Global Warming 101: Ghost Data

I'm highlighting some posts from some time back, that will give you a grounding in the science of climate change.  The data seem scientific, but get curiouser and curiouser the more you look.

-------------------------------------------


Ripogenus Dam  (October 26, 2009)

The science is settled.

OK, so what's with the Ripogenus Dam?

You don't get much more rural than that. Way, way up the west branch of the Penobscot river in Maine, it's the sort of place that Boy Scouts go for week long canoe voyages through the wilderness. 1972, Troop 47, a dozen fellow teenagers and I spent a week a hundred miles from any other living soul.


In 1972, there was a Weather station at the Ripogenus Dam. It collected temperature readings every day. Those readings were included in NASA's GISS temperature reading data set. Its readings were included in GISS until 2006, along with data from thousands of other weather stations. There's really only one little problem.

The Ripogenus Dam weather station was decommissioned in 1995.

So for ten years, GISS reported temperature readings from a station that didn't exist. How? Filnet.

Part of the USHCN data is created by a computer program called “filnet” which estimates missing values. According to the NOAA, filnet works by using a weighted average of values from neighboring stations. In this example, data was created for a no longer existing station from surrounding stations, which in this case as the same evaluation noted were all subject to microclimate and urban bias, no longer adjusted for. Note the rise in temperatures after this before the best sited truly rural station in Maine was closed.

"Urban bias" is the technical term for when a weather station reads artificially high temperatures because the station is situated in an urban location where there are lots of buildings and parking lots to absorb the heat from the sun. Filnet took temperature readings from other weather stations - stations in urban locations where reading are higher because of the surrounding asphalt heat collectors - and used it for the most rural station in the state.

Remember how 1998 was the "warmest year in a millennium"? Well, it was warmer than it would have been if the Ripogenus Dam's readings hadn't come from Millinocket.

The question is: just how unreliable is the data? Lots.

How can we trust NOAA/NASA/Hadley assessment of global changes given these and the other data integrity issues? Given that Hadley has destroyed old original data because they were running out of room in their data cabinet, can we ever hope to reconstruct the real truth?

Given that there are only 30 or 40 stations that have been providing temperature readings from the Civil War up to today, given that "adjustments" are made to the temperature data via an arcane and opaque process and may represent most or all of the warming in the 20th Century, given that scientists refuse to release their data (or the raw - unadjusted - data has been destroyed), given how some of the data sets rely on tree rings from a single tree, how can we trust the data?

The Ripogenus Dam weather station was giving ghost readings for a decade. How many other non-existent weather stations are still generating new data? The World wonders.

UPDATE 31 October 2009 17:20: David linked. Thanks! Take a look around - there's a lot more on this sort of thing here.