Watts Posts Paper Claiming "Spurious" Warming in U.S. Climate Record

[skierinvermont]

What are "pro-AGW papers"? I thought there were science papers. You introduce the aspect of debate and motive when implying there are sides to the issue.

When a study is done on ocean CO2 absorptivity for instance, do you feel the purpose of the study is to prove or disprove AGW? The aggregate of all studies tends to support AGW so of course more funding goes into those studies looked at in hindsight.

Tell me Watts' motive is not first to help dispel the concept of significant global warming by proving the U.S. temperature record to be questionably accurate. Let's hope his study, which just happens to coincide with the release of the BEST study, just like 'Climategate' just happened to coincide with 'Copenhagen' bears some scientific fruit.

Agree strongly with this post. ORH is confusing good science papers with "pro-AGW papers."

[The_Global_Warmer]

So I am assuming Watts used 0z, 6z,12z, and 18z?

But didn't adjust for data skewing from the time of the OBS? Or he took obs from different times from the same station? like 12Z and 15Z.

Using 0z, 6z, 12z, 18z would tend to be bias towards max cooling and overlook max warming more often throughout the US on a yearly basis. Especially in Spring and Fall when 21Z tends to be around max temp.

While 12z stays closer to min temp more often.

Can someone help elaborate if I am anywhere near it?

[ORH_wxman]

The TOBS issue is legit with the Watts paper, but it will likely still leave a large discrepency even when you apply it given that the TOBs trend from his time period ('79-'08) is on the scale of hundreths of a degree per decade. I don't believe there has been any TOBs adjustments since 1990 anyway.

[tacoman25]

The UAH database lists the linear trend for the continental United States at +0.23C/decade since 1979. See: http://vortex.nsstc....t2lt/uahncdc.lt. This is lower than the trend for USHCN reported by NCDC, but also significantly higher than the trend reported in the Watts study.

Just for the heck of it, I did my own analysis of the UAH data using the data to create a chart in Excel. I came up with a trend of +0.0019C/month, which comes out to +0.228C/decade, or +0.23C/decade when rounded to two significant figures. I've attached the chart below, which shows an unmistakable warming trend for the continental U.S. Also, I'm not sure whether the NCDC figure includes Alaska & Hawaii, in addition to the contiguous 48 states. My understanding is that Alaska has generally been warming more rapidly than the Lower 48. So if it includes those two states, that could account for some of the discrepancy. Although Alaska is only one state, it is the largest in area.

Start the analysis in 1980 and end in 2009 and you'll find a different trend. 1979 was a very cold year for the U.S., while 2011 was very warm.

[Entropy]

I think most of the available evidence points towards a legitimate warming of the climate system. Anthony raises some good points about the validity of the land-based temperature records, so I thought maybe a look at the temperature of an inland lake would be a good place to see whether the warming signal is robust -- or if it is just an artifact of urban warming, or siting problems. I knew NWS BUF has an archive of water temperatures taken continuously at the same location since 1927. I realize this is only one site, but it should be a reliable indicator of warming in the Lake Erie vicinity. I exported all of the data since 1960 to Excel, converted it to Celsius, and displayed it as an anomaly relative to the 1960-2012 mean. As you can see, the annual mean water temperature at the site is trending up 0.29C per decade. So far, in 2012, the mean water temperature has been 2.8C above the 1960-2012 average. If this holds, it would be the warmest on record. The current record is +1.8C, set in 1998.

I also looked at the number of days in which the water temperature was 32F. To do this, I used the =COUNTIF(source, value) function to return the number of days with a 32F reading for each year. These readings are taken 35' below the surface of the lake and usually read 32F when the surface is substantially ice-covered. There has been an observed decrease of one such day per year since 1960. The most such days was 139 in 1963, followed closely by 138 in 1971. The least 32F readings in a full year is 0 in 1998, followed closely by 1 day in 2002. So far, there has been no such days in 2012. Since 1990, there have only been three years (1993, 1995, and 2000) in which the temperature reached 32 before the end of December, so it is unlikely that 2012 will see any such days.

I also returned the maximum value using the MAX(source, value) function. The annual maximum water temperature has been trending upwards, approximately 0.5F per decade. The warmest reading is 80 in 1988 and 2011. The coldest is 70 in 1992, which was the summer following the eruption of Mt. Pinatubo.

[Entropy]

Start the analysis in 1980 and end in 2009 and you'll find a different trend. 1979 was a very cold year for the U.S., while 2011 was very warm.

I did it for the 1979-2008 period in the other chart, which I believe was the time frame Anthony looked at in his study. The linear trend was actually a bit higher at the end of 2008, because 2009 and 2010 were cooler than normal in the U.S.

[PhillipS]

Start the analysis in 1980 and end in 2009 and you'll find a different trend. 1979 was a very cold year for the U.S., while 2011 was very warm.

I believe Entropy used the same period as the Watts study for an apples-to-apples comparison.

We all understand that by cherrypicking start and end dates very different trend values can be generated.

[ORH_wxman]

I think the UAH trend is not far from what you'd get with the Watts study if you properly took into account TOBS. Though TOBS may be a bit less than that, but its not cut and dry because TOBS has a higher rate of occurance in rural stations than urban stations. The menne et al (2010) study showed a TOBS of 0.02C per decade between 1985-2006 for USHCN stations which is a bulk of the Watts time period. If you add on a bit because its more common in rural stations and also a bit because we go back to 1979 and not 1985, you can probably make up at least a good chunk of the discrepency between his value and the UAH trend. We'll have to wait until someone actually runs that on his data before we know what the adjustment would be.

[WeatherRusty]

Papers on Time of Observation Bias

[stellarfun]

From what I see, attempts to draw correlations between temperature and the spectrum of changes in the physical geography of the United States are few to non-existent..

Irrigated agriculture in the United States in acres

19 million in 1920

19 million in 1930

50 million in 1970

63 million in 2000

Forest land area in acres

1630: 1,045 million (obviously estimated)

1907: 759 million

1997: 746 million

2007: 671 million (difference between 1997 and 2007 may largely be due to reclassification)

Urban land area in acres

1945: 15 million acres

2007: 61 million acres, 3 percent of the nation's land area

Urban land area increase by decade

1960s 9 million acres

1970s 13 million acres

1980s 9 million acres

1990s 8 million acres

Rural residential land area in acres

73 million in 1997

103 million acres in 2007

While I don't have irrigated agriculture values for 1945, I suspect the growth in acres in irrigated agriculture over the past 60 years approximates the growth in urban land area. And I believe irrigated land tends to lower ambient temperature.

And has Watt done a sensitivity analysis, by decade, that correlates changes in temperature with changes in the urban land area?

[Jonger]

From what I see, attempts to draw correlations between temperature and the spectrum of changes in the physical geography of the United States are few to non-existent..

Irrigated agriculture in the United States in acres

19 million in 1920

19 million in 1930

50 million in 1970

63 million in 2000

Forest land area in acres

1630: 1,045 million (obviously estimated)

1907: 759 million

1997: 746 million

2007: 671 million (difference between 1997 and 2007 may largely be due to reclassification)

Urban land area in acres

1945: 15 million acres

2007: 61 million acres, 3 percent of the nation's land area

Urban land area increase by decade

1960s 9 million acres

1970s 13 million acres

1980s 9 million acres

1990s 8 million acres

Rural residential land area in acres

73 million in 1997

103 million acres in 2007

While I don't have irrigated agriculture values for 1945, I suspect the growth in acres in irrigated agriculture over the past 60 years approximates the growth in urban land area. And I believe irrigated land tends to lower ambient temperature.

And has Watt done a sensitivity analysis, by decade, that correlates changes in temperature with changes in the urban land area?

Forestland should show an increase since 1997.

[stellarfun]

Jonger1152, I found another source that has a total without the reclassification: 752 million acres of forest in 2007, up from 734 million acres in 1987. The increase in forest land over this 20 year period is quite comparable to the increase in urban land area.

For the majority of months, forested land lowers the air temperature when compared to open land.

For the effects of irrigation on air temperature, see:

http://journals.ametsoc.org/doi/abs/10.1175/2007JCLI1755.1