Decadal climate predictability in the West African monsoon region and in the cyclogenesis region of the tropical Atlantic

The project DEPARTURE intends to assess decadal climate predictability in the West African monsoon region and the Atlantic region of tropical cyclogenesis closing the gap between nowcasting efforts and long-term climate change projections. Major motivations are the economic relevance of hurricane activity and the sensitivity of African economies and food production systems to future climate changes at the decadal to multi-decadal time scale in an expected climate change hot-spot region. The dynamical downscaling is done with the regional climate models (RCMs) REMO, CCLM and WRF on 50 km horizontal resolution based on SST-driven ECHAM6/MPIOM and, possibly, HadCM3 simulations. The RCMs are forced by sea surface temperatures (SSTs), increasing greenhouse gas (GHGs) concentrations, aerosols and land cover changes (LCCs) to utilize the forecast potential from oceanic, atmospheric and land-surface boundary conditions. Step by step, the extended boundary conditions are incorporated in the RCMs and sensitivity studies and three-member ensembles, respectively, are realized (see Fig. 1). In addition, fully coupled ocean-atmosphere simulations with REMO are carried out to assess the importance of an interactive ocean and a SVAT (Surface-Vegetation-Atmosphere-Transfer) module is applied in CCLM to estimate the effect of interactive soil and vegetation processes. For the hindcast experiments we rely on the decadal ECHAM6/MPIOM simulations for the 1960-2015 period provided by the MiKlip community. The realised transient hindcast simulations allow for a first assessment of forecast skill and uncertainty in the RCM multi-model ensemble. Based on global climate model simulations developed within the MiKlip framework operational decadal forecasts are conducted as 10-ensemble runs from each RCM over the period 2015-2025 allowing a probabilistic forecast for quantification of uncertainties. Finally, all RCM simulations and the driving ECHAM6/MPIOM (and HadCM3) runs are validated with observations of African rainfall and Atlantic tropical cyclones, mainly focusing on measures of hurricane activity, the hydrological cycle in sub-Saharan Africa, extreme events and key atmospheric processes involved in rainfall generation and tropical cyclogenesis. In particular, the relative impacts of the various boundary conditions are quantified and tested by analysis of variance and Bayesian statistics. Systematic model errors are eliminated by statistical post-processing for impact research.

Goals

• Implementation of past and future boundary conditions including GHGs, aerosols, LCCs, SSTs, soil and vegetation processes
• Performance of transient sensitivity studies and ensemble runs for each RCM in the hindcast period 1960-2015
• Assessment of forecast skill and uncertainties within the RCM multi-model ensemble concerning hurricane activity, hydrological cycle and impacts of various boundary conditions
• Statistical post-processing of RCM output for impact research
• Performance of probabilistic decadal forecasts for the period 2015-2025 with 10-ensemble runs for each RCM