PREDICTION AND EARLY WARNING DIVISION

Skilful climate prediction and timely early warning forms the foundation of any operational climate centre. ICPAC provides 10 days, monthly and seasonal climate prediction products that are required by IGAD, the NMHSs, and other national, regional and international partners. The Division uses two empirical techniques to provide climate outlooks namely dynamical and statistical approaches. Some of the climate modeling methods requires enormous computing power and skilled human capacity. Improvement of ICPAC and national computing capacity, human resources and development of new prediction tools are the main priority of the Division. The other priority is to develop products required for specific sector applications.

Statistical climate prediction methods

The statistical prediction methods are based on empirical relationships between rainfall over specific parts of GHA and some global / regional/ local climate system indices. The major climate system indices that are used at ICPAC include evolution of monsoons, medium and upper level winds, Madden-Julian Oscillation (MJO), Quasi biennial oscillation (QBO), El Niño / Southern oscillation, Indian Ocean dipole IOD, tropical cyclones, sea surface temperature gradients among many others that have been derived from general circulation. Most of the statistical methods still do not explain most percentages of rainfall variances. A lot of efforts must therefore be made to improve the skills of the prediction models. At the moment ICPAC is collaborating with a number of partners such as IRI, NCEP/CPC, UKMO, FSU, NCMRWF among others to address climate prediction and early warning challenges in the region. The activities of the unit during 2005-2010 period include:

Documentation of the methods and associated applications
Develop improved prediction techniques for the ten-day outlook.
Provide forecasts on temperatures and rainfall onset, cessation and distribution of dry/wet spells.
Continuous verification of the prediction products to assess the skill of the forecasts disseminated to the users.
Basic diagnostic research to improve the prediction products for various time scales
Develop pilot projects to assess and communicate successful use of forecasts.
Improvement of statistical techniques on the downscaling of prediction products to meet user requirements.
Ensure user feedback is incorporated into forecast developments.
Continue to improve the capacity of the member states on statistical and other new and emerging climate prediction technologies.
Enhance collaboration with other relevant regional and international centres engaged in climate prediction and application services

Dynamical climate prediction approaches

The climate modeling and numerical weather prediction methods used at ICPAC provide the evolutionary spatial and temporal evolutionary dynamics of weather and climate evolutions over the GHA region.

Currently, ICPAC is running the Regional Spectral Model (RSM) in a dynamical downscaling mode using a mini-super computer, an IBM RS6000. The RSM is nested in the low-resolution global model ECHAM 4.5 run operationally at the International Research Institute for Climate Prediction (IRI). The IRI global model provides the initial and lateral boundary conditions for the regional model, and the global model outputs are provided by the IRI. Statistical downscaling of the global model (ECHAM4.5) is also being done at ICPAC through a package that has been coded at IRI that uses Model Output Statistics (MOS).

The ultimate objective of the section is to use an array of global climate models (GCMs), Regional Climate Models (RCMs), Numerical Weather Prediction Models (NWP), and other modeling technologies including downscaling and Super Ensemble techniques. The activities scheduled for the 2005-2010 period include:

Enhancing of NWP at ICPAC and NMHS.

Adoption of operational FSU super ensemble NWP products for the GHA at ICPAC and NMHS.
Adoption of 24 hour operational NWP models from other centers such as ECMWF, NCEP, NCMRWF, UK Met Office, and Germany
Training of National experts in NWP
Assist NMHS to source for software and hardware facilities for NWP operations

Enhancing Medium Range NWP at ICPAC in Support of 10day bulletins and NMHS.

Set up and update continuously the operational databases necessary for the medium range and super ensemble real time runs.
Extension of the FSU super ensemble NWP to 10days and its use in ICPAC dekadal bulletin.
Verify and use 10day NWP products from other centers: COLA, NCMRWF and others and run Eta model operationally at ICPAC.
Under take capacity building training on medium range NWP and it’s operational challenges.

Climate modelling and downscaling for seasonal products at ICPAC

Operational use of the ICPAC-IRI calibrated global ECHAM model and dynamical downscaling with the Regional Spectral Model (RSM) for prediction of October-December seasons.
Operational use of Model Output Statistics (MOS) of the global model as statistical downscaling at ICPAC and NMHSs in seasonal forecasts.
Creation and analysis of the ECHAM model database for seasons MAM, and JJA and evaluation of the skill of the ECHAM and RSM for these seasons.
Configuration and adoption of FSU Super ensemble model products for seasonal forecasts.
Identification of other models and modeling techniques being used operationally at other centers such as ECMWF, NCEP, BMRC, NCMRWF, UKMET Office, Max Planck Institute of Meteorology, and NCSU among others, and the customisation of those models in the region.
Acquire a global model that can be run at ICPAC.
Continuation of regional capacity building in dynamical climate modeling.
Acquisition of PCs and clustering technology for parallel processing. «top

[BACK | HOME ]