New Arctic weather station installation
Well.. we could move all the thermometers:
“NOAA . . . systematically eliminated 75% of the world’s stations with a clear bias towards removing higher latitude, high altitude and rural locations, all of which had a tendency to be cooler,” the authors say. “The thermometers in a sense, marched towards the tropics, the sea, and to airport tarmacs.”
“Scientifically” they say this is cool because they, say, take a thermometer from Toronto and “interpolate” it for what the Arctic would be. In other words, “guess”.
Oh.. .interpolation can be useful. Most images you see, that look pretty real, have interpolated data. But… interpolation looks at past data sets and would not capture changes in the places that used to be measured but that are independent from places where stations remain.
Weather is big and complex, interpolation is guessing.
January 25th, 2010 at 8:59 am
http://www.lgt.lt/geoin/doc.php?did=cl_subsurface
Boreholes and time for land. It’s anyone’s guess about the other 3/4ths!
January 25th, 2010 at 9:33 am
Let’s distinguish interpolation from extrapolation which are both valid techniques. Interpolation estimates within the bounds of whatever data you already have. Extrapolation estimates beyond the data range you already have. Each has its place and a way to estimate its accuracy. Estimates that are given from either technique should accompanied by the accuracy estimate.
If all the data points are south of Toronto, any estimates north of Toronto would be extrapolations, which does not therefore necessarily make them wrong. How good they are depends on how good the measured data are, what you know about the regression function, and how far north your are trying to estimate.
That kind of extrapolation from a regression function is different from estimating by applying a math model of the world which is much more complex but more able to accommodate variations in the input forces that create the temperatures.
January 25th, 2010 at 9:38 am
No matter how good the interpolation or extrapolation is, it will also be based on historic relations and will not capture changes in the weather – which is sort of the whole point of the exercise.
January 25th, 2010 at 10:14 am
If by “changes” you mean time variation in addition to spatial variation, it’s true that by definition any future time is outside the data range and therefore it is an extrapolation. If you are relying on a regression model that combines space and time, you can at least estimate the error in extending to the future.
But if you are using a math model that correctly describes the governing differential equations,the laws of physics and chemistry, and the physical properties of all pertinent matter, and you believe that neither those equations nor the laws will change in the future, you can feel a lot better about predicting the future in the absence of some unexpected driving force (asteroid impact, volcano eruptions, extreme radiation from space, sun changes, etc).
That’s in principle. In real life, many of the details of physics and chemistry are only vaguely known, such as the absorption of CO2 by the oceans and their creatures. Thus, the math models are a continuing project to discover those details.
January 25th, 2010 at 10:18 am
Ultimately, it seems that if we are to base trillions of dollars and millions of lives in the balance on this science, they might want to spring for a few more weather stations.
I suspect, however, that interpolation/extrapolation/guessing meets their grant and political needs better.
January 25th, 2010 at 11:05 am
Good idea, more data beats less data. Give some credit to the free press for keeping the scientists honest, although there’s not much the good scientists can do about political players focusing on real and imaginary data errors. Where big money and power are at stake, you can expect a lot of chicanery and posturing.
January 25th, 2010 at 1:36 pm
One could argue that any measurement on earth is interpolation, points around a sphere’s surface. Extrapolation would then be data from the moon or sun for example. It must have been fun to predict storms in Spain when the earth was thought to be flat and finite!