How to Convert a Skeptic
Richard A Muller
www.BerkeleyEarth.org Richard Muller 18 April 2013 Berkeley Earth Scientific available online at Rohde et al., Geoinfor Geostat: An Overview 2013, 1:2 Papers http://dx.doi.org/10.4172/gigs.1000103 Geoinformatics & www.BerkeleyEarth.org our papersGeostatistics: An Overview Research Article a SciTechnol journal
Geoinformatics & largest collection of raw data. According to the summary provided by 2012, 1:1 Berkeley Earth Temperature the Intergovernmental Panel on Climate Change (IPCC), the mean Rohde et al., Geoinfor Geostat: An Overview global surface temperature (including land and oceans) has increased Averaging Process 0.64 ± 0.13°C from 1956 to 2005 at 95% con!dence [1]. In a review http://dx.doi.org/10.4172/gigs.1000101 Geostatistics: An Overview Robert Rohde1, Richard Muller1,2,3*, Robert Jacobsen2,3, Saul of temperature changes over land areas, the IPCC summarized four a SciTechnol journal Perlmutter2,3, Arthur Rosenfeld2,3, Jonathan Wurtele2,3, Judith reconstructions of the global land average temperature as having Curry4, Charlotte Wickham5 and Steven Mosher1 trends ranging from 0.188 ± 0.069°C/decade to 0.315 ± 0.088°C/ decade over the time interval 1979 to 2005 [1]. However, some of this Wickham et al., Geoinfor Geostat: An Overview range re"ects methodological di#erences in how “land average” was http://dx.doi.org/10.4172/gigs.1000104 only mean temperatures; however, the three groups reported changes Abstract de!ned and over what regions it was computed. Research Article that ranged from 0.81 to 0.93°C when estimating the increase in land A new mathematical framework is presented for producing maps $e three major groups that produce ongoing temperature 2013, 1:2 and large-scale averages of temperature changes from weather temperatures during the 2000s decade relative to the 1950s decade. reconstructions are the NASA Goddard Institute of Space Science station thermometer data for the purposes of climate analysis. The As described below, we estimate this change as 0.90 ± 0.05°C (95% (NASA GISS), the National Climate Data Center at the National "dence). method allows inclusion of short and discontinuous temperature A New Estimate of the Average con Oceanic and Atmospheric Administration (NOAA NCDC), and records, so nearly all digitally archived thermometer data can be Research Article Earth Surface Land Temperature Methods and Materials used. The framework uses the statistical method known as Kriging the joint project of the UK Meteorological O%ce Climatic Research 2,3, to interpolate data from stations to arbitrary locations on the Earth. Unit and the Hadley Centre at the University of East Anglia (Hadley/ In this paper we present results for the Earth’s land surface Geoinformatics & 1,2,3* Robert Jacobsen2,3, CRU). $eir annual land-surface temperature histories are presented Geostatistics: An Overview Spanning 1753 to 2011 4 outliers. Statistical uncertainties are calculated by subdividing 1, Richard A. Muller 2,3, Arthur Rosenfeld temperature only, based on analysis of monthly averages at each in Figure 1A, as well as the available uncertainties in Figure 1B. NASA 1 3 the data and comparing the results from statistically independent Robert Rohde , Saul Perlmutter and Charlotte Wickham station. We gathered and merged monthly and daily thermometer GISS does not publish an uncertainty speci!c to their land-surface Elizabeth Muller 2,3, Donald Groom subsamples using the Jackknife method. Spatial uncertainties measurements from 14 databases to arrive at a collection of 14.4 data product. In Figure 1A we show that these groups report a range Jonathan Wurtele from periods with sparse geographical sampling are estimated by million mean monthly temperature observations from 44,455 sites. of best values from 0.81 to 0.93°C when estimating the increase in ese data have now been posted online in calculating the error made when we analyze post-1960 data using During this process duplicate! stations present in the 14 databases were similarly sparse spatial sampling. Rather than “homogenize” the land temperatures for the 2000s decade relative to the 1950s decade, the Global Hi detected and eliminated. Abstract with reported 95% uncertainties of roughly 0.15 to 0.2°C. stations are located in cities with a populationa SciTechnol greater than 50,000. journal a uniform format at www.BerkeleyEarth.org, along with a description data; the data are then broken into two parts at those times, and the storical Climatology Network Monthly (GHCN-M) If the typical urban station exhibited urban heating of the magnitude We report an estimate of the Earth’s average land surface of the merging and duplicate removal method. For stations that parts treated as separate records. We apply this new framework to During the second half of the twentieth century weather of Tokyo, this could result in a severe warming bias in global temperature for the period 1753 to 2011. To address issues report only daily data (and not their own monthly average) we the Global Historical Climatology Network (GHCN) monthly land monitoring instruments of good quality were widely deployed, yet the temperature dataset, and obtain a new global land temperature averages using urban stations. To avoid this bias the urban heating of potential station selection bias, we used a larger sampling of performed the average. We removed only short records (less than 1 quoted uncertainty on temperature change during this time period Wickham C stations than had prior studies. For the period post 1880, our Groom D 1, Rohde R contribution to global temperature change should be isolated to the year) and records from sites with missing or highly uncertain location 3 estimate is similar to those previously reported by other groups, # 36,866 stations that we used in our analysis. in close agreement with prior estimates made by the groups at , Jacobsen R2, Muller RA greatest extent possible. metadata; that le 3, 3, although we report smaller uncertainties. The land temperature rise $ered from that of the previous groups in NOAA, NASA, and at the Hadley Center/Climate Research Unit in 4, Perlmutter4* ,S Wurtele J from the 1950s decade to the 2000s decade is 0.90 ± 0.05°C (95% 3, Our analysis approach di Abstract 3,4 4 !e goal of this paper is to evaluate the urban heat island and Rosenfeld, Curry A J contribution to the Berkeley Earth Surface Temperature global air by 0.89 ± 0.06°C in the difference of the Jan 2000-Dec 2009 5, increased during the last century. Diurnal variations decreased several ways. Rather than adjust (homogenize) individual records for The effect of urban heating on estimates of global average 3 temperature average over land. Detailed analyses of average land known and presumed discontinuities (e.g. from instrument changes for statistical and spatial uncertainties). land surface temperature is studied by applying an urban-rural temperature time series of the Earth’s surface (T but not understood. The period of 1753 to 1850 is marked by and station moves), we split the records into portions occurring by three major teams: the NASA Goddard Institute for Space Science #er such apparent discontinuities, creating! essentiallye detection Keywords sudden drops in land surface temperature that are coincident before and a !is procedure was completelyscalpel. automated to Earth temperature dataset compilation of 36,869 sites from 15 (GISS), the National Oceanic and Atmospheric Administration different publicly available sources. We compare the distribution with known volcanism; the response function is approximately two records from one. Earth surface temperature; Kriging; Global warming; Temperature (NOAA), and the collaboration between the Hadley Centre of the UK 1.5 ± 0.5°C per 100 Tg of atmospheric sulfate. This volcanism, of linear temperature trends for these sites to the distribution for a reduce human bias; we call this approach the homogenization Met O avg ) have been reported combined with a simple proxy for anthropogenic effects (logarithm % Anglia (HadCRU).ce and the Climatic Research Unit of the University of East of the CO 2 of such breakpoints followed procedures similar to those used by with one-third of the stations in the US and worldwide having a the e existing groups, but the traditional adjustment step was omitted in "ect of urban heating on their global averages. Abbreviations !ey di addition of a solar forcing term. Thus, for this very simple model, favor of simply dividing the time series into two pieces at any apparent!e of the three groups is that the" urbaner in the heat methods island contribution used to account to their for solar forcing does not appear to contribute to the observed global breakpoints. We also split records when there was a gap in record NASA: National Aeronautics and Space Administration; GISS: series of the Earth’s average land temperature are estimated using global averages is much smaller than the observed global warming. warming of the past 250 years; the entire change can be modeled continuity greater than 1 year in duration, and at times when changes Goddard Institute of Space Science; GHCN: Global Historical !e topic is not without controversy. We ask whether the presence by a sum of volcanism and a single anthropogenic proxy. The rural subset; the difference of these is consistent with no urban of urban stations results in overestimates of warming in the Berkeley in station location or time of observation were documented. Climatology Network; IPCC: Intergovernmental Panel on Climate heating effect over the period 1950 to 2010, with a slope of -0.10 ± !e conclusion residual variations include interannual and multi-decadal variability Change Earth Surface Temperature global land average. scalpel approach avoids explicit adjustment of" thet. It data, is possible the process to use usually called “homogenization”, although it increases the number of Global warming; Kriging; Atlantic multidecadal oscillation; Keywords !e approach of the GISS team is to identify urban, “peri- Keywords: parameters that are used to create the best Introduction urban” (near urban) and rural stations using satellite images of Amo; Volcanism; Climate change; Earth surface temperature; Diurnal the scalpel approach because our reconstruction method depends While there are many indicators of climate change, the long- nighttime lights [ variability less on long duration samples than do the methods applied by prior term evolution of global surface temperatures is perhaps the metric Figure 1: (Top Panel) Comparison of annual land-surface average by subtracting a two-part linear trend based on comparison to an !e 36,866 records were split, on average, 3.9 times% toagged create the temperature anomalies for the three major research groups [2-4]. For this 8] Urban and peri-urban stations are then adjusted groups. that is both the easiest to understand and most closely linked to the purpose, the Hadley / CRU simple average has been used rather than average of nearby rural stations. Introduction 179,928 record fragments. When we detected other problems (e.g. quantitative predictions of climate models. It is also backed by the the more widely cited latitudinal-band weighted average, as the simple Introduction global average is a reduction of about 0.0°C in warming over the undocumented changes from Celsius to Fahrenheit) we average is more similar in methodology and results to the other averages period 1900-2009. !e average Earth surface temperature is a key indicator of presented here. (Bottom Panel) The 95% percent uncertainty estimation the !e result of the adjustment on their changes; the raw uncorrected data are available online in a separate !e Urban Heat Island (UHI) e climate change. Previous estimates have been reported by three major *Corresponding author: Richard A. Muller, Berkeley Earth Project, 2831 Garber annual values provided by Hadley/CRU and NOAA. NASA GISS does not temperatures in a city are o ! "le. As is standard practice for the existing climate analysis groups, St., Berkeley CA 94705, USA, Tel: 510 735 6877; E-mail: [email protected] e NOAA group does not perform a speci groups: the National Oceanic and Atmospheric Administration London was the in their most recent analysis, GHCN-M version 3. seasonality was removed from each time series prior to averaging computation, and the most recent available NOAA uncertainty for land-only "ect describes the observation that automated pair wise comparison proce (NOAA), the National Aeronautics and Space Administration Received: January 22, 2013 Accepted: February 27, 2013 Published: March data terminates in the late 1990s [5]. then many cities have$rst urban been identi heat#en islandhigher to than be documentedin its rural surroundings. [ in order to better isolate the small long-term trends from the large 05, 2013 for documented and undocumented changes in station records, ]. According to the well-known example is Tokyo where the temperature has risen much Goddard Institute for Space Studies (NASA GISS),1-8 and the Climatic annual cycle. For this purpose each record was adjusted by removing $c urban adjustment Research Unit of the University of East Anglia (CRU); the analyses of more rapidly in the city than in nearby rural areas: Fujibe estimates cycles with 1-year periods and higher harmonics; the unadjusted data excess warming of almost 2°C/100yr$ compareded as urban to heat the restislands of Japan [ [ dure to make adjustments these groups share many common features [ All articles published in Geoinformatics & Geostatistics: An Overview are the property of SciTechnol, and is protected by 1] but since !ey use an are also available on the website. ! summary provided by the by the Intergovernmental Panel on Climate International Publisher of Science, copyright laws. Copyright © 2013, SciTechnol, All Rights Reserved. e warming of Tokyo is dramatic when compared to a global average w of land- Technology and Medicine as seen in 2-5]. A Change (IPCC), the mean global surface temperature (including land In order to minimize statistical uncertainties, we developed a !e Matlab program that implements di" Figure and oceans) has increased 0.64 ± 0.13°C from 1956 to 2005 at 95% computer program we call Berkeley Average that could take advantage erences between 1. !urban and rural areas, including absorption of ]. !e IPCC did not provide a similar revie sunlight, increased heate storage UHI e of arti "dence [9 of all 179,928 record fragments. "ect can be attributed to many physical con re-radiation by buildings, absence of plant transpiration, di Berkeley Average is available on www.BerkeleyEarth.org. 6] e average, the surface of the Earth was divided in air circulation, and other phenomena [ Published: *Corresponding author: To perform th $ December 02, 2012 cial surfaces, obstruction of Accepted: Urban areas are heavily overrepresented in the siting of September 24, 2012 temperature stations: less than 1% of the globe is urban but 27% of Received: 7]. "erences December 07, 2012 *Corresponding author: All articles published in Geoinformatics & Geostatistics: An Overview are the property of SciTechnol, and is protected by copyright laws. “Copyright © 2012, SciTechnol, All Rights Reserved. Received:
International Publisher of Science, 14, 2013 February 13, 2013 Technology and Medicine Accepted: March 11, 2013 Published: 1 International Publisher of Science, March Earth Atmospheric Land Surface Temperature Technology and Medicine All articles published in Geoinformatics & Geostatistics: FigureAn Overview 1: are the property of SciTechnol, and is protected by copyright laws. Copyright © 2013, SciTechnol, All Rightsto Reserved.a global Annualland average running for mean 1900-2010. of monthly temperatures at Tokyo compared 1 2 and Station Quality in the Contiguous United
3
4 States Decadal Variations in the Global Atmospheric Land Temperatures2, 1 4 5 , Donald Groom 1,2,3, Judith Curry 3 ! , 2 1,2, Robert Rohde 6 Richard A. Muller 1,2 submitted to 1,2, Saul Perlmutter 3 5 GeoInformatics and GeoStatistics Robert Jacobsen , Jonathan Wurtele 7 1,2 4 , Charlotte Wickham Arthur Rosenfeld 8 5 Richard A. Muller 1,2,3 9 , Jonathan Wurtele 6 Perlmutter2,3 2,3 , Arthur , Robert Rohde 1 Rosenfeld2,3 10 , Judith Curry , Robert Jacobsen 4 2,3 , Donald Groom , Saul Richard Muller 11 3 18 April 2013 Steven Mosher , Charlotte Wickham 1 5, 12 Corresponding Author Address:
13 Richard A. Muller
14 Berkeley Earth Surface Temperature 15 2831 Garber St. Project
16 Berkeley CA 94705
17 email: [email protected]
18
2Lawrence Berkeley
Group, Berkeley Earth Surface 3Novim 1 Dept. of Physics, University of California, Berkeley CA 94720; 4Georgia Institute ept. of National Laboratory, Berkeley CA 94720;
Temperature Project, 211 Rametto5 Road,University Santa of Barbara California, CA 92104Berkeley, ; now at D
of Technology, Atlanta GA 30332;
Statistics, Oregon State University. Correspondence for all authors should be sent to
The Berkeley Earth Surface Temperature Project, 2831 Garber St., Berkeley CA 94705. Converted Skeptic