Climate Modelling - The Answer Is Time Travel

Feb 7 - Posted by Tory Aardvark

Never one to stop flagellating a dead horse, The Guardian are getting excited because a Climate Scientist Witchdoctor has found that by adding 2 lines of programming code to a process, he can now actually predict weather 5 whole days in advance.

Cast your mind back a year or so ago, when we were expected to believe the computer models for decades in advance, now the Warming Alarmists are celebrating being able to predict 5 days in advance, bit of a shortfall in time spans, but it is religion not science, so what the hell time to start banging gongs and barking at the moon again.

Apparently current weather and climate models do a decent job

    As you read these words, time is flowing smoothly, passing from one moment to the next with no breaks. Reassuring, isn't it?

    But in computer models of weather and climate, time has to jump forward in little steps to allow the next set of temperature, rain and other conditions to be calculated from the last. This stepping, Paul Williams at the University of Reading assures me, has the technical name of 'leapfrogging'.

    Both weather and climate models currently do a decent job. Weather can be predicted pretty accurately up to about four days ahead and the broad patterns of global climate are well replicated in climate models. That's because over past decades a multitude of researchers have tested and refined the models, improving how they represent the atmosphere and oceans and their interaction.

Broad patterns of climate can be modelled, that's not what we were sold with rising sea levels, vanishing glaciers and global climate catastrophe, strange how the story changes and no one is supposed to notice, could well be time to tell Al Gore about his new clothes.

    The addition of just a couple of lines of code to the model led to the five-day weather forecast being as accurate as the old four-day forecast. That would mean 24 hours more notice of what's on the way. And to get climate models right, you need to first get the weather right, so good news there too.

A couple of lines of code has improved the model for 5 days accurate forecasting, written with all the hope and ignorance of one not schooled in software development, anyone who has ever worked on large sophisticated programs knows that the number of lines of code required to increase the program's ability increase exponentially, as it becomes more sophisticated. At some point the whole thing will become largely unmaintainable due to the sheer number of lines of code, at which point you have to start again with a new design, development and testing process which can take years.

Essentially what the Guardian have missed, or ignored is that current climate models are rubbish and patched to hell to give the results that Michael Mann and Phil Jones want, and that any new climate models are probably 10 years away from being debugged and their results calibrated.