

-1)

Climatology

Date of forecast

ENSO phase

198

9 (n

eutr

al)

GCM-based

198

2 (E

l N

iño

)19

88

(La

Niñ

a)

1Jun 1Jul 1Aug Harvest1May 1Jun 1Jul 1Aug Harvest1May1Jun 1Jul 1Aug Harvest1May

Observed90th percentile75th percentile50th percentile25th percentile10th percentile

Advances: Downscaling &

Upscaling• Spatial climate downscaling:

– Methods advancing

– Uncertain impact on skill

• Crop model upscaling:

– Understanding and methods for aggregating point models

– Increasing set of reduced form large-area models

Predictability (r) of groundnut yields with large area model, W India. Challinor et al., 2005. Tellus 57A:198-512

0.0

0.2

0.4

0.6

0.8

Co

rre

latio

n

point ~1°×1°~2°×1° stateScale

Jan-MarApr-Jun

Obs. vs. pred. rainfall, Ceará, NE Brazil, as function of aggregation. Gong et al., 2003. J. Climate 16:3059-71.

Opportunities & Challenges:Crop Models Within Climate Models

• Run crop models within GCM or RCMs

• Allow crop to influence atmosphere– Alternative land surface scheme– Intended benefit is atmosphere response to

crop

• Likely to require calibration of crop results for foreseeable future

• Match scale of climate model grid

Opportunities & Challenges:Remote Sensing, Spatial Data Bases

• Enhanced georeferenced soil, land use, cultivar data bases

• Assimilation of real-time, contiguous antecedent weather into forecasts

• Estimation of cropped areas, dates

• Correction of simulated state variables

• Eventual farm-specific crop forecasts?

1960 1970 1980 1990

observedpredicted

0

1

2

3

4

5

0

1

2

3

4

5

0

1

2

3

4

5

1960 1970 1980 1990

Stochastic disaggregation + k nearest neighbors, 1 PC

Regression

k nearest neighbors, 2 PCs

Stochastic disaggregationr = 0.57

r = 0.53

r = 0.53

r = 0.58

r = 0.55

Opportunities & ChallengesRobustness of Alternative Approaches?

Hansen & Indeje, 2004. Agric. For. Meterol. 125:143

Opportunities & Challenges: Forecast Assessment and Uncertainty

• Does predictability (climate and impacts) change from year to year?– Artifact of skewness?– Real impacts of climate state?– Captured by GCM ensembles?

• Interpretation of forecasts based on categorical vs. continuous predictors?

• Consistency of hindcast error vs. GCM ensemble distributions?

Are differences in dispersion real?

ENSO phase

Dec

embe

r rai

nfal

l

La Nina neutral El Nino

ENSO phase

La Nina neutral El Nino

Raw Transformedskewness 1.243 -0.032p ENSO influence on: means 0.0001 *** 0.0004 *** dispersion 0.0001 *** 0.91 n.s.

Junin, Argentina, 1934-2001

Opportunities & Challenges: Forecast Assessment and Uncertainty

• Does predictability (climate and impacts) change from year to year?– Artifact of skewness?– Real impacts of climate state?– Captured by GCM ensembles?

• Interpretation of forecasts based on categorical vs. continuous predictors?

• Consistency of hindcast error vs. GCM ensemble distributions?

Opportunities & Challenges:Climate Research Questions

• Past prediction efforts driven by skill– Relative shifts– Large areas– 3-month climatic means

• Stimulating interest in “weather within climate”– Skill at sub-seasonal time scales– Higher-order rainfall statistics– Shifts in timing, onset, cessation– Methods to translate into weather realizations