This sounds like good science being done. When you are trying to predict something like carbon emissions, standard practice is to run a few different scenarios. If you run 5 models, one with predicted CO2, and the others with higher and lower values, of course some of your scenarios will be unlikely. You're doing that on purpose so that you have a ability to see how wide the range of possible results is even with uncertainty in CO2.