An Interpretation of the Origins of the 2012 Central Great Plains Drought


Published on

An assessment report which describes the morphology of the 2012 summer U.S. central Great Plains drought, placing the event into a historical context, and providing a
diagnosis of its proximate and underlying causes.

Published in: News & Politics, Technology
  • Be the first to comment

  • Be the first to like this

No Downloads
Total views
On SlideShare
From Embeds
Number of Embeds
Embeds 0
No embeds

No notes for slide

An Interpretation of the Origins of the 2012 Central Great Plains Drought

  1. 1. An Interpretation of theOrigins of the 2012 CentralGreat Plains Drought Assessment Report NOAA Drought Task Force Narrative Team Lead: Martin Hoerling Co-Leads: Siegfried Schubert & Kingtse Mo 20 March 2013
  2. 2. Composed by the “Narrative Team”1 of the NOAA Drought Task Force2 in partnership with theNational Integrated Drought Information System (NIDIS)Lead: M. Hoerling Co-Leads: S. Schubert, and K. Mo1 A. AghaKouchak , H. Berbery, J. Dong, M. Hoerling, A. Kumar, V. Lakshmi, R. Leung, J. Li, X. Liang, L. Luo, B.Lyon, D. Miskus, K. Mo, X. Quan, S. Schubert, R. Seager, S. Sorooshian, H. Wang, Y. Xia, N. Zeng2 Organized by the NOAA Modeling, Analysis, Predictions and Projections Program (MAPP) of Office ofOceanic and Atmsopheric Research/Climate Program OfficePublished March 2013This report is available online at: Credits: ‘Montana Farmland’ (cover), courtesy BLM; ‘Cracked Earth’ (cover), courtesy Wikimedia Commons, by TomasCastelazo; ‘Small Corncob’ by Christina Reed, USDA; ‘Dried Sunflower Field’, ‘Sunrise Over Cornfield’, ‘Mountain Field’ and ‘FlowerSeeds’ by Rolf Reiser (© 2013 Rolf Reiser); ‘Field of Dry Soybeans’ by Scott Bauer, USDA/ARS; ‘Cattle on the Range’ and ‘StuntedCorn Crop’ by Tim McCabe, USDA/NRCS; ‘Native Grasses’ by Lynn Betts, USDA/NRCS; ‘Bodega Bay’ by Barb DeLuisi, NOAA; ‘DryRiver Bed’, courtesy Wikimedia Commons, by Gin E.; ‘Agricultural engineers inspecting soil cracks’ by Scott Bauer, USDA/ARS).
  3. 3. An Interpretation of the Origins of the 2012 Central Great Plains Drought Assessment Report Contents Executive Summary 01 The Drought’s Morphology 03 The 2012 Drought’s Impact 07 The Historic 2012 Drought and its Antecedent Conditions 11 Proximate Causes for the 2012 Drought 17 Underlying Causes for the 2012 Drought 23 Prediction for the Summer 2012 33 Summary Comments and Additional Questions 37 Acknowledgments 40 Contributing Authors 41 Additional Reading 42
  4. 4. This report describes the morphology of the 2012summer U.S. central Great Plains drought, placingthe event into a historical context, and providing adiagnosis of its proximate and underlying causes.
  5. 5. Executive SummaryThis report describes the morphology of the 2012 produced a substantial summertime dry signal oversummer U.S. central Great Plains drought, placing the central Great Plains during 2012. Official season-the event into a historical context, and providing al forecasts issued in April 2012 did not anticipatea diagnosis of its proximate and underlying caus- this widespread severe drought. Above normales. Precipitation deficits for the period May-August temperatures were, however, anticipated in climate2012 averaged over the central Great Plains were models, though not the extreme heat wave thatthe most severe in the instrumental record since occurred and which was driven primarily by the ab-1895, eclipsing the driest summers of 1934 and sence of rain. Our integrative assessment of histori-1936 that occurred during the height of the Dust cal data, climate simulations, and seasonal forecastsBowl. The drought developed suddenly, with near thus paints a picture of an extreme drought eventnormal antecedent precipitation during winter and that may not have had extreme forcing as its cause.spring over the Great Plains giving little forewarning The interpretation is of an event resulting largelyof the subsequent failed rains. The event did not from internal atmospheric variability having limitedappear to be just a progression or a continuation of long lead predictability. This is a characteristic quitethe prior year’s record drought event that occurred different from that of the prior year’s southern Plainsover the southern Great Plains, but appeared to be drought that spanned October 2010-August 2011,a discrete extreme event that developed in situ over and for which appreciable early warning capabilitythe central U.S. The proximate cause for the drought existed owing to a strong sensitivity of that region towas principally a reduction in atmospheric mois- La Niña conditions. The outcome and value of suchture transport into the Great Plains from the Gulf of an assessment, beyond scientific merits of betterMexico that normally provides the major source of understanding what produced the 2012 drought,water vapor for the region in summer. Processes that is two-fold. It clarifies whether such drought couldwould provide air mass lift and condensation during have been anticipated, and it suggests investmentsthe wet season over the Great Plains were mostly that may lead to better guidance on mitigatingabsent, including a lack of frontal cyclones in the effects of future drought. Assessments of this sortearly stages of drought development followed by a help inform scientific pathways for creating moresuppression of deep convection in mid-late summer actionable information for stakeholders that areowing to large scale subsidence and atmospheric vulnerable to drought-related hazards, even whenstabilization. forecast skill is expected to be low.Climate simulations and empirical analysis suggestthat neither the effects of ocean surface tempera-tures nor changes in greenhouse gas concentrations 1
  6. 6. Absent were the usual abundance ofslow soaking rain systems and eveningthunderstorms that characterize GreatPlains climate from May through August,and as a result surface moisture conditionsgreatly deteriorated.
  7. 7. The Drought’s MorphologyDrought conditions developed rapidly over thecentral Great Plains during late spring 2012, and US Drought Monitorintensified in summer. The tracking of droughtseverity via the U.S. Drought Monitor revealedextreme drought to be initially confined to thesouthern Plains in November 2011, a remnant ofthe record setting drought of Texas and Oklahomathat began in late 2010 (Fig. 1, top left). In Fall 2011,only a narrow swath of moderate drought extendednorthward thru eastern Kansas to Minnesota, andno extreme drought existed over the central Plains.While some concerns existed that the southernPlains drought might expand northward into thegrain belt, little indications to this effect wereinitially observed. Indeed, much of the centralGreat Plains became drought-free by May 2012 Figure 1. U.S. Drought Monitor maps for select periods during 2011- 12. First color level (D0, yellow) denotes abnormally dry, and last(Fig. 1, top right), and considerable recovery was color level (D4, dark red) denotes exceptional drought. See http://even occurring over the Southern Plains. There for more details.were also concerns about the possible effects ofunusually high surface temperatures over the Great of extreme drought. Conditions became comparablePlains during March on soil moisture conditions. to those experienced a quarter-century earlierNonetheless, estimates of the monthly averaged during 1988 by a previous generation of inhabitants,column soil moisture1 over the contiguous US for and the combination of rainfall deficits and highApril did not reveal extreme soil moisture deficits temperatures even rivaled those observed by theirover the central Great Plains, with conditions forebears during the Dust Bowl.resembling the map of the 1 May U.S. Drought Consistent with the Drought Monitor maps, theMonitor. But then the expected rainy season Palmer Drought Severity Index (PDSI; Palmer 1965)failed. Absent were the usual abundance of slow for August 2012 (Fig, 2. left) identifies the coresoaking rain systems and evening thunderstorms region of the drought to be the central Plains region,that characterize Great Plains climate from May with the most extreme moisture deficits occurringthrough August, and as a result surface moisture over the western Plains. A central U.S. epicenterconditions greatly deteriorated. By early September for the drought is also affirmed by the May-August(Fig. 1, bottom left), estimates of surface moisture standardized rainfall deficits (Fig. 2, middle) with -2conditions revealed that over three-quarters of standardized departures being widespread fromthe contiguous U.S. was experiencing at least Colorado to Missouri.abnormally dry conditions with nearly half of theregion (the central Plains in particular) experiencing Much of the dry region also experienced hotsevere-unprecedented drought. In this way, the temperatures (Fig. 2, right). The combination ofcomfort of having entered late spring virtually low rainfall and high temperatures is typically seen_________________drought-free was abruptly replaced by the distress during summertime droughts over the central U.S.1 Monthly averaged column soil moisture is estimated routinely at CPC using a one-layer “bucket” modeldriven by monthly precipitation and temperature. See Huang et al. (1996) in Additional Reading. 3
  8. 8. Interpretation of Origins of 2012 Central Great Plains Drought Surface Anomalies May-Aug 2012 Versus 1895-2000 Longterm Average NOAA/ESRL PSD and CIRES-CDC NOAA/ESRL PSD and CIRES-CDC NOAA/ESRL PSD and CIRES-CDC -5.0 -3.0 -1.0 1.0 3.0 5.0 -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 Figure 2. The Palmer Drought Severity Index (PDSI) for August 2012 (left), the standardized precipitation departures for May-August 2012 (middle), and the surface temperature departures for May-August 2012 (°C, right). Data source is the NOAA U.S. Climate Divisions. The historical relationship between rainfall and drought, accumulated above normal rainfall for the temperature deficits (Fig. 3) suggests, however, that prior 6-month period through summer 2012. This 2012 could have been appreciably warmer (perhaps greatly improved their soil moisture balance, and by ~1°C) given the severity of rainfall deficits alone. the Drought Monitor indicated northeast Texas to The scatter plot shows that 2012 was the driest summer in the historical record Central US (-34 mm departure), though the temperature May-Aug PPT vs. May-Aug Tmp anomaly of +2°C was exceeded by two prior summers -- 1934 and 1936. Indeed, although the 2012 summer experienced less rainfall over the central Great Plains than in either 1934 or 1936, those years were about 0.5°C warmer. Temperature (oC) Daily rainfall time series from observations taken at weather stations across the Great Plains (Fig. 4) illustrate the timing of drought onset. Consistent with the Drought Monitor tracking, the event commenced suddenly in May. Further, the core period of the drought appears to be May-August 2012. The daily time series reveal that after a period of near to above normal winter and early spring precipitation at most stations over the central Great Plains, rains abruptly failed in May. For instance, Precipitation (mm) there were virtually no rainy days at Cedar Rapids, Figure 3. The observed relationship between May-August IA during May. Likewise, July saw no measurable averaged rainfall departures (mm, x-axis) and surface temperature rain at Omaha, NE. Both are climatologically wet departures (°C, y-axis) over the U.S. central Great Plains. Reference months, so the lack of any rain was a severe loss. period is 1895-2012. The 2012 departures are -34mm, and +2.1°C, and shown by the red asterisk. Dashed line is the linear relation Likewise, the western Plains sites of Goodland, KS between temperature and precipitation variability. Note that for and Cheyenne, WY saw only infrequent rains of light extreme dry conditions, temperatures are appreciably warmer than predicted by this linear fit. May-August departures are intensity during July and August. By contrast, Dallas- averages over the multi-state region (WY, CO, NE, KS, MO, IA). Data Fort Worth, which was near the center of the 2011 source is the NOAA U.S. Climate Divisions.4
  9. 9. Interpretation of Origins of 2012 Central Great Plains Droughtbe drought-free in May 2012. Oklahoma City also curves in Fig. 4, reveals the period from Septembershowed strong signs of recovery from the 2011 thru February to be normally dry over central Plains.drought with above average rains falling through Thus, it is unlikely that sufficient precipitation couldMay 2012, but skies cleared and June through July materialize in that period to redress the severewas virtually rain-free. deficits accumulated during the normally wet season of late spring/summer. In this sense, whileAs of this writing, drought conditions that in hindsight we might speak with confidence aboutestablished by the end of summer 2012 remain the time of drought onset, judgment on its durationmostly in place. Neither the termination nor must await the outcome of the 2013 wet season.the duration of this drought is yet known. Theclimatological rainfall, illustrated by the smooth Daily Precipitation 1 Jan 2012 – 31 Dec 2012 Cheyenne, WY Omaha, NE Goodland, KS Cedar Rapdis, IA Oklahoma City, OK Dallas–Ft. Worth, TX Figure 4. Daily precipitation time series during 2012 for indicated stations. For each station, top panels show the climatological precipitation (smooth curve), the actual 2012 precipitation, and their difference (color shading; brown denotes a deficit, green a surplus). Lower panels show the occurrences of daily precipitation events. Data source is NOAA Climate Prediction Center. 5
  10. 10. It is expected that water supply reductions in thesemi-arid western portions of the drought wherereservoir storage was depleted by lack of rainswill also have long-term impacts, as will livestockhealth and its long term effect on herd stocks.
  11. 11. The 2012 Drought’s ImpactThe suggestion that 2012 was a “flash drought”, at An additional comment regarding corn yieldsleast concerning its rapid onset over the central during 2012 helps to illustrate the severity of thePlains, is supported by the above-mentioned drought’s impact. The USDA indicated that thetime series of daily rainfall and the sequence of 2012 yield of about 123 bushels per acre was thedrought monitor maps. Impacts also emerged quite lowest since 1995. But even that confirmation ofswiftly. Loss estimates by the end of July 2012, greatly compromised production fails to convey thebefore drought severity peaked, were $12B (www. severity of crop failure. Fig. 5 shows the time ). It of U.S. corn yield (per acre) since 1866, the mostremains to be seen if the economic effects of the prominent feature of which is the growth in yield2012 drought will approach prior events, including since about WWII as a consequence of improvedthe 1988 drought that inflicted $78 billion in losses agricultural practices and more productive andand the 1980 event that caused $56 billion in losses heartier strains of seed.   However, 2012 corn yield(adjusted for inflation to 2012 dollars) (www.ncdc. fell strikingly below the recent trend line. The ). Broad sectors were crop yield deficit and the implied climatic impactaffected, and continue to be affected, by the 2012 was a historic event. Figure 6 shows the annual yielddrought. Notable for the swiftness of impacts was departures (computed relative to appropriate trendthe reduction in crop yields caused by lack of timely lines).  In terms of absolute loss in bushels of cornrains, as discussed further below. Also, curtailment production, no single year since 1866 experiencedof commerce on major river systems occurred owing so large a curtailment as occurred during reduced water flow, a situation that continues The 43 bushel/acre productivity loss, though onlymany months after the drought. It is expected that 26% less than expected by USDA, equates to thewater supply reductions in the semi-arid western total U.S. productivity of 1960. If measured as a %portions of the drought where reservoir storage was deficit as is shown in Fig. 6, then 2012 was about thedepleted by lack of rains will also have long-term second most severe curtailment of corn productionimpacts, as will livestock health and its long term on record, eclipsed only by 1901, and comparable toeffect on herd stocks. the decline in 1936.Preliminary USDA estimates of farm and food It is from such historical data that the USDA offeredimpacts of the 2012 drought (www.nass.usda. its initial expectation, in spring 2012, that annualgov) indicate corn yield (per acre of planted crop) corn yield would be about 166 bushels per acre.was about 123 bushels. This is 26% below the 166 That outlook was based mainly on extrapolatingbushel yield expectation that the USDA had at the the recent trend in corn yields. This is a reasonablecommencement of the growing season. Likewise, prediction given that year-to-year variationssoybean yields were estimated at 39 bushels, 10% are mostly small relative to the trend “signal” ofbelow the early season projection of 44 bushels. This relentlessly improved yields. Of course, thesewas the lowest soybean yield since 2003. Owing to variations—relative to trend—are mostly the resultthe late onset of drought conditions over the Central of interannual climate variability. The questionPlains, wheat production was not significantly is thus whether this drought could have beenimpacted. Drought conditions adversely impacted anticipated, and if actionable prediction of climatepasture growth and range land quality, which when impacts on crop yield might have been rendered.combined with elevated corn and soymeal prices,adversely affected livestock and draft capacity, asituation that will unfold over several years( 7
  12. 12. Interpretation of Origins of 2012 Central Great Plains Drought Figure 5. Historical U.S. corn yields from 1866 to 2012 (bushels/acre). Linear fit to different segments of the time series shown in solid lines, including regression formula. The 2012 yield is plotted in the blue circle, based on August estimates. Subsequent data revised the 2012 yield downward to about 123 bushels. Data source is USDA. Figure 6. Historical U.S. corn yield deficits from 1866 to 2012 (bushels/acre). Deficits computed relative to the trend lines of Fig. 1. All years having greater than a 20% deficits are highlighted and shown with red circles. The 2012 yield deficit is plotted in the large red circle, based on August estimates (the circle sizes are not proportional to deficit magnitudes). Subsequent data through the end of the growing season revised the 2012 yield deficit8 downward to about -26%. Data source is USDA.
  13. 13. The 2012 crop yield deficit and theimplied climatic impact was a historicevent. In terms of absolute loss inbushels of corn production, no singleyear since 1866 experienced so large acurtailment as occurred during 2012.
  14. 14. Summertime Great Plains rainfall has been in an upward trend since the early 20th Century, and the last major drought occurred 25 years ago in 1988. The 2012 drought thus was a —climate surprise—, and would not have been anticipated from simple considerations of central U.S. rainfall behavior in the recent past.10
  15. 15. The Historic 2012 Droughtand its AntecedentConditionsBy measures of rainfall deficits, the summer of 2012 Central USwas an unprecedented year. Fig. 7 shows the 1895-2012 time series of May-August rainfall departuresaveraged over the multi-state region (WY, CO, NE, KS,MO, IA) that experienced the most severe drought Millimetersconditions in 2012. The deficit in rainfall in 2012was -34.2 mm, which was about 53% of the region’slong-term mean rainfall (73.5 mm). This deficit brokethe record of -28.4 mm observed in 1934, and corre-sponds to a 2.7 standardized deficit.The 2012 event would not have been anticipated Degrees Celisusfrom simple considerations of central U.S. rainfallbehavior in the recent past. The 1930s droughts layin distant memory, and though not forgotten, havebeen suggested to have resulted from unique con-ditions of that era. These included remote effects oftropical sea surface temperatures, land use practicesand the potential feedbacks that abundant soil-relat-ed aerosols may have exerted on rainfall. An import- Figure 7. 1895-2012 time series of May-August central Great Plainsant role for random atmospheric internal variability rainfall departures (mm, top) and surface air temperature departures (°C, bottom). Reference period is 1895-2011. Black curve is a 9-pointhas also been proposed. Summer rainfall has shown Gaussian filter. The area is comprised of the 6-State region of WY, CO,a general upward trend in the recent period, and the NE, KS, MO, and IA. Data source is the NOAA U.S. Climate Divisions.last 2 decades were noted more by their abundantsummer rainfall, than by severe deficits. The 2012 rived product. Depletion of soil moisture associateddrought thus appears to be a climate surprise from with the prior southern Plains drought was espe-such empirical considerations alone. cially evident over Texas and Oklahoma in Fall 2011. Soil conditions were also estimated to be dry overBut did early warning signs exist, for instance in the northern Plains from Fall 2011 thru early springthe sequence of seasonal events that immediate- 2012. By contrast, antecedent spring soil moisturely preceded the 2012 drought? Figure 8 presents over the central Plains regions of Missouri, Kansas,estimates of the seasonal soil moisture anomalies, and Nebraska were mostly near normal. It is likelybased on the CPC one-layer “bucket” land surface that unusually warm early spring temperatures overwater balance model. The derived soil moisture the Plains and upper Midwest dried soils, especial-conditions are estimates for a column of about 1.6 ly in the top layers, though this cannot be readilymeter depth, and though few representative mea- discerned from the column integrated estimatessurements of actual soil moisture are available over that are derived from the CPC one-layer land surfacethe US, validation against in situ soil moisture data model. It is evident, however, that the distinguish-over Illinois has shown realistic variability in the de- ing characteristic of the model-derived soil mois- 11
  16. 16. Interpretation of Origins of 2012 Central Great Plains Drought ture for the U.S. was one of overall dryness by early the central Great Plains during 2012 have exhibited Spring 2012,, whereas the central Great Plains had robust precursors and coherent temporal and spatial near normal soil moisture. evolutions, compositing methods are applied. From the historical time series, the prior driest May-August Figure 9 illustrates the seasonal precipitation anom- periods are identified. The 10 driest years (including alies for the 12-months that preceded May-August 2012), ranked in order of their rainfall deficits, were: 2012. Much of the southern and central Great Plains 2012, 1934, 1936, 1901, 1976, 1913, 1988, 1953, experienced near normal precipitation during the 1911, and 1931. Perhaps not surprisingly, 5 of these period October 2011 thru April 2012. This precipita- (1901, 2012, 1936, 1934, and 1988) also rank among tion significantly improved soil moisture conditions the top 5 years suffering the most severe corn yield over the southern Plains by spring 2012 (see Fig. 8), curtailment. and was responsible for the amelioration of drought severity over this region as indicated by the Drought For these 9 historical cases, averages of precipitation Monitor (see Fig. 1). The question of how soil mois- for the 12 months preceding and the 12-months ture conditions may have affected precipitation is following their peak central Great Plains May-August difficult to assess from the empirical data alone, rainfall deficits are calculated. The lead-lag compos- and it is unclear from this analysis alone what if any ites of precipitation patterns for these cases (ex- affect the dry soil conditions may have had upon the cluding 2012) are shown in Figs. 10 and 11, respec- summer drought intensification. More will be said tively. There is no appreciable dryness in the prior about that when various seasonal forecast systems summer over Texas in this composite (Fig. 10, top are evaluated in section 6. Suffice it to state here that left); suggesting that southern Plains drought such the region of most severe moisture deficits existing as occurred in 2011 is not a necessary condition for over the southern Plains during fall 2011 into win- subsequent central Great Plains drought. There is ter 2012 experienced substantially above normal some indication for prevailing dryness in the an- precipitation during the subsequent winter/spring tecedent conditions across the central Great Plains 2012 period. Precipitation was thus mainly driving as a whole, however. Likewise, the aftermath of a recovery in soil moisture through spring 2012, central Great Plains summer drought also reveals a whereas the antecedent deficiencies in soil moisture tendency for below average precipitation. These dry appeared not to inhibit precipitation processes. signatures are partly related to the fact that several of the individual driest central Plains summers in the There are additional lines of diagnosis from which composite were immersed within dry epochs than one can examine the question of whether anteced- spanned much of the 1930s and also from the late- ent drought over the southern Plains in 2011 may 1940s through the mid-1950s. On average, however, have set in motion a sequence of unavoidable the composite shows no appreciable rainfall anoma- climate events that strongly determined the fate of ly over the central Great Plains in the summer fol- subsequent central Plains summer rainfall. Here the lowing a severe drought (Fig. 11, lower right panel). instrumental record dating to 1895 is examined to In this empirical sense, the composite indicates little probe for historical evidence on how southern Plains basis to expect that central Plains drought would droughts typically evolve, and especially if there necessarily recur during the subsequent summer. is any support to a hypothesis that these have a propensity to spread throughout the Great Plains re- gions as part of a typical life cycle. To address the ex- tent to which droughts of the type that occurred in12
  17. 17. Interpretation of Origins of 2012 Central Great Plains DroughtEstimated 2012 Soil Moisture: MJJ 2011 to May - Aug 2012 Figure 8. U.S. seasonal soil moisture anomalies (mm) during the 12-month period antecedent to the occurrence of dry May- August conditions over the central Great Plains (lower right panel). Soil moisture has been estimated by driving a one-layer bucket water balance model with observations of monthly temperature and precipitation. The data set spans 1948-present, and the method is described in Huang et al. (1996).Observed 2012 PPT Departures: MJJ 2011 to May - Aug 2012 Figure 9. U.S. seasonal precipitation anomalies (mm) during the 12-month period antecedent to the occurrence of dry May- August conditions over the central Great Plains (lower right panel). Note also the prior severe rainfall deficits in summer of 2011 over the southern Great Plains. Data source is the NOAA U.S. Climate Divisions. 13
  18. 18. Interpretation of Origins of 2012 Central Great Plains Drought Historical Composite PPT Departures: MJJ Yr-1 to May - Aug Yr 0 Figure 10. As in Fig. 9, except for the composite U.S. seasonal precipitation anomalies (mm) during the 12-month period antecedent to the occurrence of dry May-August conditions over the central Great Plains. Based on the average of the 9 driest May-August events during 1895-2011, including 1934, 1936, 1901, 1976, 1913, 1988, 1953, 1911, and 1931. Data source is the NOAA U.S. Climate Divisions. Historical Composite PPT Departures: May - Aug Yr 0 to JJA Yr+1 Figure 11. As in Figure 10, except for the composite U.S. seasonal precipitation anomalies (mm) during the 12-month period subsequent to the occurrence of dry May- August conditions over the central Great Plains. Based on the average of the 9 driest May-August events during 1895-2011, including 1934, 1936, 1901, 1976, 1913, 1988, 1953, 1911, and 1931. Data source is the NOAA U.S. Climate Divisions.14
  19. 19. The Central Plains drought of 2012was not a progression or northwardcreeping of the prior year’s SouthernPlains drought event. There were nostrong indicators that an extremedrought event was poised to spreadover the Central Plains in 2012.
  20. 20. As is common with droughts, atmosphericmoisture in both absolute and relativemeasures is typically deficient, and 2012 wasno exception. A second, and often inexorablylinked factor is the absence of processes thatproduce rainfall over the central Plains.
  21. 21. Proximate Causes for the2012 DroughtWhy did the 2012 drought happen the way it did? high 700 hPa specific humidity located in the cen-This is meant as a simple starting query towards tral and western Great Plains (top left) (though thisinterpreting the drought, though recognizing that moisture is also related to the nocturnal low level jetanswers to this question alone may provide lit- in the western Great Plains).tle predictive understanding. As is common withdroughts, atmospheric moisture in both absolute During late spring/summer 2012 the typical north-and relative measures is typically deficient, and 2012 ward 700 hPa meridional wind along the Gulf Coastwas no exception. A second, and often inexorably was much reduced (Fig. 12 bottom right). The sea-linked factor is the absence of processes that pro- sonal mean anomaly of about -1 m/s (anomalousduce rainfall over the central Plains. These include equatorward flow) was 50% of the magnitude of thespringtime low pressure systems and their attending typical northward flow. There was thus an apprecia-warm and cold fronts that act to lift air masses and ble reduction in the typical moisture transport intoproduce widespread rains. During summertime, the the continent. Consistent with this, the summertimekey process involves thunderstorms that normally 700 hPa specific humidity was anomalously low inoccur with considerable frequency and from which the Great Plains (bottom left). Departures of aboutthe majority of precipitation falls in July and August. -0.5 g/kg over the Great Plains were on the orderBoth of these mechanisms were largely absent or of a 10% reduction of climatological water vaporinoperative to considerable degree in 2012 over the content. Of course, the general absence of migratorycentral Great Plains. low pressure systems across the central Plains would have entailed a similar lack of large scale air mass lift-A principal source of water vapor in summer over ing and precipitation, while simultaneously reducingthe central U.S. is the Gulf of Mexico region, with the influx of Gulf moisture.vapor-laden air transported inland and northwardsto the continent’s interior by mean southerly winds. Analysis of relative humidity provides anotherFigure 12 illustrates this latter feature using the long- indication of the extent to which dryness prevailedterm mean 700 hPa meridional (north-south com- in the lower troposphere during summer 2012 overponent) wind (top right) which shows a peak 2 m/s the Great Plains. The top panels of Fig. 13 show themagnitude immediately on the coast of southwest climatological relative humidity at 850 hPa (left) andTexas. This is partly related to mean transport linked 700 hPa (right). Note in particular the 700 hPa axis ofto the clockwise air motion around the subtropical high relative humidity that normally characterizeshigh located over the Atlantic Ocean. The influx is the Great Plains region from northern Texas to Can-also related to the integrated effects of migratory ada (top right). This feature was essentially absentmid-latitude storm systems, especially in the spring- during summer 2012, with departures of -10% run-time when they exhibit a geographically preferred ning from northern Texas to Montana (lower right).cyclogenesis in the lee of the southern Rocky Moun- The relative humidity was even further reduced attains and then track northeastward to the Great 850 hPa with widespread deficits of greater thanLakes. It is in association with the circulation around -10% almost exactly matching the scale of the rain-such storms that Gulf of Mexico moisture is intermit- fall departures (see Fig. 2). It is worth noting that thetently, but strongly, drawn northward. These mean relative humidity reductions at 850 hPa were some-and transient features are thus primarily responsible what greater than one would have surmised fromfor the influx of moisture that maintains the axis of just the fractional change in specific humidity. This 17
  22. 22. Interpretation of Origins of 2012 Central Great Plains Drought May – August 2012 700 hPa Figure 12. (top) Observed climatological May-August 700 hPa specific humidity (left, g/kg) and 700 hPa meridional wind magnitude (right, m/s) (bottom). Anomalous May-August 2012 700 hPa specific humidity (left. g/kg) and anomalous 700 hPa meridional wind magnitude (right, m/s). Data source is the NCEP/NCAR reanalysis, and graphics from the NOAA/ESRL Physical Sciences Division. Anomalies relative to 1981-2010 reference. is because the low troposphere temperatures were summer 2012. This is further affirmed by the month- especially warm, and so the water holding capacity ly 500 hPa height anomalies for May, June, July and increased even while the actual water vapor content August (Fig. 14). In May and June (top panels), a was diminished. zonal ridge of high pressure anomalies inhibited the typical southward push of cold fronts from Canada Water vapor deficiencies alone need not guarantee that often serve to organize widespread rains. July drought, as mechanisms that induce convergence saw a somewhat different pattern, though no less and air mass lift can still operate from time to time to effective in inhibiting rainfall. An intense anticyclone yield precipitation events. But, recall from the station was centered over the northern Plains region, pre- rainfall times series (Fig. 4) that some locations saw venting frontal incursions while also stabilizing the rather remarkable sequences of 30-60 days with- atmosphere and inhibiting deep convection that out precipitation, an indication that rain-producing typically contributes appreciably to mid-summer mechanisms and triggers for ascent were scarce in rainfall totals. The August 500 hPa height pattern,18
  23. 23. Interpretation of Origins of 2012 Central Great Plains Droughtthough also drought producing, was yet different The upper-level circulation broadly favored large-again from both June and July. A deep Ohio Valley scale descent during summer and inhibitedtrough acted to inhibit Gulf of Mexico moisture the normal occurrence of spring storms. Wheninflow (as seen in the seasonal map of 700 hPa me- conditions favorable for rainfall were present,ridional wind anomalies), while subsidence over the the depleted moisture in the low tropospherewestern Great Plains was enhanced on the western limited rainfall amount. Together, these conditionsedge of this low pressure system. Note that this dry conspired to create a 4-month sequence of recordAugust pattern was also a cool pattern for the central rainfall reduction over the central Great eastern Plains, which may account for the fact that The impression is also rendered of a sequence ofthe May-August 2012 mean temperature anomalies unfortunate events. There was considerable monthlywere not greater than would have been surmised variability in the upper level circulation (perhapsgiven the severity of rainfall deficits. belying the impression that such a sustained and May - August 2012 Figure 13. (top) Observed climatological May-August 850 hPa relative humidity (left,%) and 700 hPa relative humidity (right, %) (bottom). Anomalous May-August 2012 850 hPa relative humidity (left. %) and anomalous 700 hPa relative humidity (right, %). Data source is the NCEP/NCAR reanalysis, and graphics from the NOAA/ESRL Physical Sciences Division. Anomalies relative to 1981-2010 reference. 19
  24. 24. Interpretation of Origins of 2012 Central Great Plains Drought extreme drought must have been the consequence intensified through July and August as summertime of some strong sustained forcing), yet each of these convection was inhibited. Since the end of summer, patterns in their own manner squelched rainfall- the normal dry season emerged, and soil moisture inducing processes over the central Plains. And so it conditions remain depleted. As this report is being began, with drought emerging suddenly in May as written, the 2013 rainy season is anxiously awaited. late spring storms avoided the region entirely, and 2012 500 hPa Geopotential Height Departures Figure 14. Observed monthly 500 hPa geopotential height anomalies (m) for May, June, July, and August 2012. Data from the NCEP/NCAR reanalysis, and graphics from the NOAA/ESRL Physical Sciences Division.20
  25. 25. Underlying causes refer to root causes, within achain of factors, that lead to an outcome. Climatescientists are especially interested in identifyingsuch causes because they can entail useful long-lead predictability. The report examines sea surfacetemperature (SST) and sea ice conditions, and alsothe chemical composition of the atmosphere, aspotential underlying causes for the drought over thecentral Plains in summer 2012.
  26. 26. Climate simulations and empirical analysis suggestthat neither ocean surface temperatures nor changesin greenhouse gases induced a substantial reductionin summertime precipitation over the central GreatPlains during 2012. Diagnosis of historical data,climate simulation data, and seasonal forecastspaint a picture of an extreme drought that may nothave had extreme forcing as its cause and that hadlimited long lead predictability.
  27. 27. Underlying Causes for the2012 DroughtWhy did drought occur over the central Great Plains reflects mostly the long-term trend in SSTs (whichduring summer 2012 (and what caused the proxi- have been warming in the latter half-century inmate conditions discussed above)? We have already particular). The effect of this trend on the compos-surmised, from empirical analysis, that the central ite arises because of the inhomogeneous temporalPlains drought was unlikely part of a multi-year sampling of drought events in the historical recorddrought life cycle that began over the southern with only two of the nine prior severe droughts oc-Plains in late 2010 and evolved northward. Here we curring after 1953 (to minimize the influence of thisexplore whether particular forcings, including sea trend, the composite SST anomalies in Fig. 15 weresurface temperature (SST) and sea ice conditions, calculated relative to a 1901-1990 reference thatand also the chemical composition of the atmo- brackets the years of the 9-case sample).sphere, may have contributed to the occurrence of adrought over the central Plains in summer 2012. Nonetheless, several of the prior summer droughts occurred in the immediate aftermath of wintersConcerning SST forcing, it is useful to first examine experiencing cold equatorial Pacific SSTs. Examina-the state of global oceans that attended prior his- tion of a SST index that is used to monitor the occur-torical Great Plains droughts. Figure 15 shows the rences of El Niño (warm) and La Niña (cold) tropicalseasonal SST anomaly composite that is based on Pacific events reveals that the preceding winters of 3the same sample of the 9 prior driest summers used cases (1910/11; 1933/34; 1975/76) were moderate Lato construct the antecedent precipitation maps. Niña events. However, two other severe droughts oc-Though this composite reveals global SSTs to be curred after wintertime El Niño conditions (1930/31;cool overall in all seasons, the magnitudes are weak. 1987/88), while the remaining 4 cases were neutralThe composite SST coolness is less indicative of a with respect to ENSO’s phase.coherent pattern of interannual forcing, but instead Historical Composite SST Departures: MJJ Yr–1 to May–Aug Yr 0 Figure 15. As in Fig. 10, except for the composite seasonal SST anomalies (°C) during the 12-month period antecedent to the occurrence of dry May-August conditions over the central Great Plains. Based on the average of the 9 driest May-August events during 1895-2011, including 1934, 1936, 1901, 1976, 1913, 1988, 1953, 1911, and 1931. Reference period is a shorter 1901-1990 period in order to reduce effects of the long term SST warming trend. Data source is the monthly NOAA Merged Land-Ocean surface temperature analysis (MLOS). 23
  28. 28. Interpretation of Origins of 2012 Central Great Plains Drought Central US May–Aug PPT vs. May–Aug Tmp 1895-2012, N=118 Figure 16. The linear correlation between an index of observed May-August U.S. central Great Plains. summer rainfall (see Fig. 6) and May-August surface temperatures. Period of analysis is 1895-2011. Statistically significant correlations are confined to the central U.S. where there is a strong inverse correlation between summer rainfall and summer land surface temperature. Data source is the monthly NOAA Merged Land-Ocean surface temperature analysis (MLOS). Consistent with the weak evidence for a coherent However, global SSTs have appreciably changed precursor SST condition attending summertime (principally warmed) since the last major central central Plains drought, evidence for a strong simulta- Plains drought of 1988. Shown in Fig. 17 are the SST neous SST effect is not found either. Figure 16 pres- anomaly maps during 2012 (using the same 1901- ents the correlation between the index of central 1990 reference), from which the material difference Great Plains summer precipitation with summertime from the SSTs seen in the 9-case historical composite ocean surface and land surface temperatures for the is obvious. One point of similarity with the histori- entire 1895-2011 period (Fig. 16). The weak posi- cal composites is coolness in the equatorial central tive correlations with ocean surface temperature Pacific in the preceding winter. Otherwise, owing in variability seen over the tropical east Pacific are not part to the warming trend and perhaps also due to statistically significant, nor are most of the correla- low frequency decadal ocean variability, the 2012 tions in other ocean basins significantly different drought occurred in concert with an appreciably from zero. The empirical results thus suggest that warmer ocean in most basins than was the case for SST variations, at least those observed during the any prior historical drought. last century, have likely failed to consistently pro- duce May-August drought occurrences over the Has this overall ocean warming altered the probabil- central Great Plains. This diagnosis of historical data ities for U.S. summertime drought? Recognizing that paints an overall picture in which ocean conditions most of the prior severe Great Plains droughts hap- have not strongly constrained the variations of sum- pened before 1950 when global climate as a whole mertime central Great Plains precipitation. There is was appreciably cooler, it becomes important to thus little compelling evidence that past droughts of examine the particular attributes of climate forcings this type have had a coherent pattern of sea surface that operated during 2012 and assess if they served temperature forcing. to condition the probability for severe drought over24
  29. 29. Interpretation of Origins of 2012 Central Great Plains Droughtthe central Great Plains in 2012. The warm SSTs in fective during summer. The phrase “perfect ocean” isthe Atlantic basin during 2012 are noteworthy, and thus more figurative, and does not connote an elixirrecent studies point to a summertime U.S. climate explaining the cause for all droughts. In particular,sensitivity to Atlantic forcing. Also, the tropical-wide as will be shown subsequently, the issue of centralSST pattern of the past year has many of the attri- U.S. summertime drought as relates to ocean forc-butes of the so-called “perfect ocean for drought” ing appears to be rather distinct from the SST forc-pattern. This consists of an increased zonal contrast ings conducive for cold season precipitation in thein SSTs between the eastern equatorial Pacific and southern portion of the US. What may be “perfect”the Indo-west Pacific.   In a published study, titled for understanding some region’s drought sensitivitythe “Perfect Ocean for Drought” (see Additional to ocean states, may be flawed and defective forReading) an analysis was conducted on how SST understanding droughts in other seasons and overconditions during 1998-2002 affected precipitation different regions.over the US (especially the southern regions thatspanned California to Florida) and other mid-latitude A few more comments on the attribution ofregion’s of the Northern Hemisphere. Those resem- droughts to particular forcing patterns is in order.bled the conditions seen during 2012, with abnor- In what has perhaps become jargon, several phras-mally warm Indo-West Pacific Ocean conditions and es or phenomena in addition to “perfect oceanabnormally cold east Pacific conditions. However, for drought” are getting increasingly circulated asthe SSTs that were deemed to be effective in drying “explanatory” for causes of events such as droughts.a widespread portion of mid-latitudes during the These include ENSO (the El Nino-Southern Oscilla-turn of the century drought likely did so via tropi- tion phenomenon that is associated with interan-cal-extratropical climate linkages that were endemic nual warm or cold states of the tropical east Pacificto the winter/spring season, and are unlikely as ef- ocean), PDO (the Pacific Decadal Oscillation that is Observed 2012 SST Departures: MJJ 2011 to May–Aug 2012 Figure 17. As in Fig. 15, except for the SST anomalies (°C) during the 12-month period antecedent to the occurrence of dry May-August 2012 central Great Plains drought. Reference period is 1901-1990. Data source is the monthly NOAA Merged Land-Ocean surface temperature analysis (MLOS). 25
  30. 30. Interpretation of Origins of 2012 Central Great Plains Drought associated with decadal cool or warm states of the Second is the so-called “tail response”, a sensitivi- Pacific Ocean especially north of 20°N), or global ty that reveals how the probability of a particular warming (the rise in average surface temperatures threshold exceedence (e.g., the odds of eclipsing a over the world land areas and oceans). In the subse- prior record value) changes as a consequence of the quent material of this section, we attempt to rig- specified forcing. orously test the connection between ocean condi- tions and also the state of external radiative forcing Key to this modeling technique for assessing the that operated during 2012 and the occurrence of impact of boundary conditions is an ensemble ap- drought over the central Plains. proach, whereby the period of simulation is repeat- ed a multitude of times. Here simulations that have The question of whether the particular SST condi- been repeated 20 times (a 20-member ensemble), tions in 2012 may have exerted a more substantial, and which differ from one another only in the ini- and potentially predictable influence on summer tial atmospheric conditions in January 1979 but in U.S. precipitation is addressed using climate simula- which identical time evolving forcings are specified, tions. Global atmospheric models that are run over are analyzed. The strategy is to average the monthly the period 1979-2012 are used herein. These are variability across the 20 members in order to deter- continuous simulations, begun from atmospheric mine the mean response to specified forcings. Note initial states in January 1979, and conclude in De- that the process of averaging eliminates the random cember 2012. The only constraining information internal variability of the atmosphere, and facilitates representing observed conditions in these simu- identifying the coherent signal from the forcing. lations is the sea surface temperature, sea ice, and However, analysis of the statistical distribution of external radiative forcing. These are specified in all 20-members is likewise important especially for the atmospheric model as monthly time evolving discerning how the frequency of extreme events is boundary conditions from January 1979- December affected by specified forcing. In this assessment, the 2012. Because the forcings are typically of a time use of 20-member ensemble simulations may be scale that is much longer than the time scale of at- adequate for estimating the coherent mean signal, mospheric variations, the atmospheric sensitivity to however, it is unlikely sufficient for estimating how such forcings is judged to be potentially predictable the statistics of extreme events are affected. This to the extent that such boundary forcings are them- should be kept in mind when judging the reliabil- selves predictable. The forcing conditions may act ity of model-based diagnoses. It must also be em- to influence the year-to-year variability of the atmo- phasized that a more thorough assessment would sphere and also the probabilities of certain extreme require the use of multiple models in order to min- conditions (e.g. severe drought), and the purpose of imize the possible influence of a particular climate the experiments is to quantify their influence. Cli- models’ biases, and larger ensemble sizes to better mate simulations of this type are referred to as ‘AMIP separate forced changes from unforced internal (Atmospheric Model Intercomparison Project) exper- variability. iments’, and are designed to determine the sensitiv- ity of the atmosphere, and the extent to which its The model used is the NCAR CAM4 global climate temporal evolution is constrained by known bound- model, with the simulations performed at a 1° ary forcings. (~100 km) resolution. Monthly varying SSTs and sea ice are based on a global monthly 1° analysis, and There are two particular aspects of the sensitivity the specified external radiative forcings consist of that are of interest. First is the mean response to the greenhouse gases (e.g. CO2, CH4, NO2, O3, CFCs), specified forcings, a sensitivity that reveals how the aerosols, solar, and volcanic aerosols. The latter most likely (e.g. median) outcome for a particular employ observed estimates through 2005, and then season changes as a consequence of the forcing. an emission scenario thereafter (RCP6.0, a moderate26
  31. 31. Interpretation of Origins of 2012 Central Great Plains Drought CAM4 2012 PPT Departures: MJJ 2011 to May–Aug 2012 Figure 18. As in Figure 9, except the simulated U.S. seasonal precipitation anomalies (mm) during the 12-month period antecedent to the occurrence of observed dry May-August conditions over the central Great Plains. The simulated May-August rainfall anomalies are shown in the lower right panel, and simulations for the prior season are shown chronologically in the other panels. Simulations based on NCAR CAM4 forced with observed SST, sea ice, and external radiative forcing. Plots show the 20-member ensemble average, and anomalies are relative to the model’s 1981-2010 climatology.emissions scenario pathway). The model output has tions generated at 12 operational forecast centers,been interpolated to U.S. climate divisions to facili- indicating that the weak signal in the CAM4 runs istate comparison with observations. unlikely a symptom of model bias. It is also worth noting that CAM4 simulations exhibit a strongerThe simulated ensemble mean precipitation anoma- signal of reduced summer rainfall anomalies in 2011lies for May-August 2012 are shown in Fig. 18 (lower over the southern Plains (Fig. 18, top left), whichright), as are the simulated precipitation anoma- though considerably less than the observed drynesslies for the prior 12 months. Only a weak signal of in 2011, suggests a stronger SST influence on thedryness is simulated during summer 2012 over the prior drought that spanned the southern Plains.central Great Plains, with the area-averaged anomalyover the 6-state index region in the CAM4 ensem- Consistent with a weak signal of reduced season-ble average being an order of magnitude weaker al mean rainfall, the overall distribution of thethan the observed anomaly. The particular SSTs of 20-member CAM4 simulations indicates a shift2012 thus appeared not to force the seasonal mean toward drier states. The box-whisker display in Fig.rainfall reduction over the central Plains, and this 19 shows, in the far right side, the distribution ofweak sensitivity implies that the most likely outcome the 20 realizations for summer 2012. Note that thefor central Plains precipitation in summer 2012 was extreme driest member, shown by a red asterisk,close to its climatological normal value. Further ranks among the driest model simulations for anyanalysis to be presented in section 6 will show a year during 1979-2012. Indicated hereby is that thesimilar weak signal in the ensemble rainfall predic- probability of an extreme dry summer over central 27
  32. 32. Interpretation of Origins of 2012 Central Great Plains Drought CAM4 Central US May–Aug Precipitation 1979-2012 Figure 19. Box-whisker plots of the May-August CAM4 simulated central Great Plains rainfall anomalies for 1979-2012. Extreme wet and dry members are shown with blue and red asterisks, Millimeters respectively. The horizontal dashed lines are the model’s 1-standardized departures of May-August rainfall. Green circles plot the observed rainfall anomalies for each year. The area is comprised of the 6-State region of WY, CO, NE, KS, MO, and IA. Year Great Plains may have been elevated during 2012. period into equal halves and taken the difference However, the ensemble size is too small to derive between the post and pre-1996 ensemble mean reliable estimates of the change in probability for CAM4 rainfall. This pattern bears considerable extreme threshold exceedences during 2012. An resemblance to the summer 2012 U.S. pattern of additional question, unresolved by the current set of simulations, is whether the tail probabilities in CAM4 May–Aug PPT: (1996-2012) minus (1979-1995) 2012 changed beyond what would be expected from the simple shift in the mean value of the statis- tical distribution. While these are technical matters laying beyond the scope of this assessment, they do touch on a fundamental science question — how do particular forcings affect not only the mean state of climate but also its modes of variability and the statistics of extreme events? There is an indication from CAM4 runs that there has been a consistent (albeit weak) dry signal each year during the past decade, and within each year’s distribution, the extreme driest member was been considerably lower than in prior decades. Figure 19 also shows the distribution of model rainfall simulations for each year since 1979, and the consistency of a mean dry signal after 1999 is Figure 20. The simulated change in May-August rainfall (mm) apparent. There is a coherent spatial scale to the for (1996-2012) minus (1979-1995) based on the 20-member ensemble mean CAM4 runs. Note that this change pattern of simulated summertime rainfall change, shown in simulated dryness is quite similar to the pattern of 2012 summer Fig. 20 where we have simply divided the simulation rainfall anomalies (see Fig. 2).28
  33. 33. Interpretation of Origins of 2012 Central Great Plains Drought CAM4 Central US Extreme Dry Probability CAM4 Central US Extreme Wet ProbabilityProbability Density Function PPT Departure (mm) Probability Density Function PPT Departure (mm) Figure 21. The probability distributions of the rainfall departures Figure 22. Same as Figure 21, except the probability distributions of the (mm) for the driest May-August central Great Plains CAM4 member in rainfall departures (mm) for the wettest May-August central Great Plains each year’s simulations during 1996-2012 (red curve) and for 1979- CAM4 member in each year’s simulations during 1996-2012 (red curve) 1995 simulations. There is a 20-member ensemble for each year, and and for 1979-1995 simulations. There is a 20-member ensemble for each the driest member has been extracted. Each PDF is thus based on 17 year, and the wettest member has been extracted. Each PDF is thus samples, which are displayed as red asterisks in Fig. 18. The area is based on 17 samples, which are displayed as blue asterisks in Fig. 18. comprised of the 6-State region of WY, CO, NE, KS, MO, and IA. The area is comprised of the 6-State region of WY, CO, NE, KS, MO, and IA. rainfall anomalies (see Figure 2), though of course of change in extreme event probabilities during any much weaker magnitude. The cause for the model’s single year, an examination of these extreme event protracted dryness is not currently known, though it statistics over consecutive years appears to reveal a is temporally associated with systematic pattern of change. a shift toward mostly cooler It is a speculative yet an states of the tropical east To illustrate the change in sim- intriguing conjecture that, Pacific that occurred after the ulated extreme summer rainfall while perhaps unbeknownst statistics over the central Plains, large 1997-98 El Niño event. and undetectable from the the probability distributions With this mean rainfall reduc- observations, the recent 10-15 (PDFs) of extreme values for the tion has come an increased year period may have been 1996-2012 runs are compared risk of severe drought during one of heightened risk for the to the extreme values for the summer over the central occurrence of a record setting 1979-2012. Figures 21 and 22 Great Plains in each year of summer drought over the show the results for the extreme the CAM4 runs after the late dry and wet PDFs, respectively. central Great Plains. 1990s. It is apparent from Simulated extreme event statis- inspection of the box-whisker tics for the recent period exhibit plots that the magnitudes of the single most dry a distinct increase in severe drought probabilities ensemble member (shown by the red asterisk) have (and also a distinct decrease in excessively wet prob- been consistently lower than in the prior decades of abilities). the model simulations. Thus, although a 20-mem- ber ensemble for any individual year may not pro- It is a speculative yet an intriguing conjecture that, vide reliable information from which to discern the while perhaps unbeknownst and undetectable from 29
  34. 34. Interpretation of Origins of 2012 Central Great Plains Drought the observations, the recent 10-15 year period may that such sensitivity is not readily verifiable from the have been one of heightened risk for the occurrence observations themselves. An additional question of a record setting summer drought over the central these results pose is whether the simulated change Great Plains. The analysis of CAM4 runs does not ex- in extreme drought risk is a symptom of climate plain, however, why the particular extreme drought change forcing related to global warming. There occurred in 2012 specifically — the model runs are several indications that this behavior in CAM4 is indicate that the risks were comparably elevated in largely unrelated to the model’s sensitivity to grad- all years during the last decade. We know that no ually increasing anthropogenic forcing. One key such event has occurred in the last decade; one has indication is the rather sudden character of change to return to 1988 to have experienced a drought as in model simulations toward dry conditions in the severe as occurred in 2012. The fact that an extreme late 1990s. Though one cannot dismiss the possibili- drought did occur in 2012 may thus be largely coin- ty that a steady forcing (for instance increasing CO2) cidental, and by the very nature of extreme events, may not provoke an abrupt change in responses, its occurrence was a low probability outcome. And there are other plausible physical explanations for while even those small odds may have been hedged the shift in model behavior in the 1990s including by the particular forcings, the odds remained very natural swings in ocean states (for instance, Pacific small nonetheless. The implication from this analy- and Atlantic Ocean natural decadal SST variability). ses is that the 2012 drought may not have been es- Note also that the Great Plains surface temperature pecially predictable even a month or two in advance, responses in CAM4 reveal a rather abrupt change in an inference that is further supported by results in summertime conditions over the central U.S. after section 6 wherein the poor performance of opera- 1998, with sustained mean warmth having ensem- tional forecasts for this drought are documented. ble averaged magnitudes consistently between +0.5 to +1.0 standardized departures (Fig. 23). Further analysis of other climate models, similarly forced, would be required to build confidence in An additional indication that global warming is the realism of the CAM4 results, especially given unlikely a major factor in the 2012 central Plains CAM4 Central US May–Aug Temperature 1979-2012 Figure 23. Box-whisker plots of the May-August CAM4 simulated central Great Plains surface temperature anomalies for 1979- 2012. Extreme warm and cold members are shown with red and blue asterisks, respectively. Degrees Celcius The horizontal dashed lines are the model’s 1-standardized departures of May-August temperature. The area is comprised of the 6-State region of WY, CO, NE, KS, MO, and IA. Year30
  35. 35. Interpretation of Origins of 2012 Central Great Plains Droughtdrought is drawn from a further set of climate sim- Central US PPTulations that have been performed using the NCAR Climate Change Sensitivitymodeling system. Here the coupled ocean-atmo-sphere version of the model has been used to assessits sensitivity to the change in external radiative Probability Density Functionforcing since about 1850. This is the same modelincluded among many modeling centers’ that arecontributing to the upcoming IntergovernmentalPanels on Climate Change (IPCC) assessment of cli-mate change. This model is also part of the so-calledClimate Model Intercomparison Project – Phase 5(CMIP5). Two 500-yr long runs of CCSM4 were con-ducted, one using year-1850 radiative forcing, anda second using year-2000 radiative forcing. In theseexperiments, which are different from the atmo-spheric model simulations wherein SSTs were spec-ified, the coupled model’s ocean responds to thechange in specified radiative forcing. Broadly speak- May-August Total PPT (mm)ing, the model yields a realistic warming of globally Figure 24. The probability distributions of the May-Augustaveraged temperatures (~1.5°C) in response to this total precipitation over the central Great Plains (mm) for CCSM4change in radiative forcing. Nonetheless, the simu- equilibrium simulations using Yr1850 external radiative forcinglations do not show a shift toward mean dryness in (blue curve) and using Yr2000 external radiative forcing. Each run is 500 yrs long, and plotted are the last 400 years of results. Thesummer over the central Plains, or a systematic in- atmospheric model component in these coupled simulations iscrease (decrease) in extreme dry (wet) probabilities. the same as used for the 1979-2012 AMIP runs of CAM4. The areaFigure 24 plots the PDFs of summer central Great is comprised of the 6-State region of WY, CO, NE, KS, MO, and IA.Plains rainfall from two parallel 500-yr CCSM equi-librium runs, one using year-1850 external radiative radiative forcing did not generate an appreciable dryforcing and the other using year-2000 external radia- signal over the central Great Plains in 2012. Second,tive forcing. The mean change in summer precipita- the CCSM4 coupled model simulations using thetion is about a 1.5 mm increase over the Great Plains change in external radiative forcing between yearin the warmer climate state. The variability in mean 1850 and 2000 do not exhibit a systematic changesummer rainfall increases in the warmed climate to drier conditions. Perhaps most striking is the wide(standard deviation increases about 15%), with both range of summer central Plains rainfall that occursextreme dry and extreme wet summers increasing. within the 500 years of simulations in CCSM4 (shown by the tick marks in Fig. 24) for a particular forcingThis is not intended to be a comprehensive assess- regime. This range is far greater than any change inment of the possible effects of global warming on that range (and related statistics) associated with thethe 2012 central Plains drought, and hence results forcing change. The implication is that the signal ofhere are inconclusive. Further analysis will be re- climate change may be very small compared to thequired to assess the role of global warming on noise of the intrinsic year-to-year variability. Detect-recent and future precipitation variability over the ability of a global warming signal in the statistics ofGreat Plains using the full suite of CMIP5 models. summertime Great Plains rainfall may thus be veryA few points are nonetheless worth noting even difficult at this time.from the limited analysis presented herein. First, theCAM4 atmospheric model simulations for 1979-2012using the actual observed SST and specified external 31
  36. 36. Experimental methods are being studied thatoffer some hope for improved prediction, at leastfor short lead times, of drought conditions suchas occurred in 2012.
  37. 37. Prediction for the Summer 2012Operational Precipitation and model predictions are consistent with the retrospec- tive AMIP simulations of CAM4. Namely, both exhibitTemperature Forecast a weak signal of reduced summertime rainfall over the central U.S., but a comparatively strong signal ofGlobal Producing Centers (GPC) of seasonal climate surface warmth. In this regard, both simulation andpredictions regularly supply their data to the WMO prediction runs imply that there was an appreciablelead center for long-range predictions, which in increase in probability that the central Great Plainsturn produce various statistics of these predictions. would experience warmer than normal tempera-There are currently 12 operational climate prediction tures during summer 2012. However, this forcedcenters around the world that participate. Shown warming signal alone fails to explain the heat wavein Figures 25 and 26 are simple composites of the that occurred. The latter almost certainly resulted12-centers’ seasonal predictions for May-July 2012 mainly from rainfall’s absence, and the associatedand June-August 2012. feedbacks on temperatures that ensued due toThe multi-model predictions, based on April initial- severely depleted soil moisture. Also, the operationalizations, for May-July 2012 reveal a weak dry signal, predictions suggest that initial conditions in May,located over the north central U.S., but a strong sig- which would have begun to reflect the reducednal of warmth that spans the entire contiguous U.S. soil moisture states owing to the lack of May rain-The May 2012 initialized predictions for the June-Au- fall, failed to increase the probabilities of centralgust period show no appreciable rainfall signal, but a U.S. drought in June-August. While soil moisturecontinued widespread large amplitude warm signal. conditions may have affected some aspects of theIn many ways, the results of the initialized coupled forecasts during summer 2012 such as temperature, April 2012 Figure 25. Equal-weighted composites of 12 operational centers’ seasonal predictions for May-July 2012 for global sea surface temperature departures (°C, top left), global precipitation departures (mm, bottom left), and for North American sector precipitation departures (mm, top right) and for North American sector surface temperature anomalies (°C, bottom right). Forecasts are based on April 2012 initializations. Data source is the WMO GPC project. 33
  38. 38. Interpretation of Origins of 2012 Central Great Plains Drought May 2012 Figure 26. Same as Fig. 25, except for the June-August seasonal predictions based on May 2012 initializations. their effects on rainfall were either not systematic with initial soil conditions from NCEP Reanalysis 2 across the models, or they were weak for the central (R-2) and NLDAS. As part of ongoing research to Great Plains areas of interest. test sensitivity, here each case is facilitated with 4~5 CWRF physics configurations. The operational Simulations of Precipitation and forecast results using the existing configurations of CFSv2 and ECHAM4.5 are also shown as the first blue Soil Moisture bar in each grid. The CWRF/ERI simulation consis- tently captures the low rainfall in summer 2012 Experimental methods are being studied that offer (relative to rainfall conditions the prior year), while some hope for improved prediction, at least for short most other forecasts fail to do so. Whether the gains lead times, of drought conditions such as occurred seen in CWRF/ERI simulation mode translate into in 2012. Shown in Fig. 27 are simulations of summer- improved predictions is matter of current research. time Midwest regional mean precipitation differenc- es (2012-2011) driven by ECMWF Interim Reanalysis Also shown are the model predicted monthly evolu- (ERI) based on the CWRF model. For comparison, tions of soil moisture at 2m depth (Fig. 28; hereafter also shown are the CFSv2 operational forecasts ini- denoted as SM_2m). Not surprisingly, initial soil tialized at May 1, and ECHAM4.5 real-time forecasts conditions have the dominant impact on subse- initialized at April 1, May 1 and June 1 respectively quent soil moisture conditions for about the first34