The answer can be simple. And it can be extremely complicated.
So complicated that I'm not even sure if I'm sharing science or pseudo-science.
But I shall try my best to explain.
Climate does not equal weather.
In a way (but not so in a way), weather is a part of climate.
I heard a really useful analogy once (and hopefully my brain made the right connections).
Imagine an empty swimming pool.
You're dumping loads of water into the pool.
You measure how big the pool is; how much water you need; how quickly is the pool filling up; what would happen if you poured even more water in?
That's climate.
Then you see how many tiny ripples are produced; how many bubbles are produced; how does the water splash around?
That's weather.
In climate, you look at things on a larger, more macro scale.
But you really need to focus and look real close when it comes to weather.
That's why we're more confident at looking at the big picture (climate) and not sweating the little details (weather).
To be more specific, weather is everything happening in the atmosphere at any moment.
Climate is just a statistical average of the climate over a long long period of time (Ie: what's the average temperature of 2014?).
A lot of times (at least for me), I see the results of climate models in the form of temperature.
"Ohhhh if we keep emitting carbon dioxide, the temperature will increase by x amount."
Why don't they ever give us something more "weather forecast-y"? Like air circulation?
That's because air movement relies on the laws of Dynamics, something we don't fully understand yet.
However, what we do understand is the Law of Thermodynamics.
That's why we're more confident in predicting temperature, of all the variables to look at.
That's definitely a bonus.
We use what we know as an indicator of future climate, and rely less on something we're unsure of.
At least then we can call our results 'credible' and don't risk getting any 'risky' predictions.
Another way to look at this is through the problems that arise from predicting weather and climate.
As I've mentioned previously, after predicting at 5-7 days' worth of weather, Mr. Chaos Theory kicks in and we get predictions that might be very different from what will actually happen.
We can also call this 'high sensitivity to initial values'.
Our tools aren't accurate enough to provide measurements that will minimise this sensitivity.
If we have more accurate tools, maybe we could predict a month into the future before Mr. Chaos Theory screws us up.
This applies to weather, and not climate (according to some).
In the climate system, it's believed that if we create a realistic model environment, everything will 'work out in the end'.
At some point, no matter how crazy the system goes, it will reach an equilibrium, a stable state.
This realistic model environment includes things such as concentration of greenhouse gas in the atmosphere, amount of solar radiation we're receiving, the flow of air currents in the atmosphere etc.
If those values (we call them 'boundaries') are set to the same as our Earth, we can get a pretty accurate reading on things.
So weather prediction is a 'initial value' problem.
And climate prediction is a 'boundary value' problem.
These sort of problems are really opinion-driven.
Some people think that climate modelling is purely a 'boundary' problem.
But others, like my wonderful lecturer, think that it's a 'boundary' AND 'initial value' problem.
Geez, I hope I explained that well enough (and true enough).
It would be depressing to end this 'trilogy' on an uncertain note.
So I'll write about a simple, yet fairly effective solution scientists have developed.
It's called "Ensemble Modelling".
What you do is basically, run several models at the same time.
But with each of them, you use some slight different initial values.
At the end of it, you can see how the models reacted differently to the similar inputs.
If you graph that up, you'll get a range of possible outcomes (Figure 1).
And that's really helpful, because replicating the modelling run several times gives a sense of consistency to our results.
Depending on how you use ensembles, you can either test to see how consistent your models are, compared to others (by putting the same inputs in and observing what comes out); or how slight changes in input values can change the results (by using the same model but with slightly different inputs).
This simple, yet computationally mind-blowing, technique has enabled us to become more confident in our forecasting and it gives us a rough idea of how large our uncertainties are.
Thanks to this, we've been able to do weather predictions up to 10-14 days into the future (instead of 5-7).
This is only a small step in creating reliable climate models, but our innovations won't stop there.
We'll continue developing better techniques and stronger supercomputers.
And one day, we might actually be able to plan our fishing trips months in advance.
 |
| Figure 1. The model output from a ensemble modelling run. The different-coloured lines represent different climate models used in this process. Although every model produced different results, once we overlap them, we can actually observe a rough area of what the possible climate might be. After that, they calculate the mean of all these results (the thick black line). Ensembles also tell us how big our uncertainty is. The wider the coloured-band areas are, the more uncertain these projections will be. Image taken from RealClimate. |
And that concludes this 3-part blog post about my assignment: climate modelling.
Funny thing is, I've already submitted my assignment a week ago before writing this.
Let's just say, by the end of the week I can't be stuffed with writing a good essay anymore.
Besides, it's only worth 10%.
Now I have another paper to write. But I probably won't blog about it.
Just because it looks incredibly dry and boring.
Plus, I wouldn't know how to blog about it even if I tried.
Good news is, I'm already halfway through the semester.
Just a little (actually 50%) more!
Listening to Hey Ya! - Glee Cast
Back to Writing,
TK
0 comments:
Post a Comment