Difference between revisions of "Random starting conditions"
Revision as of 16:53, 11 April 2008
Randomised Initial Conditions
To what extent are the proposed negative feedbacks able to robustly control the individual properties of the Earth System? If they exert robust control then it shouldn’t matter where the system starts from; they should converge on a steady-state (the same steady-state) regardless.
This can be examined for each JModel by changing the initial conditions (these are the values of the different state variables at the start of a model run). To examine the effect of changing one initial condition by itself, launch the model then click on the ‘Initial Conditions’ button. Edit the value in the relevant text box, click ‘Apply and Close’ and then click ‘Run’ on the main window.
After having done this, examine the resulting plots to see if all plots eventually reach a plateau (no change in the value of the variable with time) by the end of the run. If all variables do plateau out, the system has converged to a steady-state. In order to determine whether or not it is always the same steady-state that they converge to, the final values at the end of each run can be examined by clicking ‘Show Final State’ on the main window. The final values can be compared to those for the default run (which starts off at equilibrium) by clicking 'Default' on the 'Initial Conditions' window and then 'Apply and Close' and re-running the model.
A second, more stringent, test is to examine the effect of changing all the initial conditions simultaneously. This can be done by bringing up the ‘Initial Conditions’ window again, then clicking on ‘Rand’ followed by ‘Apply and Close’ and re-running the model.
This website is in its early stages of use. If you find it difficult to run a model in the way described, or find any other problems, your feedback will help us improve the site for future users.