date: Tue Feb 26 17:01:29 2008 from: Phil Jones subject: Re: Early Temperature in the US (1776-, 1804-05, 1810-16) and in to: FMims@aol.com Forrest, Only picked this up. It is a pity the water vapour measurements don't extend further back in time. Early expedition records and explorer's logs are difficult to compare with modern measurements, as they are often short in duration. When making comparisons you have to remember they are essentially long runs of weather and the long-term change from the 19th century to the present is only about 1 deg C. This 1 deg C is small compared to weather variability. Attached is another paper on early US weather obs. There are a lot of issues to consider with early observations. I also tried years ago to look at early Antarctic weather obs from the explorer days. The day-to-day variability of Antarctic temperatures almost defeats anything being said, except perhaps in the Peninsula region. Cheers Phil At 15:30 26/02/2008, you wrote: Dr. Jones, I very much appreciate the time you have taken to reply to my questions. I had not seen your intriguing paper in NATURE on surface humidity trends since 1973. Would that this paper could have extended several decades back beyond 1973 so that we could see the impact, if any, of the cooling of the 60s and 70s on RH trends. I will look more into this as time permits. Of special interest is your group's use of fewer stations than GISS and other groups. I agree completely with this "quality beats quantity" approach. Based on extensive personal experience measuring the enormous variation of temperature in and around "heat islands" ranging in size from a single brick to the nearby city, I simply find it incredible to presume that temperature records corrupted by the presence of asphalt, concrete and buildings can be corrected using simple regressions if at all. As you seem to be doing, a single, properly sited rural station uncorrupted by the presence of nearby buildings and roads is far better than a host of corrupted and supposedly corrected other sites. (I am familiar with various papers and plots concluding otherwise.) Regarding early temperature observations in the US, Susan Solomon's papers on temperature measured by the Lewis and Clark Expedition and by President Thomas Jefferson show close agreement with modern measurements, a fact I noticed in 1991 while studying Jefferson's temperature record for 1810-1816 (which includes the eruption of Tambora). According to Solomon: "It is not surprising that Lewis and Clarkâs observations during a single winter two centuries ago lie within the range of twentieth-century data, in spite of global climate change." Susan Solomon and John S. Daniel (2004). Lewis and Clark: Pioneering Meteorological Observers in the American West, Bulletin of the American Meteorological Society 85, 1273-1288. And: "A great deal of further analysis beyond the scope of the present work would be needed to determine whether the Jefferson or Madison observations may be of scientific value in the context of the modern challenge of global warming, but it is interesting that the Madison record (which is far more complete than that of Jefferson) is not inconsistent with cooler conditions in the late 18th century, particularly in summer." Susan Solomon, John S. Daniel and Daniel L. Druckenbrod (2007), Revolutionary Minds, Thomas Jefferson and James Madison participated in a small ârevolutionâ against British weather-monitoring practices, American Scientist 95, 430-437. I am just completing the writing of a rather long book on the history of Hawaii's Mauna Loa Observatory (going back to 1794 and the Menzies expedition) and was intrigued to find that the temperature at Hilo in February 1841 (US Exploring Expedition) is in close agreement with modern measurements. Here is what I wrote, and I would be most appreciative if you will offer criticism: "Some of the meteorological and geophysical measurements made by the Ex. Ex. on Mauna Loa and at Hilo can be compared with those made today, in particular the elevations of Mauna Loa and Mauna Kea, the atmospheric pressure, the temperature measured by the shaded thermometers and the water temperature of Hilo Bay. "For example, from January 31 to February 28, 1841, the mean temperature measured at Observatory 1 at Hilo, was 70.93 F (21.6 C) (Wilkes, 1851, 486). The mean temperature from 1946 to 2006 about 0.8 mile (1.3 km) away at the Hilo National Weather Service station was 71.3 F (21.8 C) (WRCC, 2006), a difference of +0.4 F (0.2 C). Very close agreement is also found in the temperature of the water measured from the USS Vincennes in Hilo Bay near Coconut Island and at a modern site in the bay. The mean water temperature for February 1841 was 73.09 F (22.8 C), which is slightly warmer than the present February mean of 72 F (22.2 C), according to the National Oceanographic Data Center (2007). The difference of 1.1 F (0.6 C) should be considered provisional, for the depths at which the temperature was measured may have differed. From Chapter 3, "Fifty Years of Monitoring a Changing Atmosphere: The Story of Hawaiiâs Mauna Loa Observatory." In December 2007 I placed temperature loggers at the site of Observatory 1 and as close as I could get to the official Hilo Weather Station (at the airport), and the difference is very close to the difference in the 1841. Thus, there seems to have been a minimal change between the single February of 1841 and the mean of the past 40 years of February. Best regards, Forrest Forrest M. Mims III [1]www.forrestmims.org [2]www.sunandsky.org Geronimo Creek Observatory Phone: 830-372-0548 In a message dated 2/26/2008 3:04:04 A.M. Central Standard Time, p.jones@uea.ac.uk writes: Forrest, All cities have UHIs, but this doesn't always mean that they are warming more that their rural neighbours. It depends how long they have been where they are. London, UK has a UHI of about 1.5 deg C compared to rural neighbours, 40km north and south of the city. The one longish record in Central London doesn't warm more than the rural sites though over the last 50 years. Vienna is similar as are a number of large European cities, as they have been where they are a very long time. The fact that they have a UHI doesn't mean therefore that they will necessarily have an additional urban-related warming trend. This is only Europe though, but it does seem to apply to some large Asian cities as well. In North America, the situation is different as the cities weren't there in most cases 100 years ago. I did recently see a paper on a long daily site in Kansas (Manhattan, KS) where a group has developed a record back to 1828 from observational data. It showed a warning since 1828 of 2.25 deg C, but the warmest years are in the 1930s Dust Bowl years. The point of the paper was that long records can be developed for the USA (not quite as far back as Europe) and that they show long-term warming. I think the authors have submitted the paper to a journal. As I said earlier, it is important to know which sites we've used and there is a list on our web site. Surfacestations.org does show many poorly located sites in the US, but there are loads of sites in the US, so there is no need to use the sites which have problems and/or numerous site moves. We don't use as many US sites as NCDC or GISS, as our aim isn't to use all, but a selection over the US. We put more effort into many other regions of the world, where coverage is less and it is important to make best of the sparser record. You mention water vapour. Globally this is rising - at the surface where there are longish records. Here's a paper on this. The dataset on this is in submission to J. Climate. Also, here's another paper from 1997. This shows that if the main aim of a study was to measure the global average T from land regions, it is only necessary to use a much more limited number of sites. This is because temperature between stations is generally very well correlated. Statistically this means that lots of stations aren't independent of each other, so there are only a limited number of spatial degrees of freedom. When using fewer stations, however, you have to make sure they are correct and fully homogeneous. Best Regards Phil At 18:04 25/02/2008, you wrote: Dear Dr. Jones, Thanks very much for the PPT slides and new links. There is considerable new information (for me, at least) in "Uncertainty estimates in regional and global observed temperature changes: a new dataset from 1850." I will study this and the other items you suggested before proceeding. Regarding the two PPT slides you sent, the significance of the black line in the histogram is certainly apparent. However, I was also struck by the similarity in the downward correction in the chart at left to many GISS charts of the USA. Briefly, early temperatures are adjusted downward, sometimes by up to several degrees C, while recent temperatures are not adjusted (as in the PPT you sent), even at sites with serious problems. I looked at the annual T for San Antonio, Texas, on the NWS web site. I loaded the data from 1885-2007 into Excel, did a linear regression and found that T has increased by only 0.22 F since 1885. I then looked at 1990 to 2006 and found an increase of 0.06 F. I then found a paper from the Southwest Research Institute showing a 3 degree F increase in San Antonio compared with neighboring but small New Braunfels. In short, subtracting the urban heat island effect from San Antonio leaves a significant cooling. Confidence in this is enhanced by my personal measurements on the porch of my rural office. This site does not meet NWS standards, but nothing has changed since 1990 when I began measuring the temperature. The linear regression yields a decline of 2 degrees F between February 4, 1990 to a week ago. I have also measured the total column water vapor (or precipitable water) here since Feb 4, 1990 using a simple, highly stable instrument , and there is a slight downward trend of 0.06 cm/decade (even including a spike around 1999). Please see the attached chart. (One of the papers on this instrument is F. M. Mims III, An inexpensive and stable LED Sun photometer for measuring the water vapor column over South Texas from 1990 to 2001, Geophysical Research Letters 29, 20-1 to 20-4, 2002.) Of course my site and nearby San Antonio are only points on the globe. But at least I understand them and well and have the full record of physical locations of the San Antonio weather station. Combining these facts with the many deficient sites revealed at [3]www.surfacestations.org is why I wrote to your organization. The most interesting record you sent is the surface ocean data. This seems to show an increase only about 0.1 C less than the land record, which is good agreement. I will look more into this to learn about sampling methodology. Best regards, Forrest Forrest M. Mims III [4]www.forrestmims.org [5]www.sunandsky.org Editor, The Citizen Scientist [6]www.sas.org/tcs In a message dated 2/25/2008 10:56:32 A.M. Central Standard Time, p.jones@uea.ac.uk writes: Dear Forrest, Before you go too far, I would suggest you look at the paper by Brohan et al I mentioned to you in the earlier email. You can get it from [7]http://www.cru.uea.ac.uk/cru/data/temperature/ in the list of references on the page. Have a look at Figure 4. I'm attaching a couple of ppt slides, I sent to a Chinese colleague a few months ago. In the first ppt Figure 4 is repeated on the right. Ignore the left plot and the red.blue lines on the right. Just look at the black line. The text of the paper explains what the red/blue lines if you want to find out. What this Figure is is a histogram of adjustments to temperature records we've applied and also those that have been applied by the Canadians at AES. So what is there is a count of adjustments of a certain value. There are few around zero and then peaks at about +0.5 to +1.5 and -0.5 to -1.5 degrees Celsius. What this means is that making adjustments for site changes tends to average out, some sites move to warmer locations, some to cooler locations. The second ppt is a similar type of plot - the frequency of adjustments (the black line). This is US work from NCDC Asheville. Again the adjustments tend to cancel. overall, when looking at the US Lower-48 station average. When I say cancel out, I mean the trend in the time series of the before and after (adjusted) stations is much the same. So, we have taken the whole issue very seriously. The people at Climate Audit don't think we have, but we've tried to summarise the results of all the adjustments we've applied to instrumental data around the world in diagrams like these. It is clear that you need each record to be homogeneous and just be affected by weather and climate (and not exposure, site changes etc). Averaged over a large region, though, if the large-scale average series was all you were interested in, it all doesn't make much difference. So this doesn't work for single stations, but it does for large averages. There is one other reason, that the NH land temperature average is essentially correct. It agrees with the complementary record from marine data taken by ships. So we've taken it very, very seriously. We spent person years on it in the 1980s and 1990s It is only when you do this, you begin to realise what needs to be adjusted and what does not. We are making and have made numerous adjustments. You have to read numerous papers to find out what they are. I have also told you where the list of sites we use is located. Cheers Phil ___________________________________________________________________________________ Delicious ideas to please the pickiest eaters. [8]Watch the video on AOL Living. Prof. Phil Jones Climatic Research Unit Telephone +44 (0) 1603 592090 School of Environmental Sciences Fax +44 (0) 1603 507784 University of East Anglia Norwich Email p.jones@uea.ac.uk NR4 7TJ UK ----------------------------------------------------------------------------