F. Replay Strategies within ADAS

The idea of replaying the ADAS to a given analysis is applicable to multiple circumstances. The primary situation is one of trying to obtain extra output from a cycling ADAS that has perhaps been missed when the experiment first ran. Another type of replay arises when changes being implemented in the system are not expected to affect the ensemble too drastically so that in a hybrid methodology would simply replay to ensemble to that of an existing experiment; also useful to get a first glimpse of what changes might bring to the system without having to incur the cost of running the ensemble. Other cases may represent a desire to study the consequences of changes in the aerosol analysis without bothering with direct impact to the meteorological analyses; or vice versa, when the user simply wants to replay to existing aerosol analyses without bothering about the effects of changes to aerosols due to changes in the meteorology, and finally further motivations that may arise from the creativity of every user.

Replaying hybrid ADAS to given ensemble

In recent years, the ADAS replay strategy that has been mostly exercised and found to be important in getting preliminary results from first-order testing is that of ensemble-replay. Clearly, this strategy applies to systems that run some form of hybrid ensemble-variational methodology (either 3D or 4D).

Replaying a new experiment to the existing ensemble of another is controlled by a file named atmens_replay.acq. This is a single-line file providing the templated name of the location of a previously

Replaying ADAS to its own outputs

Replaying ADAS to alternative analyses

Replaying ADAS to alternative aerosol analyses