VErification, Calibration and Assessment of Predictability of medium-range climate predictions using satellite data

Of central importance for the development process of the MiKlip system is the data and process based validation of the decadal scale prediction system during the three development stages of MiKlip. Furthermore, the generation of the decadal forecasts is based on an ensemble approach which requires the deduction of suitable probabilistic information from the different ensemble simulations. Of crucial importance is the evaluation of atmospheric water cycle components including clouds and related changes in the radiation budget. Additionally, the representation of extreme precipitation events requires the accurate simulation of precipitation variability.

The VECAP project contributes by the provision of a basic suite of statistical methods for the evaluation of the decadal hindcast and forecast experiments. According to the MiKlip validation strategy the suite enables a standardised validation of the forecast system using scores and skill scores. The suite will continuously be upgraded during the runtime of the project aiming at a comprehensive data and process based validation.

The project contributes to the evaluation of the decadal scale prediction system by comparison of the simulations with reanalysis and satellite products. Virtual satellite measurements calculated from the climate simulations for comparison with satellite based observations such as inferred from the Infrared Atmospheric Sounding Interferometer (IASI) and the Tropical Rainfall Measuring Mission precipitation radar (TRMM-PR) are developed and used for the evaluation of climate feedbacks and precipitation variability. While the focus of the project is on the evaluation of the global model system the developed methods will potentially also be suited for application to the regional model simulations.  

Goals

• Elaborate a concept for standardised validation according to the MiKlip validation strategy (cooperation with EnsDiVal) to validate the simulations in an (i) absolute sense against observations using scores and (ii) in a relative sense against the previous version of the forecast system using skill scores together with observations and a predefined set of data/index values derived jointly from simulations and observations focusing on ECVs (GCOS Essential Climate Variables, cf. http://www.wmo.int/pages/prog/gcos).
• Provision of a suite of statistical methods for the evaluation of ensemble simulations (handed over to INTEGRATION) which will be gradually extended during the runtime of the project (cooperation with EnsDiVal, INTEGRATION).
• Development of methods for the extraction of stochastic and probabilistic information (probabilistic post processing), quantify the level of uncertainty, validation and calibration against observations on the basis of the available samples of single forecasts.
• Implementation of the COSP satellite simulator package, development and inclusion of additional satellite simulators which will provide the simulated reflectivities of the TRIMM-PR and the simulated outgoing longwave spectrum as observed by the IASI instrument.
• Data based validation, assessment of the predictability, consideration of observation based dataset uncertainties, calibration of the probabilistic information against observations.
• Process based evaluation, identification of predictable spatial and temporal scales, structures and related processes, identify processes which lead to the generation of uncertainty using classical observations and satellite-based products, validation of energy cycle components at the atmosphere-ocean interface and of the tropical water cycle (cooperation with AODA-PENG). 
• In-depth analysis of processes relevant for (a) predictability and (b) uncertainty growth for guiding the development process of the MiKlip system.
• Exemplary adaptation of methods for regional model output.