nai

Project Progress

Many environmental processes contribute to the reflected stellar and emitted thermal spectrum of a planet. To simulate these processes for a range of extrasolar terrestrial planets, the VPL will incorporate radiative transfer, climate, chemical, geological, exogenic, and biological algorithms. Given a series of initial conditions, these methods will use a forward time-stepping approach to march the environment toward a plausible equilibrium state. The radiative transfer, climate and chemical modules are based on well tested, existing codes. The other modules are being developed in parallel with the model integration activity.

A modular VPL software architecture design will:

1. minimize changes to existing codes,
2. allow flexible evolution rates (time-splitting) between processes,
3. allow parallel development of algorithms for environmental processes, and
4. optimize the combined model for parallel computing

This architecture is illustrated in Figure 1. The “Initialize” block contains an assumed initial state (stellar type, distance from star, environmental characteristics) and specific process-control information and convergence criteria. The major time-step loop contains a series of processors (exogenic, geology, chemistry, climate, biology) connected by “translators”. Each translator takes the output from the previous module, combines it with information from the initiation module and an evolving common database, and writes input files that can be read by the next processor. The rapidly evolving processors (e.g. chemistry, climate, biology) might execute internal time-step loops between major time-steps to ensure stability. Time-stepping continues until convergence criteria are satisfied. The VPL then outputs a spectrum of the planet for evaluation by the instrument simulator.

{{ 1 }}

This year we defined the basic architecture, and the requirements for the Translators and Common database. Integration of the chemistry and climate models will happen first. The remainder of the modules will initially be specified as static “state vectors”, and later replaced as the full models come on line.