Decision Systems Research developing decision tools
IRI Underpinning Activity
Climate Prediction Applications WorkshopFlorida State University
9-11 March 2004
N. WardAnd
C. F. Ropelewski
Centrality of exploring and influencing sectoral decisions,based on information that draws on sound biophysical science
IRI’s mission is to enhance society's capability to understand, anticipate and manage the impacts of seasonal climate fluctuations, in order to improve human welfare and the environment, especially in
developing countries.
The Discussion
•Scope of the Underpinning Activity
•Illustrations of its presence in Regional Project Settings- NE Brazil and Philippines Water management- Farm level agriculture - Support at regional level for agriculture issue
•Role of training / capacity building
Components of the Work
•Development of Decision Strategies and Tools (DST)
•Methodologies to extract relevant environmental information to feed DST
•Testing of DST based on forecasts/information over past years
•Experimental implementationDeveloping decision support informationEvaluating and learning through implementation
Simulating the Expected Improvements in Reservoir Management
Example for Reservoir in Ceara, NE Brazil
(collaboration led by Assis de Souza Filho, FUNCEMEand Upmanu Lall, IRI
additional contributions, especially Sankar Arumugam, IRI)
Reliability – Yield Curve
Reliability-Yield Curve
0
200
400
600
800
1000
1200
20 30 40 50 60 70 80 90 100
Reliability
Yie
ld (
hm
3)
Forecast 1989Forecast 1993Climatology
-40.0
-35.0
-30.0
-25.0
-20.0
-15.0
-10.0
-5.0
0.0
5.0
1950 1960 1970 1980 1990
Year
Sp
ill/
Sto
rag
e M
ax (
in %
)
0
1000
2000
3000
4000
5000
6000
7000
Ob
serv
ed F
low
s (m
3 /s)
Forecast - Climatology (Reliability = 0.9)
Observed Flows
Spill (Reliability = 0.9)Spill (Reliability = 0.9)
Simulating the benefits of using climate forecast informationin the operation of a reservoir in Ceara over 1950-2000
SEASONAL WATER ALLOCATION AND RESERVOIR OPERATION UTILIZING CLIMATE
INFORMATION BASED STREAMFLOW FORECASTS
A aerial view of the Angat Hydroelectric Plant
Courtesy of Mr. Rodolfo German (Angat dam)
Hydroclimatology
0
50
100
150
200
250
300
350
JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC
Month
Str
ea
mfl
ow
(in
hm
3)
0
50
100
150
200
250
300
350
400
450
500
Ra
infa
ll (
mm
)
StreamflowRainfall
3-months lag correlation
(Nino3.4,QJJAS) = -0.20
(Nino3.4,QOND) = -0.51
JJAS – 30%
OND – 46%
Reservoir Operation Strategy
ANGAT H.E. PLANT
150
160
170
180
190
200
210
220
JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC
EL
EVA
TIO
N (
m)
1996 1997 1998 1999 2000 2001 UPPER LOWER
StatisticalDownscalingResults forSri Lanka, 1951-80 Verification
Map shows correlationskill (shading) alongwith contours of elevation
Need for cautionin regions of complex terrain
0
1
2
3
4
5
Gra
in yie
ld, M
g/h
a
1970 1975 1980 1985 1990 1995Year
observed hindcast
0
20
40
60
Pre
cip
ita
tio
n,
cm
a
b
r = 0.50
r = 0.62
Statistical Downscaling from General Circulation Model (GCM) output to(a) Predict Oct-Dec seasonal rainfall total(b) Predict Crop yield
daily weather generator conditioned on GCM predicted wind field –
resulting daily weather sequences used to drive a crop model
Example for a Site in Kenya
Examples of Decisions Represented in Farm-level Models
Field-scale crop management decisions:
Cultivar selectionPlanting datePlanting densityAmount and timing of nitrogen fertilizer applicationLivestock stocking rates
Farm-scale management decisions:
Land allocation among cropsFeed management (pasture planting and fodder purchase)Borrowing for production inputs (planned)Allocation of household labor among farm vs. non-farm activities (planned)
Objective 2: Tailored ProductsProblem area User(s) Requirement Application
1) Areas of high Rift Valley Fever (RVF) outbreak risk
Red Sea Livestock Trade Commission
Predict RVF risk areas 3-6 months in advance
Identify and treat RVF outbreaks before regional trade barriers are imposed
2) Livestock fodder availability
Pastoral communities in northern Kenya and southern Ethiopia
Narrow the confidence interval of the current Livestock Early Warning System (LEWS) 90 day fodder outlook using a seasonal forecast
Provide improved 90 day early warning to nomadic pastoralists and sedentary agro-pastoralists of expected fodder conditions
3) Livestock fodder availability
Organization of African Union/Inter-African Bureau of Animal Resources (OAU/IBAR)
Provide 3-6 month early warning of major regional fodder shortages
Support IBAR livestock purchase programs
4) Pastoralist livelihood system stress
USAID, other donors and international emergency assistance organizations
Simulate climate shock impacts on pastoralist livelihood systems and food security
Contingency and operational assistance planning
Greater Horn of Africa Project
0
0.5
1
1.5
81 83 85 87 89 91 93 95 97
Year
ind
ex predicted
observed
Using Global Climate Model Output to Predict Vegetation
The index is the satellite estimated Normalized Difference Vegetation Index (NDVI)Index shown is the October-December average for Northeastern Kenya
Positive values indicate enhanced greenness in vegetation
Correlation between predicted and observed = 0.84
Using Global Climate Model Output to Predict NDVI on a 1 degree lat x 1 degree lon grid across Kenya
Skill (indicated by shading) is very good in most parts of the domain,less to the NE of Lake Victoria in region of complex orography
(to be further investigated)
Shading Indicates Correlation between predicted and observed NDVI time-series over 1981-98
Contours are elevationon 1 lat x 1 lon grid
Example of pageFrom online course
Collaborationwith CCNMTL
IRI is exploring the potential of online courses
http://iri.columbia.edu/outreach
Advanced Training Institute on Climatic Variability and Food Security
Palisades, New York, USA
8 - 26 July 2002
Dr. James Hansen, Training Institute Director
The Advanced Training Institute on Security is designed to equip young developing country professionals with expertise in agriculture and food security to apply advances in climate prediction to their home institutions' ongoing efforts to address climate-sensitive aspects of agricultural production, food insecurity and rural poverty
•Examples of the underpinning activity within emergingend-to-end regional projects
• Methodological advances to contribute to fasterimplementation in other regions
00
10001000
20002000
30003000
40004000
50005000
60006000
70007000
2929 55 1212 1919 2626 22 99 1616 2323 22 99 1616 2323 3030 66 1313 2020 2727 44 1111 1818 2525
Date
Cas
es
JanuaryDec. February March April May
The forecasting principle Early warning for changes in the incidence of the
meningococcal meningitis disease(Example for West African Sahel region)
not preventable
Vaccination starts
Early_______________ Peak_________________Late______________
+Dust
-Rain______-Rain______
Feb-Apr Precipitation and Near-surface wind Predicted by a Regional Climate Model at 10km Resolution
across Taiwan and surrounding ocean
For case study years 1983 (generally WET) MINUS 1971 (generally DRY).
Rainfall difference in mm per dayas predictedby the model for 1983relative to 1971
Illustrates theNeed for cautionIn regions of Complex terrain