When I ride my bicycle from an open space trail to the Safeway parking lot less than 100 yards away, I frequently see 5-10 degrees difference in temperature. Asphalt and snow removal make huge differences. The USHCN UHI adjustment is about 0.1 degrees F over the last century.
NOAA docs show 8F UHI in Phoenix, 80 times higher than what NCDC adjusts for.
Ignoring a host of other issues about TOBS, according to USHCN V2 docs the USHCN V2 TOBS adjustment is supposed to look like this – positive since 1980, and slightly negative prior to that.
This is somewhat similar to V1 (dashed line below) except for a different offset, and a smoking gun of very different behavior after 1990.
But the actual implementation of V2 TOBS (thick blue line below) is very different from the supporting docs. The actual adjustment is about 0.2C lower than the graph in the specification (red and thin blue) lines. It is abundantly clear that their code does not correctly implement the specification.
U.S. Data Since 1895 Fail To Show Warming Trend
By PHILIP SHABECOFF, Special to the New York Times Published: January 26, 1989
After examining climate data extending back nearly 100 years, a team of Government scientists has concluded that there has been no significant change in average temperatures or rainfall in the United States over that entire period.
The study, made by scientists for the National Oceanic and Atmospheric Administration was published in the current issue of Geophysical Research Letters. It is based on temperature and precipitation readings taken at weather stations around the country from 1895 to 1987.
U.S. Data Since 1895 Fail To Show Warming Trend – New York Times
NOAA now shows about half a degree warming during that same time period. They have rewritten their own history.
I spent the evening comparing USHCN V1 and V2 graphs, and discovered a huge discrepancy between their V1 and V2 adjustments.
This is their current US graph. Note that there is a discontinuity at 1998, which doesn’t look right. Globally, temperatures plummeted in 1999-2000, but they didn’t in the US graph.
Climate at a Glance | Time Series
It doesn’t look right, because they made a huge change going from USHCN V1 to V2.
In V1 they adjusted recent temperatures upwards (thin line below) and made little adjustment to older temperatures.
But when they switched to V2, they started adjusting older temperatures downwards, and left post-2000 temperatures more or less intact. This created a huge jump (greater than one degree) downwards for all years prior to 2000. You can see what they did in the animation below.
Blue line is thermometer data. Thin red line is V1 adjusted. Thick red line is V2 adjusted. Thermometer data and V2 data are normalized to the last five years, creating a small offset on the Y-axis. They created a significant warming by reversing polarity of the adjustment in the pre-2000 years. How did a peer-reviewed positive adjustment suddenly become a negative adjustment?
A big clue has been right in front of my eyes all along. Note that measured data below shows that by 2008, temperatures were back down to the 1989 level. But in the NCDC data, 2008 is half a degree warmer than 1989.
Bottom line is that there is clearly a huge error in the USHCN adjustments which has added a non-existent one degree hockey stick warming to the official US temperature record, and I now know just where to look for it in their code.
FOIA time.
This is what the adjusted temperatures would look like, without the error. Ignoring the fact that the adjustments themselves are probably not valid.
NOAA made a big deal about 2012 blowing away all temperature records, but the temperature they reported is the result of a huge error. This affects all NOAA and NASA US temperature graphs, and is part of the cause of this famous shift.
According to USHCN 1 docs, the total adjustment is supposed to be about 0.5F upwards, and flat after 1990.
ts.ushcn_anom25_diffs_urb-raw_pg.gif (650×502)
But in USHCN2, the adjustments are much larger, downwards, and actually accelerate after 1990. The USHCN2 adjustments are supposed to be similar to USHCN1, and the TOBS adjustment is supposed to be identical.
This graph shows the USHCN1 (red) and USHCN2 (blue) adjustments at the same scale and offset. Note that the adjustments went from being positive in V1 to being negative in V2. Also note that in V1, the adjustments went flat after 1990, and in V2 they rise exponentially after 1990.
RSS satellite temperatures show that by 2008, US temperatures had cooled down below what they were during all of the non-volcano years of the 1980’s and 1990’s.
RSS does not show the NCDC discontinuity, and NCDC is diverging from RSS at almost 1C/century
Here is an animation of the complete set of USHCN2 adjustments, which turn a 90 year cooling trend into a warming trend.
Visualizing How USHCN Hides The Decline In US Temperatures | Real Science
According to the USHCN2 docs, the TOBS adjustment is supposed to be almost identical to USHCN1. But in the actual data, it isn’t even close. Points to a software bug.
ftp://ftp.ncdc.noaa.gov/pub/data/ushcn/v2/monthly/menne-etal2009.pdf
There are additional problems evident in the docs. The USHCN2 docs show TOBS continuing up after 1990, but the USHCN1 docs show TOBS going flat after 1990. USHCN1 showed 2ºF TOBS, but USHCN2 shows 2ºC TOBS. It is supposed to be the same adjustment, and according to USHCN1 docs should have gone flat after 1990 – but is almost double the USHCN1 adjustment.
ts.ushcn_anom25_diffs_pg.gif (650×502)
The next graph shows the actual implementation of V2 TOBS vs. the spec. The implementation (thick blue line) is offset down from the specification (thin red and blue lines) by about a quarter of a degree. How did a positive adjustment after 1980 turn into a negative adjustment? Fuzzy math?
The actual TOBS adjustment above was calculated by subtracting the raw temperatures ushcn.tavg.latest.raw.tar.gz from the TOBS adjusted temperatures ushcn.tavg.latest.tob.tar.gz.
Bottom line is that the NCDC US temperature record is completely broken, and meaningless. Adjustments that used to go flat after 1990, now go up exponentially. Adjustments which are documented as positive, are implemented as negative.
Record temperatures can’t be explained by TOBS, homogenization, or any of their other fake adjustments. The vast majority of all-time record maximum temperatures were set before 1960, and never matched again.
The USHCN V2 docs say that they did not change the TOBS adjustment from V1.
Given these verifications, the Karl et al. (1986) TOB adjustment procedure was used in HCN version 2 without modification
ftp://ftp.ncdc.noaa.gov/pub/data/ushcn/v2/monthly/menne-etal2009.pdf
The V1 docs show less than 0.4F TOBS increase since 1920, going flat after 1990.
The V2 docs show more than 0.3C exponential increase since 1920, with no leveling after 1990. Same adjustment specs, same data set, widely divergent numbers. Obviously an error.
The next graph overlays the V1 TOBS (black dashed line) on the V2 TOBS at the same scale. They are supposed to be identical and match closely until 1950, but diverge after 1950. The V2 version does not go flat after 1990, and by 2007 is double what it is supposed to be. The times of observation from 1995 obviously did not change after 1999. NOAA appears to be pulling a fast one.
This doesn’t even consider the fact that TOBS assumes that all observers are idiots. Almost anyone operating a min-max thermometer will recognize in a matter of days that they have to reset the thermometer at night. I figured this out when I was about seven years old. TOBS is not a subtle problem, and I suspect that anyone with an IQ higher than a turnip will quickly figure out how to avoid it.
The New York Times said in 2000 that the North Pole was ice-free for the first time in 50 million years
Ages-Old Icecap at North Pole Is Now Liquid, Scientists Find – New York Times
They also reported that it was ice-free in 1926, which must have been 50 million years ago.
h/t to Jimbo
NOAA claims that the raw US minimum temperature trend from 1895-2007 is 0.054/C per decade
ftp://ftp.ncdc.noaa.gov/pub/data/ushcn/v2/monthly/menne-etal2009.pdf
This is complete BS. The actual 1895-2007 raw minimum trend is 0.020/C per decade. Whatever they are using for “raw” is not the raw data.
They then adjust tamper the trend to 0.90C/data, which means that they have increased the trend to 450% of its measured value – indicating a S/N ratio close to zero.
Maximum temperatures are even worse. They increase the trend to 700% of its measured value (0.009C/decade.)
No legitimate scientist would release complete crap like this, which is why NOAA does release it – without any warnings.
