Weather Prediction, Climate Prediction. What’s the Diff?
If scientists can’t accurately predict the weather next week or the week after, how can they predict the climate in 10 or 20 years? Good question. The answer lies in apples and oranges.
PopSci.com welcomes back Dr. Bill Chameides, dean of Duke’s Nicholas School of the Environment. Dr. Chameides blogs at The Green Grok to spark lively discussions about environmental science, keeping you in the know on what the scientific world is discovering and how it affects you – all in plain language and, hopefully, with a bit of fun. Now, PopSci.com partners with The Green Grok to bring you exclusive new blog posts a week before they hit the Grok’s blog. Give it a read and get in on the discussion!
Scientifically speaking, the difference between weather prediction and climate prediction is the difference between an “initial value problem” and a “boundary value problem.” Let’s see if I can explain in English.
While weather and climate both focus on temperature, wind, cloudiness, rain or snow, the way these properties are used is quite different. The National Center for Atmospheric Research defines the two like so:
- “Weather is the mix of events that happen each day in our atmosphere.”
- “Climate … is the average weather pattern in a place over many years.”
When Time and Place Are Critical
When you want to know the weather, time and place are critical. You are interested in what is going to happen in the immediate future (not sometime in the next month or two) and in your vicinity (not 1000 miles away).
If the TV weather person announced it was going to rain somewhere in your state sometime next month, I suspect you’d find that prediction a little less than satisfactory. But the latter is essentially what a climate prediction is, and the methodology to arrive at it is fundamentally different from predicting the weather.
What Goes into Weather Predictions
Imagine you are a center fielder on a baseball team. The batter hits a fly ball your way and it’s your job to catch it. To do so, you need to figure where in center field the ball is headed and when it’s going to get there.
If you’re a good outfielder, a crack computer in your brain gathers up essential data — like the speed of the bat as it hits the ball, the sound of the impact, and the ball’s initial direction — and in a split second calculates the ball’s trajectory.
But to do this well, it’s essential that the input into your computer – what scientists would call the initial values – is complete and accurate. If the glare of the sun or stadium lights obscure your view of the ball’s initial flight, your ability to accurately predict where the ball is going and when it will get there is impaired.
Predicting the weather is similarly dependent on the initial values you specify in the computer model used to make the prediction. These initial conditions include temperature, wind speed, wind direction, and precipitation rates everywhere in your model – essentially everywhere in the atmosphere. The values for these parameters can’t be made up; they must come from real data. Today these data come from the global meteorological network run by countries around the world and largely coordinated by the World Meteorological Organization. This network includes surface meteorological stations, balloon measurements, shipboard measurements, and space-borne platforms.
Despite its very impressive size, the network is limited; we can only make meteorological measurements in so many locations and these measurements are not perfectly accurate. Thus, the initial conditions input into our weather models are imperfect, and so our weather predictions are inaccurate — and would be even if our understanding of the physics of the weather were perfect.
Because the effects of imperfect initial conditions on weather simulations tend to grow, the longer the weather model is run into the future, the less accurate the prediction. Predictions of the weather just a week or two in advance, let alone decades, become highly problematic.
What Goes into Climate Predictions
Less concerned with exact time and place, predicting climate focuses on spatially and temporally averaged conditions.
Unlike the earlier example of the outfielder who must know exactly where the ball is heading and when it will get there, climate prediction is more akin to predicting at the beginning of the game how many times a ball will be hit to center field sometime in the first three innings. Initial conditions like the speed of the bat or direction of the ball as it leaves the first batter’s bat are not going to help very much. The critical factors are the speed and direction of the wind, the properties of the ball and the bat, the strength of the pitcher and the batters, and the dimensions of the field – factors that scientists call boundary conditions.
So while predicting the weather depends critically on getting the initial state of the atmosphere right, predicting the climate does not. Which is not to say that climate prediction is easy. It’s not.
Predicting climate accurately depends on getting a host of those boundary conditions correct, many of which relate to the atmosphere’s energy. They include the amount and strength of sunlight reaching the Earth, the reflectivity of the Earth’s surface, the movement of heat in the oceans, and the opacity of the atmosphere to terrestrial radiation as a result of greenhouse gases. And for this reason, getting long-term, accurate observations of, for example, the variations in the sun’s output of energy over time is critical for understanding past climate change. Uncertainties in how the sun’s output will change in the coming decades limits our ability to predict future climate with complete confidence. However, such decadal variations in the sun’s output are irrelevant to predicting tomorrow’s weather.
There are other fundamental differences between weather and climate predictions. Some of these relate to mechanisms. For example, accurate weather predictions require a good simulation of the processes that lead to precipitation from a cloud since whether or not it rains at a specific location on a specific day is relevant. For climate predictions, the specifics of the cloud-to-rain process are less important. Far more important is getting right the reflective properties of the cloud since these affect the planet’s long-term energy budget. Again, both of these inputs present difficult but different challenges.
And that’s why comparing the limitations of weather predictions with those of climate predictions is a little or a lot like comparing apples and oranges.