However, it does seem to be a programming task, so let me guess just a bit.
There are a couple parts to a simple computer model.
1. The data that is supplied to the model.
2. The weight that is given to each in the input parameters when the model is executed.
3. The calculations that the model does to produce some output.
If you wanted to check the sensitivity of the model, you'd first run it with all of the parameters set as near as you can get to real world conditions.
Then you'd vary one condition, or weight at a time, to see how it affected the output.
And you'd do that for all of the weights and input parameters, so that you would be able to see how well the model compared to real world conditions.
Keep in mind, we have fairly good weather data for several decades.
What you'd like to get is a model that, given the data for the 1980s, produced the climate that has already occurred in the 1990s.
And then do the same for each decade, or other reasonable period for which reliable data is available.
When the model works well for all the tested time periods, then you have a pretty good idea that the weights you've used for the various inputs look reasonable,
You are going to see several comments saying that models don't work, or are too complicated, or that climatologists don't know what they're doing. All from people who actually don't know what they're talking about.
Computer models generally have more than one input value that can be changed. Climate models are no different.
So what you could do is set up the model so that is uses something close to real world conditions. Then you select one of the inputs, and vary it slightly and note the results. Then you vary it some more and note the results again.You keep repeating this process.
Eventually, you will be able to plot the change in your input parameter against the associated change in output. For instance, you might change the input parameter that sets the current level of CO2 and then see how this affects the temperature.
If your model is very sensitive to a particular parameter (that is a small change in that parameter causes a big change in output) then youi have to be sure that you set that value as accurately as you can.
If you can vary an input parameter by a large amount and the output hardly changes then you don't need to know that so accurately. In fact you mighht consider dropping it from the model altogther.
In the case of climate, models this process is not straightforward for two main reasons:
1. The models take a long time to run. So getting even one result will consume a lot of computer time.
2. Climate models don't always produce the same output for any given set of inputs. There is a random element to them so typically many modelling runs have to be averaged to produce an average model output.
You build a model of the climate, then you see the end result in amount of warming. This tells you how sensitive the planet is to increasing CO2. For example you may get 3C of warming or 5C of warming. Changing the inputs parameters will get you a different result. You can even get below 1C if you modify how the model treats clouds.
It amounts to confirmation of your beliefs, in that with modelling you can achieve any result you wish, by altering the inputs.
Climate sensitivity should be calculated using empirical data, and that would be very dificult also.
Lets face it until we can also determing climate sensitivity to a whole range of other factors, such as solar cycles, ocean cycles, cloud cover, volcanoes etc, isolating sensitivity to CO2 is difficult, however as CO2 continues to escalate, and the other factors go through their cycles, we should be able to estimate in a few years time by comparing the temperature record.
the paleo climate record
