Climate field reconstruction (CFR) methods; MBH 1998 and 1999 Hockey stick graph

1 climate field reconstruction (cfr) methods; mbh 1998 , 1999

1.1 mann, bradley , hughes 1998

1.1.1 publicity , controversy on publication of mbh98


1.2 pollack, huang , shen, jones et al. 1998
1.3 mann, bradley , hughes 1999

1.3.1 critique , independent reconstructions


1.4 ipcc third assessment report, 2001

climate field reconstruction (cfr) methods; mbh 1998 , 1999

variations on composite plus scale (cps) method continued used produce hemispheric or global mean temperature reconstructions. 1998 complemented climate field reconstruction (cfr) methods show how climate patterns had developed on large spatial areas, making reconstruction useful investigating natural variability , long term oscillations comparisons patterns produced climate models. cfr method made more use of climate information embedded in remote proxies, more dependent cps on assumptions relationships between proxy indicators , large-scale climate patterns remained stable on time.

related rigorous statistical methods had been developed tree ring data, harold c. fritts publishing 1991 study , 1991 book showing methodology , examples of how produces maps showing climate developments in north america on time. these methods had been used regional reconstructions of temperatures, , other aspects such rainfall.

as part of phd research, michael e. mann worked seismologist jeffrey park on developing statistical techniques find long term oscillations of natural variability in instrumental temperature record of global surface temperatures on last 140 years; mann & park 1993 showed patterns relating el niño–southern oscillation, , mann & park 1994 found later termed atlantic multidecadal oscillation. teamed raymond s. bradley use these techniques on dataset bradley & jones 1993 study aim of finding long term oscillations of natural variability in global climate. resulting reconstruction went 1400, , published in november mann, park & bradley 1995. able detect multiple proxies varying in coherent oscillatory way, indicating both multidecadal pattern in north atlantic , longer term oscillation of 250 years in surrounding region. study did not calibrate these proxy patterns against quantitative temperature scale, , new statistical approach needed find how related surface temperatures in order reconstruct past temperature patterns.

mann, bradley , hughes 1998

for postdoctoral research mann joined bradley , tree ring specialist malcolm k. hughes develop new statistical approach reconstruct underlying spatial patterns of temperature variation combining diverse datasets of proxy information covering different periods across globe, including rich resource of tree ring networks areas , sparser proxies such lake sediments, ice cores , corals, historical records.

their global reconstruction major breakthrough in evaluation of past climate dynamics, , first eigenvector-based climate field reconstruction (cfr) incorporating multiple climate proxy data sets of different types , lengths high-resolution global reconstruction. relate data measured temperatures, used principal component analysis (pca) find leading patterns, or principal components, of instrumental temperature records during calibration period 1902 1980. method based on separate multiple regressions between each proxy record (or summary) , of leading principal components of instrumental record. least squares simultaneous solution of these multiple regressions used covariance between proxy records. results used reconstruct large-scale patterns on time in spatial field of interest (defined empirical orthogonal functions, or eofs) using both local relationships of proxies climate , distant climate teleconnections. temperature records 50 years prior 1902 analysed using pca important step of validation calculations, showed reconstructions statistically meaningful, or skillful.

a balance required on whole globe, of proxy data came tree rings in northern mid latitudes, largely in dense proxy networks. since using of large numbers of tree ring records in have overwhelmed sparse proxies polar regions , tropics, used principal component analysis (pca) produce pc summaries representing these large datasets, , treated each summary proxy record in cfr analysis. networks represented in way included north american tree ring network (noamer) , eurasia.

the primary aim of cfr methods provide spatially resolved reconstructions essential coherent geophysical understanding of how parts of climate system varied , responded radiative forcing, hemispheric averages secondary product. cfr method used reconstruct northern hemisphere mean temperatures, , results closely resembled earlier cps reconstructions including bradley & jones 1993. mann describes least scientifically interesting thing rich spatial patterns, aspect got attention. original draft ended in 1980 reconstructions went far, anonymous peer reviewer of paper suggested curve of instrumental temperature records should shown present include considerable warming had taken place between 1980 , 1998.

the mann, bradley & hughes 1998 (mbh98) multiproxy study on global-scale temperature patterns , climate forcing on past 6 centuries submitted journal nature on 9 may 1997, accepted on 27 february 1998 , published on 23 april 1998. paper announced new statistical approach find patterns of climate change in both time , global distribution, building on previous multiproxy reconstructions. authors concluded northern hemisphere mean annual temperatures 3 of past 8 years warmer other year since (at least) ad1400 , , estimated empirically greenhouse gases had become dominant climate forcing during 20th century. in review in same issue, gabriele c. hegerl described method quite original , promising , verify model estimates of natural climate fluctuations , important step towards reconstructing space–time records of historical temperature patterns .

publicity , controversy on publication of mbh98

release of paper on 22 april 1998 given exceptional media coverage, including questioning whether proved human influences responsible global warming. mann agree highly suggestive of inference. said our conclusion warming of past few decades appears closely tied emission of greenhouse gases humans , not of natural factors . proxy data inherently imprecise, , mann said have error bars. sizable 1 gets farther in time, , there reasonable uncertainty in given year. there quite bit of work done in reducing these uncertainties. climatologist tom wigley welcomed progress made in study, doubted if proxy data ever wholly convincing in detecting human contribution changing climate.

phil jones of uea climatic research unit told new york times doubtful adding 150-year thermometer record extend proxy reconstruction, , compared putting apples , oranges; mann et al. said used comparison thermometer record check recent proxy data valid. jones thought study provide important comparisons findings of climate modeling, showed pretty reasonable fit proxy evidence. commentary on mbh98 jones published in science on 24 april 1998. noted used available long term proxy climate series, , if new multivariate method of relating these series instrumental data paper claims, should statistically reliable. discussed of difficulties, , emphasised each paleoclimatic discipline has come terms own limitations , must unreservedly admit problems, warts , all.

the study disputed contrarian pat michaels claim of warming took place between 1920 , 1935, before increased human greenhouse gas emissions. george c. marshall institute alleged mbh98 deceptive in going 1400, , not covering medieval warm period predated industrial greenhouse gas emissions. same criticisms made willie , sallie baliunas.

pollack, huang , shen, jones et al. 1998

in october 1998 borehole reconstruction published pollack, huang , shen gave independent support conclusion 20th century warmth exceptional past 500 years.

jones, keith briffa, tim p. barnett , simon tett had independently produced composite plus scale (cps) reconstruction extending thousand years, comparing tree ring, coral layer, , glacial proxy records, not estimating uncertainties. jones et al. 1998 submitted holocene on 16 october 1997; revised manuscript accepted on 3 february , published in may 1998. bradley recalls, mann s initial view there little information , uncertainty go far, bradley said why don t try use same approach used in nature, , see if push bit further? within few weeks, mann responded surprise, there amount of skill. can something, although there large uncertainties.

mann, bradley , hughes 1999

in considering 1998 jones et al. reconstruction went thousand years, mann, bradley , hughes reviewed own research , reexamined 24 proxy records extended before 1400. mann carried out series of statistical sensitivity tests, removing each proxy in turn see effect removal had on result. found proxies critical reliability of reconstruction, particularly 1 tree ring dataset collected gordon jacoby , rosanne d arrigo in part of north america bradley s earlier research had identified key region. dataset extended 1400, , though proxy dataset same region (in international tree-ring data bank) went further , should have given reliable proxies earlier periods, validation tests supported reconstruction after 1400. find out why, mann compared 2 datasets , found tracked each other closely 1400 1800, diverged until around 1900 when again tracked each other. found reason in co2 fertilisation effect affecting tree rings identified graybill , idso, effect ending once co2 levels had increased point warmth again became key factor controlling tree growth @ high altitude. mann used comparisons other tree ring data region produce corrected version of dataset. reconstruction using corrected dataset passed validation tests extended period, cautious increased uncertainties.

the mann, bradley , hughes reconstruction covering 1,000 years (mbh99) submitted in october 1998 geophysical research letters published in march 1999 cautious title northern hemisphere temperatures during past millennium: inferences, uncertainties, , limitations emphasise increasing uncertainty involved in reconstructions of period before 1400 when fewer proxies available. university of massachusetts amherst news release dated 3 march 1999 announced publication in 15 march issue of geophysical research letters, suggesting 1990s warmest decade of millennium, 1998 warmest year far. bradley quoted saying temperatures in latter half of 20th century unprecedented , while mann said go farther in time, data becomes sketchier. 1 can’t quite pin things down well, but, our results reveal significant changes have occurred, , temperatures in latter 20th century have been exceptionally warm compared preceding 900 years. though substantial uncertainties exist in estimates, these nonetheless startling revelations. while reconstruction supported theories of relatively warm medieval period, hughes said warmer intervals in reconstruction pale in comparison mid-to-late 20th-century temperatures. new york times report had colored version of graph, distinguishing instrumental record proxy evidence , emphasising increasing range of possible error in earlier times, mbh said preclude, yet, definitive conclusions climate before 1400.

the reconstruction found significant variability around long-term cooling trend of –0.02 °c per century, expected orbital forcing, interrupted in 20th century rapid warming stood out whole period, 1990s warmest decade, , 1998 warmest year, @ moderately high levels of confidence. illustrated time series line graph figure 2(a) showed reconstruction ad 1000 1980 thin line, wavering around thicker dark 40-year smoothed line. curve followed downward trend (shown thin dot-dashed line) medieval warm period (about warm 1950s) down cooler little ice age before rising sharply in 20th century. thermometer data shown dotted line overlapped reconstruction calibration period 1902 1980, continued sharply 1998. shaded area showed uncertainties 2 standard error limits, in medieval times rising high recent temperatures. when mann gave talk study national oceanic , atmospheric administration s geophysical fluid dynamics laboratory, jerry mahlman nicknamed graph hockey stick , slow cooling trend stick , , anomalous 20th century warming blade .

critique , independent reconstructions

briffa , tim osborn critically examined mbh99 in may 1999 detailed study of uncertainties of various proxies. raised questions later adopted critics of mann s work, including point bristlecone pines western u.s. have been affected pollution such rising co2 levels temperature. temperature curve supported other studies, of these shared limited dated proxy evidence available, , few independent. uncertainties in earlier times rose high in reconstruction @ 1980, did not reach temperatures of later thermometer data. concluded although 20th century warmest of millennium, amount of anthropogenic warming remains uncertain.

with work progressing on next ipcc report, chris folland told researchers on 22 september 1999 figure showing temperature changes on millennium clear favourite policy makers summary . 2 graphs competed: jones et al. (1998) , mbh99. in november, jones produced simplified figure cover of short annual world meteorological organization report, lacks status of more important ipcc reports. 2 fifty-year smoothed curves going 1000 shown, mbh99 , jones et al. (1998), third curve 1400 briffa s new paper, combined modern temperature data bringing lines 1999: in 2010 lack of clarity change of data criticised misleading.

briffa s paper published in january 2000 issue of quaternary science reviews showed unusual warmth of last century, cautioned impact of human activities on tree growth made subtly difficult isolate clear climate message. in february 2000 thomas j. crowley , thomas s. lowery s reconstruction incorporated data not used previously. reached conclusion peak medieval warmth occurred during 2 or 3 short periods of 20 30 years, temperatures around 1950s levels, refuting claims 20th century warming not unusual. analysis crowley published in july 2000 compared simulations energy balance climate model reconstructed mean annual temperatures mbh99 , crowley & lowery (2000). while earlier reconstructed temperature variations consistent volcanic , solar irradiation changes plus residual variability, large 20th-century warming closely agreed predicted effects of greenhouse gas emissions.

reviewing twenty years of progress in palaeoclimatology, jones noted reconstructions jones et al. (1998), mbh99, briffa (2000) , crowley & lowery (2000) showing agreement using different methods, cautioned use of many of same proxy series meant not independent, , more work needed.

ipcc third assessment report, 2001

ipcc wg1 co-chair sir john t. houghton showing ipcc fig. 2.20 hockey stick graph @ climate conference in 2005

the working group 1 (wg1) part of ipcc third assessment report (tar) included subsection on multi-proxy synthesis of recent temperature change. noted 5 earlier large-scale palaeoclimate reconstructions, discussed mann, bradley & hughes 1998 reconstruction going 1400 ad , extension 1000 ad in mann, bradley & hughes 1999 (mbh99), while emphasising substantial uncertainties in earlier period. mbh99 conclusion 1990s have been warmest decade, , 1998 warmest year, of past millennium in northern hemisphere, defined 66-90% chance , supported reconstructions crowley & lowery 2000 , jones et al. 1998 using different data , methods. pollack, huang & shen 1998 reconstruction covering past 500 years gave independent support conclusion, compared against independent (extra-tropical, warm-season) tree-ring density nh temperature reconstruction of briffa 2000.

its figure 2.21 showed smoothed curves mbh99, jones et al. , briffa reconstructions, modern thermometer data red line , grey shaded 95% confidence range mbh99. above it, figure 2.20 adapted mbh99. figure 5 in wg1 technical summary b (as shown right) repeated figure without linear trend line declining ad 1000 1850.

this iconic graph adapted mbh99 featured prominently in wg1 summary policymakers under graph of instrumental temperature record past 140 years. text stated that, in northern hemisphere, 1990s warmest decade , 1998 warmest year in past 1,000 years. versions of these graphs featured less prominently in short synthesis report summary policymakers, included sentence stating increase in surface temperature on 20th century northern hemisphere have been greater other century in last thousand years , , synthesis report - questions.

the working group 1 scientific basis report agreed unanimously member government representatives in january 2001 @ meeting held in shanghai, china. large poster of ipcc illustration based on mbh99 graph formed backdrop when sir john t. houghton, co-chair of working group, presented report in announcement shown on television, leading wide publicity.