IRI Seasonal Forecasting Update

1
IRI Seasonal Forecasting Update
2
Models Run at IRI 2-Tier

• ECHAM4.5 T42L19
• GHG Forcing will be added
• New SST scenario strategy
• ECHAM5 T42L19
• GHG Forcing will be added
• CCM3 T42L19
• CAM3/4? T42L19
• GHG Forcing will be added

3
Models Run at IRI 1-Tier

• ECHAM-MOM3 (Real-Time in next few months)
• OGCM 1.5 X 0.5 with 25 vertical layers
• GFDL ODA
• Temperature only
• Constant background error covariance
• Ensemble size 12
• Retrospective forecasts from 1982
• ECHAM-MOM4 (Development to start late spring)
• OGCM 1 X 0.33 with 40 to 50 vertical layers
• NCEP GODAS ODA (kindly provided by Dave
  Behringer)
• Temperature and salinity assimilation
• State dependant background error covariance
• Ensemble size 12
• Retrospective forecasts from 1982

4
IRI 1-Tier Multi-Model Ensemble

• Initially the current IRI 2-Tier MME will not
  include 1-Tier models
• A separate 1-Tier MME will be made
• Length of retrospective forecasts is shorter
  than 2-tier
• (1982 start versus 1957 start)
• Possible that 2-Tier and 1-Tier MME will merge
  into a single product in future

5
MULTI-MODEL PROBABILISTIC FORECASTS
Current Method - Performance-based weighting of
models, including climatology as a model -
Historical performance from AMIP-type runs -
Produces 3-Category forecasts (i.e.
Terciles) New Method - Models recalibrated
individually before combination ? Spatial bias
correction ? Local bias correction - Historical
performance from HINDCASTS (AGCMs forced with
predicted SSTs - Produces full probability
distribution
6
1. Model Calibration Spatial Bias Correction
CCA performed regionally. Results are smoothed
along overlapping areas.
7
3. PDF Flexible format of information
ECHAM4.5 2m Temperature JFM 1983 El Nino
Forecasts for the full PDF allows users to
produce probabilistic forecasts for any
category or threshold of interest.
X
8
3. PDF Flexible format of information
Probability Distribution Function (relative to
climatological PDF)
Could add user-defined categoryor threshold
boundaries to illustrate probability of those.
Cumulative Probability Distribution
Probability of Exceedance
9
statistical downscaling seasonal rainfall
statisticsIndian monsoon rainfall
seasonal total
rainfall frequency
JJAS rainfall correlation skill ECHAM4-CA made
from June 1
10
prediction skill of SW monsoon onset over
Philippines
ECHAM-CA
SST
CFS
ECHAM-MOM
11
International Research Institute for Climate and
Society Research in support of climate risk
management
Experts in the use of remotely sensed data to
establish regional climate patterns where direct
observations are missing
Leaders in the development and assessment of
forecast products.
10
graph courtesy of U. Redding
Innovators in the sectoral analysis of climate
impacts (e.g., malaria early warning tool)
Basic research to unravel and understand climate
mechanisms
12
IRI Examples of Climate Risk Management
Research and Practice

• Climate variability and agriculture in Southeast
  South America
• Improved understanding and predictability of
  climate impacts on the sector
• Collaboration with national agriculture research
  institutes in the southern cone

• Weather indexed insurance for farmers in Malawi,
  Tanzania, Ethiopia
• Improved use of agroclimatological information
  to design insurance contracts
• Advances in use of remote sensed data
  climatology to fill data voids
• Work with local farmers collectives, financial
  institutions, World Bank, Oxfam, Swiss Re

• Desert Locust Early Warning Systems
• Training of national control authorities
• Product Integration in UN Food and Agriculture
  Organizations early response system

13
IRI Examples of Climate Risk Management
Research and Practice

• Reservoir Management Tools
• Improvements in hydroelectric capacity with
  tailored climate information
• Innovative financial instruments to off-set
  impacts of water shortages
• Collaboration with reservoir managers in the
  Philippines and Chile

• Training of Sectoral and Climate Specialists
• On-going collaboration with WHO, WMO, Red Cross,
  national ministries, NGOs and research partners
  to bridge gaps between climate knowledge and
  practice

• Climate Research for Greater Social Utility
• Development and testing of forecasts and other
  products tailored to the needs of users

14
IRI and Google.org Foundation

• Three-year invited project from Google/Moore
  (March 08)
• Geographical Focus Greater Horn of Africa
• Focal country case study Ethiopia
• Climate side increase capacity to supply useful
  information products to help guide health
  interventions
• Health side increase capacity to demand and use
  climate products for more effective interventions

15
IRI and Google.org/Moore Foundation
Draws on IRIs Climate Program Environmental
Monitoring Program Data Library/Map Rooms Health
specialists Economists Educations and
Trainers Project Management
Some partners ICPAC WHO Reading
University National met agencies
16
IRI and Google.org/Moore Foundation
Building communities of practice
Ethiopia CHWG Sep 08
Ethiopia CHWG/MERIT Dec 08
Madagascar CHWG Oct 08
Kenya CHWG Dec 08
17
IRI and International Federation of Red Cross/Red
Crescent

• Goal is to use advanced climate information to
  improve disaster preparedness and response
• Provide a global six-day forecast tool for IFRC
• Form Partnerships with RC/RC national societies

18
IRI and International Federation of Red Cross/Red
Crescent
19
IRI IFRCPotential for Assessing Disaster Risks
at Regional/National Scale
Example Landslides in the Philippines
Land Cover
Recent Rainfall
Slopes, Soils
Land Cover
Exposed Pop.
Slopes, Soils
Exposed Pop.
Typhoon Fengshen, June 08