WP3 - Seamless Atmospheric Integrations

• To develop a seamless prediction system from medium-range to seasonal scales.
• Provide the regional seasonal predictions necessary for the other tasks of the project applying state-of-the-art methodologies and using existing global simulations (mainly from ECMWF ) complemented by the regional simulations performed by University of Pretoria (UP), focusing on southern Africa.
• Production of regional seasonal predictions as a multi-model multi-downscaling unified approach, including also an estimation of the uncertainty.

Task 3.1.a: Developing seamless calibrated products for disease-related variables integrating the output from medium-range, monthly and seasonal ensemble forecast systems. Calibration of model output versus local observations/satellite estimates will be complemented by projection of model-generated anomalies onto regional empirical-orthogonal functions derived from gridded observational datasets. (ECMWF, ICTP)

Task 3.1b: Examine possible gain of multi-model seasonal forecasts with statistical approaches (analogs, weather types, neural networks) in terms of health-relevant climate/weather variables and health prediction. In this task we will apply state-of-the-art methodologies to regionally projected large-scale model outputs from global models, building mainly on ECMWF simulations including hindcasts. We will also explore the capabilities of probabilistic networks as a global methodology for multi-site statistical downscaling in a multi-model framework. This task will use data produced in WP1.2. (CSIC, UP, ICTP)

Task 3.1c: Examine possible gain of seasonal forecasts with dynamical downscaling methods (regional models) in terms of health-relevant climate/weather variables and health prediction. (UCAD, CSIC)

Task 3.1d: Production of regional seasonal predictions as a multi-model multi-downscaling unified approach and estimation of the associated uncertainty. This task will produce the seasonal predictions needed by the disease models in the target regions (Senegal, Ghana and Malawi) by optimal combination of models and downscaling techniques. To this aim, bias correction for raw/downscaled monthly data tailored to the health relevant climate variables will be produced (univariate and bivariate correction methodology) and weighting procedures based on compound-weights composed of different terms (annual cycle, variability, extremes, etc.) will be applied to the different models and downscaling algorithms. Mechanisms to translate the uncertainty to the predictions will be also explored. (CSIC, UP, UCAD, ICTP)

D3.1.a - Assessment report on the skill of global seasonal predictions in Africa using a quintile interval-based validation. M9 (CSIC).

D3.1a: Assessment report on the skill of global seasonal

D3.1.b - QWeCI Statistical Downscaling Portal established and open to partners with an initial set of statistical-based seasonal predictions for the target regions, with documentation and support on good practises of use. M15 (CSIC, UCAD, UP)

D3.1b: QWeCI Statistical Downscaling Portal

D3.1c Report on the skill of dynamical predictions for southern Africa. M24 (UP)

D3.1.c – Report on the skill of dynamical predictions for southern Af

D3.1d – Regional seasonal predictions, with uncertainty estimations, in the target regions, and inclusion in the QWeCI statistical downscaling portal, accessible to partners in user-friendly formats. M24 (CSIC, UCAD)

D3.1d: Regional seasonal predictions, with uncertainty

D3.1e. Report on the seamless calibrated products for disease-related variables integrating the output from medium-range, monthly and seasonal ensemble forecast systems using ECMWF products. M28 (ECMWF).

M3.1.a Prototype seamless products from monthly to seasonal EPS systems. M12. (ECWMF)

M3.1 a: Prototype seamless products from monthly to seasonal EPS systems

Describe the characteristics of African temperature and precipitation in interannual and decadal time scales and assess and improve the state-of-the-art forecast quality with dynamical and statistical models.

Task 3.2a: Assessment of decadal predictability over Africa: use of available decadal global forecasts, comparison with uninitialized global projections (AR4 type) and with observed-SST runs; focus on the pilot study areas and on the reproducibility of the decadal Sahel drought (IC3, ECMWF).

Task 3.2b: Validation of intraseasonal variability in global decadal forecasts: frequency of dry days, monsoon onset, characterization of the PDF of precipitation (IC3, ICTP).

Task 3.2c: Model combination, bias correction and generation of reliable decadal climate information suitable to drive disease models (IC3, UCAD, UOC).

D3.2.a: Preliminary assessment of the characteristics of interannual variations of the intraseasonal variability of African temperature and precipitation in dynamical models and observations. M24. (IC3, ECMWF, ICTP).

D3.2a: Preliminary assessment of the characteristics of interannual variations o

D3.2.b: Report on the advantages in terms of forecast quality of the combination of dynamical and statistical models of interannual and decadal variability for Africa. M30. (IC3, ICTP, UOC, UCAD).

D3.2b: Report on the advantages in terms of forecast

M3.2.a Prototype seamless decadal ensemble system completed, with assessment of the predictability of precipitation and temperature over Africa in interannual and decadal time scales. M18. (IC3, ECMWF, ICTP)