Bumps and Wiggles (1): Predictions and Projections
Introduction: This starts a series on understanding climate variability, global warming, and what we might do about it. The series focuses on the past 30 years and the next 30 years. Much of what I will be writing about is derived from the work that I am currently involved with at various federal agencies. In the world of the blog, and amongst many of my friends, there is a lot of angst and anxiety about the politically motivated aspects of climate change. While this is wildly and widely amplified, there is an unprecedented amount of real, substantive work. Since I feel I have done any good I can do on that political side, I’m going to (try to) practice what I preach, do no harm, and push forward in matters of substance. I have collected a few of my entries about the political and emotional aspects of the, graciously, debate, and linked them below.
Let’s start to think about bumps and wiggles.
Back in October I wrote an entry about a figure
that was being used to make the case that IPCC
predictions are wrong because they predict that the last four years should be getting warmer in each successive year and that has not been observed.
My first reaction to that argument was that the IPCC
predictions were never intended to be a “forecast” in the spirit of a weather forecast, and that there was nothing in the record of the last four years that suggested to me any problem with the "forecasts." Spirit of a weather forecast? We have become used to “deterministic” weather forecasts, which are broadcast far and wide on, of course, Wunderground.com
. (I have heard there are other sources of weather information as well.) A deterministic forecast is one where information about a particular place at a particular time is projected; that is, the high tomorrow afternoon in Newnan, Georgia
will be 67 degrees F.
We are also used to some information that comes with a specified probability, for example, the probability of precipitation tomorrow is 70%. With a number as large as 70%, we are pretty certain it is going to rain. This introduces another type of forecast, namely, the “probabilistic” forecast which provides a range of, for example, temperature; that is, the temperature is very likely to lie within the specified range. If you think about it, even the deterministic forecasts have an implicit range. That is, we don’t think it is a bad forecast if the temperature is a couple of degrees above or below a predicted value.
The statement of a probability suggests several things. First, it is a measure of error – perhaps several types of errors. There are measurement errors, and there are errors in the formulation of forecast models. One way to derive “forecast error” is to compare a bunch of forecasts with observations and calculate how much the observations and the forecasts differ. Another factor that contributes to the specification of probabilities is what we might call “noise.” "Noise" in determining the quality of a weather forecast might be related to the presence of a lake in the middle of a small town, a thunderstorm crossing over the thermometer in the middle of the afternoon, or the vagaries of turbulent flow near the surface of the Earth. (Here are basics of modeling
for, let’s say, science-interested people.)
So back in October when I first saw the use of this figure to question the veracity of the IPCC
forecasts, I felt that there was no real challenge to the science-based conclusions about global warming. The first reason I felt this was because these “forecasts” make no attempt to represent the variability of any specific year. As all of you weather-savvy readers know, there are many sources of variability, such as El Nino. (El Nino at the Climate Prediction Center
) The climate models strive to represent the variability of a “generic” El Nino, but they do not attempt in any way to represent specific El Ninos; for example in 1983, 1998, or 2010.
The models used for the IPCC
report do contain such variability, but when they are all averaged together for the report, this variability cancels out. In this way, the variability is “noise” to the climate. When we look at the past 4 years, then we have to ask whether or not what the IPCC
projects is outside of the normal range of this climate noise. It was not, so the four year record, was to me, inconsequential.
There is another important point to make to a weather-savvy community that might read a Wunderground.com blog
. When a weather forecast is made, the forecast starts from an observed state of the atmosphere (and increasingly the ocean and the “land surface”). That is, all of those surface observations, airplane observations, balloon observations, and satellite observations are melded together to tell us what the weather “looks like.” Then this set of measurements is projected forward in time. This is called a forecast or a prediction.
The same sort of process is used in “seasonal prediction,” which is what El Nino forecasting is called. In seasonal prediction, the ocean is very important, so if we can predict that the eastern tropical Pacific ocean will warm up, then we can, minimally, suggest how the weather patterns in much of the Northern Hemisphere will respond. Again, scientists working in this field tend to call their products “predictions” or “forecasts.”
Many climate scientists, especially, careful ones, don’t call their products predictions, rather they call them “projections.” This is meant to suggest that we know that these model estimates of the future are not deterministic in the traditional spirit of a weather forecast. Rather they provide information about, perhaps, the average temperature change and a range above and below that average.
There is one more point I would like to make in this introduction. Above I mentioned “noise.” Noise is a relative term, and some people would even say that when we talk about climate, weather is “noise.” (This is a casual and erroneous statement.) Sometimes when we think of noise we think of something being random. If we think random, we think unpredictable. Some of us even resort to the more sophisticated concept of “chaos.” There is a well known theory that weather is “chaotic,” which suggests some upper limit of WEATHER prediction at a couple of weeks. Chaos and Weather Prediction from National Geographic
. This does not, let me repeat, does not mean that the Earth’s climate is a random, unpredictable system.
Think of El Nino
. If you examine the last twenty years of progress we predict El Ninos a lot better. This is because we observe and model the oceans a lot better, and there are many features in ocean variability that are “predictable” (or potentially so) for months, perhaps years, in advance. And since the ocean impacts strongly what happens in the atmosphere, we can suggest with some confidence how weather patterns will respond.
Therefore, from an atmospheric point of view, the ocean “forces” a particular regime of behavior. Thinking of the climate system, while there is a random aspect to the climate, when we talk about the surface warming, then we are talking about “forced” response that comes from he addition of enormous amounts of greenhouse gases.
With this introduction, the next entry will be about understanding bumps and wiggles in the past tens years and for the next ten years.
Here is Figure 1 from Lean and Rind:
Figure 1 from Lean and Rind
(2009), Geophysical Research Letters
. Figure taken from tinypic.com
. This figure shows the temperature record and the model representation from 1980 to 2030, the subject of this series of articles.Entries about the political and emotional aspects of the climate change debateIf lady Chatterley’s Lover, thenStrength in Many Peers“Have you no sense of decency, sir, at long last?” Trust, but VerifyScientist as AdvocateScience, Belief and the VolcanoOpinions and Anecdotal Evidence And here is Faceted Search of Blogs at climateknowledge.org